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(0x514,0x526),
|
---|
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 | CASELACE(0x2ceb,0x2ced),
|
---|
73 |
|
---|
74 | CASELACE(0xa640,0xa66c),
|
---|
75 | CASELACE(0xa680,0xa696),
|
---|
76 |
|
---|
77 | CASELACE(0xa722,0xa72e),
|
---|
78 | CASELACE(0xa732,0xa76e),
|
---|
79 | CASELACE(0xa779,0xa77b),
|
---|
80 | CASELACE(0xa77e,0xa786),
|
---|
81 |
|
---|
82 | CASELACE(0xa790,0xa792),
|
---|
83 | CASELACE(0xa7a0,0xa7a8),
|
---|
84 |
|
---|
85 | CASEMAP(0xff21,0xff3a,0xff41),
|
---|
86 | { 0,0,0 }
|
---|
87 | };
|
---|
88 |
|
---|
89 | static const unsigned short pairs[][2] = {
|
---|
90 | { 'I', 0x0131 },
|
---|
91 | { 'S', 0x017f },
|
---|
92 | { 0x0130, 'i' },
|
---|
93 | { 0x0178, 0x00ff },
|
---|
94 | { 0x0181, 0x0253 },
|
---|
95 | { 0x0182, 0x0183 },
|
---|
96 | { 0x0184, 0x0185 },
|
---|
97 | { 0x0186, 0x0254 },
|
---|
98 | { 0x0187, 0x0188 },
|
---|
99 | { 0x0189, 0x0256 },
|
---|
100 | { 0x018a, 0x0257 },
|
---|
101 | { 0x018b, 0x018c },
|
---|
102 | { 0x018e, 0x01dd },
|
---|
103 | { 0x018f, 0x0259 },
|
---|
104 | { 0x0190, 0x025b },
|
---|
105 | { 0x0191, 0x0192 },
|
---|
106 | { 0x0193, 0x0260 },
|
---|
107 | { 0x0194, 0x0263 },
|
---|
108 | { 0x0196, 0x0269 },
|
---|
109 | { 0x0197, 0x0268 },
|
---|
110 | { 0x0198, 0x0199 },
|
---|
111 | { 0x019c, 0x026f },
|
---|
112 | { 0x019d, 0x0272 },
|
---|
113 | { 0x019f, 0x0275 },
|
---|
114 | { 0x01a6, 0x0280 },
|
---|
115 | { 0x01a7, 0x01a8 },
|
---|
116 | { 0x01a9, 0x0283 },
|
---|
117 | { 0x01ac, 0x01ad },
|
---|
118 | { 0x01ae, 0x0288 },
|
---|
119 | { 0x01af, 0x01b0 },
|
---|
120 | { 0x01b1, 0x028a },
|
---|
121 | { 0x01b2, 0x028b },
|
---|
122 | { 0x01b7, 0x0292 },
|
---|
123 | { 0x01b8, 0x01b9 },
|
---|
124 | { 0x01bc, 0x01bd },
|
---|
125 | { 0x01c4, 0x01c6 },
|
---|
126 | { 0x01c4, 0x01c5 },
|
---|
127 | { 0x01c5, 0x01c6 },
|
---|
128 | { 0x01c7, 0x01c9 },
|
---|
129 | { 0x01c7, 0x01c8 },
|
---|
130 | { 0x01c8, 0x01c9 },
|
---|
131 | { 0x01ca, 0x01cc },
|
---|
132 | { 0x01ca, 0x01cb },
|
---|
133 | { 0x01cb, 0x01cc },
|
---|
134 | { 0x01f1, 0x01f3 },
|
---|
135 | { 0x01f1, 0x01f2 },
|
---|
136 | { 0x01f2, 0x01f3 },
|
---|
137 | { 0x01f4, 0x01f5 },
|
---|
138 | { 0x01f6, 0x0195 },
|
---|
139 | { 0x01f7, 0x01bf },
|
---|
140 | { 0x0220, 0x019e },
|
---|
141 | { 0x0386, 0x03ac },
|
---|
142 | { 0x0388, 0x03ad },
|
---|
143 | { 0x0389, 0x03ae },
|
---|
144 | { 0x038a, 0x03af },
|
---|
145 | { 0x038c, 0x03cc },
|
---|
146 | { 0x038e, 0x03cd },
|
---|
147 | { 0x038f, 0x03ce },
|
---|
148 | { 0x0399, 0x0345 },
|
---|
149 | { 0x0399, 0x1fbe },
|
---|
150 | { 0x03a3, 0x03c2 },
|
---|
151 | { 0x03f7, 0x03f8 },
|
---|
152 | { 0x03fa, 0x03fb },
|
---|
153 | { 0x1e60, 0x1e9b },
|
---|
154 | { 0x1e9e, 0xdf },
|
---|
155 |
|
---|
156 | { 0x1f59, 0x1f51 },
|
---|
157 | { 0x1f5b, 0x1f53 },
|
---|
158 | { 0x1f5d, 0x1f55 },
|
---|
159 | { 0x1f5f, 0x1f57 },
|
---|
160 | { 0x1fbc, 0x1fb3 },
|
---|
161 | { 0x1fcc, 0x1fc3 },
|
---|
162 | { 0x1fec, 0x1fe5 },
|
---|
163 | { 0x1ffc, 0x1ff3 },
|
---|
164 |
|
---|
165 | { 0x23a, 0x2c65 },
|
---|
166 | { 0x23b, 0x23c },
|
---|
167 | { 0x23d, 0x19a },
|
---|
168 | { 0x23e, 0x2c66 },
|
---|
169 | { 0x241, 0x242 },
|
---|
170 | { 0x243, 0x180 },
|
---|
171 | { 0x244, 0x289 },
|
---|
172 | { 0x245, 0x28c },
|
---|
173 | { 0x3f4, 0x3b8 },
|
---|
174 | { 0x3f9, 0x3f2 },
|
---|
175 | { 0x3fd, 0x37b },
|
---|
176 | { 0x3fe, 0x37c },
|
---|
177 | { 0x3ff, 0x37d },
|
---|
178 | { 0x4c0, 0x4cf },
|
---|
179 |
|
---|
180 | { 0x2126, 0x3c9 },
|
---|
181 | { 0x212a, 'k' },
|
---|
182 | { 0x212b, 0xe5 },
|
---|
183 | { 0x2132, 0x214e },
|
---|
184 | { 0x2183, 0x2184 },
|
---|
185 | { 0x2c60, 0x2c61 },
|
---|
186 | { 0x2c62, 0x26b },
|
---|
187 | { 0x2c63, 0x1d7d },
|
---|
188 | { 0x2c64, 0x27d },
|
---|
189 | { 0x2c6d, 0x251 },
|
---|
190 | { 0x2c6e, 0x271 },
|
---|
191 | { 0x2c6f, 0x250 },
|
---|
192 | { 0x2c70, 0x252 },
|
---|
193 | { 0x2c72, 0x2c73 },
|
---|
194 | { 0x2c75, 0x2c76 },
|
---|
195 | { 0x2c7e, 0x23f },
|
---|
196 | { 0x2c7f, 0x240 },
|
---|
197 | { 0x2cf2, 0x2cf3 },
|
---|
198 |
|
---|
199 | { 0xa77d, 0x1d79 },
|
---|
200 | { 0xa78b, 0xa78c },
|
---|
201 | { 0xa78d, 0x265 },
|
---|
202 | { 0xa7aa, 0x266 },
|
---|
203 |
|
---|
204 | { 0x10c7, 0x2d27 },
|
---|
205 | { 0x10cd, 0x2d2d },
|
---|
206 |
|
---|
207 | /* bogus greek 'symbol' letters */
|
---|
208 | { 0x376, 0x377 },
|
---|
209 | { 0x39c, 0xb5 },
|
---|
210 | { 0x392, 0x3d0 },
|
---|
211 | { 0x398, 0x3d1 },
|
---|
212 | { 0x3a6, 0x3d5 },
|
---|
213 | { 0x3a0, 0x3d6 },
|
---|
214 | { 0x39a, 0x3f0 },
|
---|
215 | { 0x3a1, 0x3f1 },
|
---|
216 | { 0x395, 0x3f5 },
|
---|
217 | { 0x3cf, 0x3d7 },
|
---|
218 |
|
---|
219 | { 0,0 }
|
---|
220 | };
|
---|
221 |
|
---|
222 |
|
---|
223 | static wchar_t __towcase(wchar_t wc, int lower)
|
---|
224 | {
|
---|
225 | int i;
|
---|
226 | int lmul = 2*lower-1;
|
---|
227 | int lmask = lower-1;
|
---|
228 | /* no letters with case in these large ranges */
|
---|
229 | if (!iswalpha(wc)
|
---|
230 | || (unsigned)wc - 0x0600 <= 0x0fff-0x0600
|
---|
231 | || (unsigned)wc - 0x2e00 <= 0xa63f-0x2e00
|
---|
232 | || (unsigned)wc - 0xa800 <= 0xfeff-0xa800)
|
---|
233 | return wc;
|
---|
234 | /* special case because the diff between upper/lower is too big */
|
---|
235 | if (lower && (unsigned)wc - 0x10a0 < 0x2e)
|
---|
236 | if (wc>0x10c5 && wc != 0x10c7 && wc != 0x10cd) return wc;
|
---|
237 | else return wc + 0x2d00 - 0x10a0;
|
---|
238 | if (!lower && (unsigned)wc - 0x2d00 < 0x26)
|
---|
239 | if (wc>0x2d25 && wc != 0x2d27 && wc != 0x2d2d) return wc;
|
---|
240 | else return wc + 0x10a0 - 0x2d00;
|
---|
241 | for (i=0; casemaps[i].len; i++) {
|
---|
242 | int base = casemaps[i].upper + (lmask & casemaps[i].lower);
|
---|
243 | if ((unsigned)wc-base < casemaps[i].len) {
|
---|
244 | if (casemaps[i].lower == 1)
|
---|
245 | return wc + lower - ((wc-casemaps[i].upper)&1);
|
---|
246 | return wc + lmul*casemaps[i].lower;
|
---|
247 | }
|
---|
248 | }
|
---|
249 | for (i=0; pairs[i][1-lower]; i++) {
|
---|
250 | if (pairs[i][1-lower] == wc)
|
---|
251 | return pairs[i][lower];
|
---|
252 | }
|
---|
253 | if ((unsigned)wc - (0x10428 - 0x28*lower) < 0x28)
|
---|
254 | return wc - 0x28 + 0x50*lower;
|
---|
255 | return wc;
|
---|
256 | }
|
---|
257 |
|
---|
258 | wint_t towupper(wint_t wc)
|
---|
259 | {
|
---|
260 | return (unsigned)wc < 128 ? toupper(wc) : __towcase(wc, 0);
|
---|
261 | }
|
---|
262 |
|
---|
263 | wint_t towlower(wint_t wc)
|
---|
264 | {
|
---|
265 | return (unsigned)wc < 128 ? tolower(wc) : __towcase(wc, 1);
|
---|
266 | }
|
---|
267 |
|
---|
268 | wint_t __towupper_l(wint_t c, locale_t l)
|
---|
269 | {
|
---|
270 | return towupper(c);
|
---|
271 | }
|
---|
272 |
|
---|
273 | wint_t __towlower_l(wint_t c, locale_t l)
|
---|
274 | {
|
---|
275 | return towlower(c);
|
---|
276 | }
|
---|
277 |
|
---|
278 | weak_alias(__towupper_l, towupper_l);
|
---|
279 | weak_alias(__towlower_l, towlower_l);
|
---|