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 CASELACE(0x514,0x526),
34 CASEMAP(0x531,0x556,0x561),
35
36 CASELACE(0x01a0,0x01a4),
37 CASELACE(0x01b3,0x01b5),
38 CASELACE(0x01cd,0x01db),
39 CASELACE(0x01de,0x01ee),
40 CASELACE(0x01f8,0x021e),
41 CASELACE(0x0222,0x0232),
42 CASELACE(0x03d8,0x03ee),
43
44 CASELACE(0x1e00,0x1e94),
45 CASELACE(0x1ea0,0x1efe),
46
47 CASEMAP(0x1f08,0x1f0f,0x1f00),
48 CASEMAP(0x1f18,0x1f1d,0x1f10),
49 CASEMAP(0x1f28,0x1f2f,0x1f20),
50 CASEMAP(0x1f38,0x1f3f,0x1f30),
51 CASEMAP(0x1f48,0x1f4d,0x1f40),
52
53 CASEMAP(0x1f68,0x1f6f,0x1f60),
54 CASEMAP(0x1f88,0x1f8f,0x1f80),
55 CASEMAP(0x1f98,0x1f9f,0x1f90),
56 CASEMAP(0x1fa8,0x1faf,0x1fa0),
57 CASEMAP(0x1fb8,0x1fb9,0x1fb0),
58 CASEMAP(0x1fba,0x1fbb,0x1f70),
59 CASEMAP(0x1fc8,0x1fcb,0x1f72),
60 CASEMAP(0x1fd8,0x1fd9,0x1fd0),
61 CASEMAP(0x1fda,0x1fdb,0x1f76),
62 CASEMAP(0x1fe8,0x1fe9,0x1fe0),
63 CASEMAP(0x1fea,0x1feb,0x1f7a),
64 CASEMAP(0x1ff8,0x1ff9,0x1f78),
65 CASEMAP(0x1ffa,0x1ffb,0x1f7c),
66
67 CASELACE(0x246,0x24e),
68 CASELACE(0x510,0x512),
69 CASEMAP(0x2160,0x216f,0x2170),
70 CASEMAP(0x2c00,0x2c2e,0x2c30),
71 CASELACE(0x2c67,0x2c6b),
72 CASELACE(0x2c80,0x2ce2),
73 CASELACE(0x2ceb,0x2ced),
74
75 CASELACE(0xa640,0xa66c),
76 CASELACE(0xa680,0xa696),
77
78 CASELACE(0xa722,0xa72e),
79 CASELACE(0xa732,0xa76e),
80 CASELACE(0xa779,0xa77b),
81 CASELACE(0xa77e,0xa786),
82
83 CASELACE(0xa790,0xa792),
84 CASELACE(0xa7a0,0xa7a8),
85
86 CASEMAP(0xff21,0xff3a,0xff41),
87 { 0,0,0 }
88 };
89
90 static const unsigned short pairs[][2] = {
91 { 'I', 0x0131 },
92 { 'S', 0x017f },
93 { 0x0130, 'i' },
94 { 0x0178, 0x00ff },
95 { 0x0181, 0x0253 },
96 { 0x0182, 0x0183 },
97 { 0x0184, 0x0185 },
98 { 0x0186, 0x0254 },
99 { 0x0187, 0x0188 },
100 { 0x0189, 0x0256 },
101 { 0x018a, 0x0257 },
102 { 0x018b, 0x018c },
103 { 0x018e, 0x01dd },
104 { 0x018f, 0x0259 },
105 { 0x0190, 0x025b },
106 { 0x0191, 0x0192 },
107 { 0x0193, 0x0260 },
108 { 0x0194, 0x0263 },
109 { 0x0196, 0x0269 },
110 { 0x0197, 0x0268 },
111 { 0x0198, 0x0199 },
112 { 0x019c, 0x026f },
113 { 0x019d, 0x0272 },
114 { 0x019f, 0x0275 },
115 { 0x01a6, 0x0280 },
116 { 0x01a7, 0x01a8 },
117 { 0x01a9, 0x0283 },
118 { 0x01ac, 0x01ad },
119 { 0x01ae, 0x0288 },
120 { 0x01af, 0x01b0 },
121 { 0x01b1, 0x028a },
122 { 0x01b2, 0x028b },
123 { 0x01b7, 0x0292 },
124 { 0x01b8, 0x01b9 },
125 { 0x01bc, 0x01bd },
126 { 0x01c4, 0x01c6 },
127 { 0x01c4, 0x01c5 },
128 { 0x01c5, 0x01c6 },
129 { 0x01c7, 0x01c9 },
130 { 0x01c7, 0x01c8 },
131 { 0x01c8, 0x01c9 },
132 { 0x01ca, 0x01cc },
133 { 0x01ca, 0x01cb },
134 { 0x01cb, 0x01cc },
135 { 0x01f1, 0x01f3 },
136 { 0x01f1, 0x01f2 },
137 { 0x01f2, 0x01f3 },
138 { 0x01f4, 0x01f5 },
139 { 0x01f6, 0x0195 },
140 { 0x01f7, 0x01bf },
141 { 0x0220, 0x019e },
142 { 0x0386, 0x03ac },
143 { 0x0388, 0x03ad },
144 { 0x0389, 0x03ae },
145 { 0x038a, 0x03af },
146 { 0x038c, 0x03cc },
147 { 0x038e, 0x03cd },
148 { 0x038f, 0x03ce },
149 { 0x0399, 0x0345 },
150 { 0x0399, 0x1fbe },
151 { 0x03a3, 0x03c2 },
152 { 0x03f7, 0x03f8 },
153 { 0x03fa, 0x03fb },
154 { 0x1e60, 0x1e9b },
155 { 0xdf, 0xdf },
156 { 0x1e9e, 0xdf },
157
158 { 0x1f59, 0x1f51 },
159 { 0x1f5b, 0x1f53 },
160 { 0x1f5d, 0x1f55 },
161 { 0x1f5f, 0x1f57 },
162 { 0x1fbc, 0x1fb3 },
163 { 0x1fcc, 0x1fc3 },
164 { 0x1fec, 0x1fe5 },
165 { 0x1ffc, 0x1ff3 },
166
167 { 0x23a, 0x2c65 },
168 { 0x23b, 0x23c },
169 { 0x23d, 0x19a },
170 { 0x23e, 0x2c66 },
171 { 0x241, 0x242 },
172 { 0x243, 0x180 },
173 { 0x244, 0x289 },
174 { 0x245, 0x28c },
175 { 0x3f4, 0x3b8 },
176 { 0x3f9, 0x3f2 },
177 { 0x3fd, 0x37b },
178 { 0x3fe, 0x37c },
179 { 0x3ff, 0x37d },
180 { 0x4c0, 0x4cf },
181
182 { 0x2126, 0x3c9 },
183 { 0x212a, 'k' },
184 { 0x212b, 0xe5 },
185 { 0x2132, 0x214e },
186 { 0x2183, 0x2184 },
187 { 0x2c60, 0x2c61 },
188 { 0x2c62, 0x26b },
189 { 0x2c63, 0x1d7d },
190 { 0x2c64, 0x27d },
191 { 0x2c6d, 0x251 },
192 { 0x2c6e, 0x271 },
193 { 0x2c6f, 0x250 },
194 { 0x2c70, 0x252 },
195 { 0x2c72, 0x2c73 },
196 { 0x2c75, 0x2c76 },
197 { 0x2c7e, 0x23f },
198 { 0x2c7f, 0x240 },
199 { 0x2cf2, 0x2cf3 },
200
201 { 0xa77d, 0x1d79 },
202 { 0xa78b, 0xa78c },
203 { 0xa78d, 0x265 },
204 { 0xa7aa, 0x266 },
205
206 { 0x10c7, 0x2d27 },
207 { 0x10cd, 0x2d2d },
208
209 /* bogus greek 'symbol' letters */
210 { 0x376, 0x377 },
211 { 0x39c, 0xb5 },
212 { 0x392, 0x3d0 },
213 { 0x398, 0x3d1 },
214 { 0x3a6, 0x3d5 },
215 { 0x3a0, 0x3d6 },
216 { 0x39a, 0x3f0 },
217 { 0x3a1, 0x3f1 },
218 { 0x395, 0x3f5 },
219 { 0x3cf, 0x3d7 },
220
221 { 0,0 }
222 };
223
224
__towcase(wchar_t wc,int lower)225 static wchar_t __towcase(wchar_t wc, int lower)
226 {
227 int i;
228 int lmul = 2*lower-1;
229 int lmask = lower-1;
230 /* no letters with case in these large ranges */
231 if (!iswalpha(wc)
232 || (unsigned)wc - 0x0600 <= 0x0fff-0x0600
233 || (unsigned)wc - 0x2e00 <= 0xa63f-0x2e00
234 || (unsigned)wc - 0xa800 <= 0xfeff-0xa800)
235 return wc;
236 /* special case because the diff between upper/lower is too big */
237 if (lower && (unsigned)wc - 0x10a0 < 0x2e)
238 if (wc>0x10c5 && wc != 0x10c7 && wc != 0x10cd) return wc;
239 else return wc + 0x2d00 - 0x10a0;
240 if (!lower && (unsigned)wc - 0x2d00 < 0x26)
241 if (wc>0x2d25 && wc != 0x2d27 && wc != 0x2d2d) return wc;
242 else return wc + 0x10a0 - 0x2d00;
243 for (i=0; casemaps[i].len; i++) {
244 int base = casemaps[i].upper + (lmask & casemaps[i].lower);
245 if ((unsigned)wc-base < casemaps[i].len) {
246 if (casemaps[i].lower == 1)
247 return wc + lower - ((wc-casemaps[i].upper)&1);
248 return wc + lmul*casemaps[i].lower;
249 }
250 }
251 for (i=0; pairs[i][1-lower]; i++) {
252 if (pairs[i][1-lower] == wc)
253 return pairs[i][lower];
254 }
255 if ((unsigned)wc - (0x10428 - 0x28*lower) < 0x28)
256 return wc - 0x28 + 0x50*lower;
257 return wc;
258 }
259
towupper(wint_t wc)260 wint_t towupper(wint_t wc)
261 {
262 return __towcase(wc, 0);
263 }
264
towlower(wint_t wc)265 wint_t towlower(wint_t wc)
266 {
267 return __towcase(wc, 1);
268 }
269