nsz Git - musl/blob - src/ctype/towctrans.c

   1 #include <wchar.h>
   2 #include <wctype.h>
   3 #include <stdio.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('A','Z','a'),
  14         CASEMAP(0xc0,0xde,0xe0),
  15
  16         CASELACE(0x0100,0x012e),
  17         CASELACE(0x0132,0x0136),
  18         CASELACE(0x0139,0x0147),
  19         CASELACE(0x014a,0x0176),
  20         CASELACE(0x0179,0x017d),
  21
  22         CASELACE(0x370,0x372),
  23         CASEMAP(0x391,0x3a1,0x3b1),
  24         CASEMAP(0x3a3,0x3ab,0x3c3),
  25         CASEMAP(0x400,0x40f,0x450),
  26         CASEMAP(0x410,0x42f,0x430),
  27
  28         CASELACE(0x460,0x480),
  29         CASELACE(0x48a,0x4be),
  30         CASELACE(0x4c1,0x4cd),
  31         CASELACE(0x4d0,0x50e),
  32
  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         CASELACE(0x246,0x24e),
  67         CASELACE(0x510,0x512),
  68         CASEMAP(0x2160,0x216f,0x2170),
  69         CASEMAP(0x2c00,0x2c2e,0x2c30),
  70         CASELACE(0x2c67,0x2c6b),
  71         CASELACE(0x2c80,0x2ce2),
  72
  73         CASELACE(0xa722,0xa72e),
  74         CASELACE(0xa732,0xa76e),
  75         CASELACE(0xa779,0xa77b),
  76         CASELACE(0xa77e,0xa786),
  77
  78         CASEMAP(0xff21,0xff3a,0xff41),
  79         { 0,0,0 }
  80 };
  81
  82 static const unsigned short pairs[][2] = {
  83         { 'I',    0x0131 },
  84         { 'S',    0x017f },
  85         { 0x0130, 'i'    },
  86         { 0x0178, 0x00ff },
  87         { 0x0181, 0x0253 },
  88         { 0x0182, 0x0183 },
  89         { 0x0184, 0x0185 },
  90         { 0x0186, 0x0254 },
  91         { 0x0187, 0x0188 },
  92         { 0x0189, 0x0256 },
  93         { 0x018a, 0x0257 },
  94         { 0x018b, 0x018c },
  95         { 0x018e, 0x01dd },
  96         { 0x018f, 0x0259 },
  97         { 0x0190, 0x025b },
  98         { 0x0191, 0x0192 },
  99         { 0x0193, 0x0260 },
 100         { 0x0194, 0x0263 },
 101         { 0x0196, 0x0269 },
 102         { 0x0197, 0x0268 },
 103         { 0x0198, 0x0199 },
 104         { 0x019c, 0x026f },
 105         { 0x019d, 0x0272 },
 106         { 0x019f, 0x0275 },
 107         { 0x01a6, 0x0280 },
 108         { 0x01a7, 0x01a8 },
 109         { 0x01a9, 0x0283 },
 110         { 0x01ac, 0x01ad },
 111         { 0x01ae, 0x0288 },
 112         { 0x01af, 0x01b0 },
 113         { 0x01b1, 0x028a },
 114         { 0x01b2, 0x028b },
 115         { 0x01b7, 0x0292 },
 116         { 0x01b8, 0x01b9 },
 117         { 0x01bc, 0x01bd },
 118         { 0x01c4, 0x01c6 },
 119         { 0x01c4, 0x01c5 },
 120         { 0x01c5, 0x01c6 },
 121         { 0x01c7, 0x01c9 },
 122         { 0x01c7, 0x01c8 },
 123         { 0x01c8, 0x01c9 },
 124         { 0x01ca, 0x01cc },
 125         { 0x01ca, 0x01cb },
 126         { 0x01cb, 0x01cc },
 127         { 0x01f1, 0x01f3 },
 128         { 0x01f1, 0x01f2 },
 129         { 0x01f2, 0x01f3 },
 130         { 0x01f4, 0x01f5 },
 131         { 0x01f6, 0x0195 },
 132         { 0x01f7, 0x01bf },
 133         { 0x0220, 0x019e },
 134         { 0x0386, 0x03ac },
 135         { 0x0388, 0x03ad },
 136         { 0x0389, 0x03ae },
 137         { 0x038a, 0x03af },
 138         { 0x038c, 0x03cc },
 139         { 0x038e, 0x03cd },
 140         { 0x038f, 0x03ce },
 141         { 0x0399, 0x0345 },
 142         { 0x0399, 0x1fbe },
 143         { 0x03a3, 0x03c2 },
 144         { 0x03f7, 0x03f8 },
 145         { 0x03fa, 0x03fb },
 146         { 0x1e60, 0x1e9b },
 147
 148         { 0x1f59, 0x1f51 },
 149         { 0x1f5b, 0x1f53 },
 150         { 0x1f5d, 0x1f55 },
 151         { 0x1f5f, 0x1f57 },
 152         { 0x1fbc, 0x1fb3 },
 153         { 0x1fcc, 0x1fc3 },
 154         { 0x1fec, 0x1fe5 },
 155         { 0x1ffc, 0x1ff3 },
 156
 157         { 0x23a, 0x2c65 },
 158         { 0x23b, 0x23c },
 159         { 0x23d, 0x19a },
 160         { 0x23e, 0x2c66 },
 161         { 0x241, 0x242 },
 162         { 0x243, 0x180 },
 163         { 0x244, 0x289 },
 164         { 0x245, 0x28c },
 165         { 0x3f4, 0x3b8 },
 166         { 0x3f9, 0x3f2 },
 167         { 0x3fd, 0x37b },
 168         { 0x3fe, 0x37c },
 169         { 0x3ff, 0x37d },
 170         { 0x4c0, 0x4cf },
 171
 172         { 0x2126, 0x3c9 },
 173         { 0x212a, 'k' },
 174         { 0x212b, 0xe5 },
 175         { 0x2132, 0x214e },
 176         { 0x2183, 0x2184 },
 177         { 0x2c60, 0x2c61 },
 178         { 0x2c62, 0x26b },
 179         { 0x2c63, 0x1d7d },
 180         { 0x2c64, 0x27d },
 181         { 0x2c6d, 0x251 },
 182         { 0x2c6e, 0x271 },
 183         { 0x2c6f, 0x250 },
 184         { 0x2c72, 0x2c73 },
 185         { 0x2c75, 0x2c76 },
 186
 187         { 0xa77d, 0x1d79 },
 188
 189         /* bogus greek 'symbol' letters */
 190         { 0x376, 0x377 },
 191         { 0x39c, 0xb5 },
 192         { 0x392, 0x3d0 },
 193         { 0x398, 0x3d1 },
 194         { 0x3a6, 0x3d5 },
 195         { 0x3a0, 0x3d6 },
 196         { 0x39a, 0x3f0 },
 197         { 0x3a1, 0x3f1 },
 198         { 0x395, 0x3f5 },
 199         { 0x3cf, 0x3d7 },
 200
 201         { 0,0 }
 202 };
 203
 204
 205 static wchar_t __towcase(wchar_t wc, int lower)
 206 {
 207         int i;
 208         int lmul = 2*lower-1;
 209         int lmask = lower-1;
 210         if ((unsigned)wc - 0x10400 < 0x50)
 211                 return wc + lmul*0x28;
 212         /* no letters with case in these large ranges */
 213         if (!iswalpha(wc)
 214          || (unsigned)wc - 0x0600 <= 0x0fff-0x0600
 215          || (unsigned)wc - 0x2e00 <= 0xa6ff-0x2e00
 216          || (unsigned)wc - 0xa800 <= 0xfeff-0xa800)
 217                 return wc;
 218         /* special case because the diff between upper/lower is too big */
 219         if ((unsigned)wc - 0x10a0 < 0x26 || (unsigned)wc - 0x2d00 < 0x26)
 220                 return wc + lmul*(0x2d00-0x10a0);
 221         for (i=0; casemaps[i].len; i++) {
 222                 int base = casemaps[i].upper + (lmask & casemaps[i].lower);
 223                 if ((unsigned)wc-base < casemaps[i].len) {
 224                         if (casemaps[i].lower == 1)
 225                                 return wc + lower - ((wc-casemaps[i].upper)&1);
 226                         return wc + lmul*casemaps[i].lower;
 227                 }
 228         }
 229         for (i=0; pairs[i][1-lower]; i++) {
 230                 if (pairs[i][1-lower] == wc)
 231                         return pairs[i][lower];
 232         }
 233         if ((unsigned)wc - 0x10428 + (lower<<5) + (lower<<3) < 0x28)
 234                 return wc - 0x28 + (lower<<10) + (lower<<6);
 235         return wc;
 236 }
 237
 238 wint_t towupper(wint_t wc)
 239 {
 240         return __towcase(wc, 0);
 241 }
 242
 243 wint_t towlower(wint_t wc)
 244 {
 245         return __towcase(wc, 1);
 246 }