1static const uint32_t to_ucs4[256] = {
2 [0x01] = 0x0001,
3 [0x02] = 0x0002,
4 [0x03] = 0x0003,
5 [0x04] = 0x0004,
6 [0x05] = 0x0005,
7 [0x06] = 0x0006,
8 [0x07] = 0x0007,
9 [0x08] = 0x0008,
10 [0x09] = 0x0009,
11 [0x0a] = 0x000A,
12 [0x0b] = 0x000B,
13 [0x0c] = 0x000C,
14 [0x0d] = 0x000D,
15 [0x0e] = 0x000E,
16 [0x0f] = 0x000F,
17 [0x10] = 0x0010,
18 [0x11] = 0x0011,
19 [0x12] = 0x0012,
20 [0x13] = 0x0013,
21 [0x14] = 0x0014,
22 [0x15] = 0x0015,
23 [0x16] = 0x0016,
24 [0x17] = 0x0017,
25 [0x18] = 0x0018,
26 [0x19] = 0x0019,
27 [0x1a] = 0x001A,
28 [0x1b] = 0x001B,
29 [0x1c] = 0x001C,
30 [0x1d] = 0x001D,
31 [0x1e] = 0x001E,
32 [0x1f] = 0x001F,
33 [0x20] = 0x0020,
34 [0x21] = 0x0021,
35 [0x22] = 0x0022,
36 [0x23] = 0x0023,
37 [0x24] = 0x0024,
38 [0x25] = 0x0025,
39 [0x26] = 0x0026,
40 [0x27] = 0x0027,
41 [0x28] = 0x0028,
42 [0x29] = 0x0029,
43 [0x2a] = 0x002A,
44 [0x2b] = 0x002B,
45 [0x2c] = 0x002C,
46 [0x2d] = 0x002D,
47 [0x2e] = 0x002E,
48 [0x2f] = 0x002F,
49 [0x30] = 0x0030,
50 [0x31] = 0x0031,
51 [0x32] = 0x0032,
52 [0x33] = 0x0033,
53 [0x34] = 0x0034,
54 [0x35] = 0x0035,
55 [0x36] = 0x0036,
56 [0x37] = 0x0037,
57 [0x38] = 0x0038,
58 [0x39] = 0x0039,
59 [0x3a] = 0x003A,
60 [0x3b] = 0x003B,
61 [0x3c] = 0x003C,
62 [0x3d] = 0x003D,
63 [0x3e] = 0x003E,
64 [0x3f] = 0x003F,
65 [0x40] = 0x0040,
66 [0x41] = 0x0041,
67 [0x42] = 0x0042,
68 [0x43] = 0x0043,
69 [0x44] = 0x0044,
70 [0x45] = 0x0045,
71 [0x46] = 0x0046,
72 [0x47] = 0x0047,
73 [0x48] = 0x0048,
74 [0x49] = 0x0049,
75 [0x4a] = 0x004A,
76 [0x4b] = 0x004B,
77 [0x4c] = 0x004C,
78 [0x4d] = 0x004D,
79 [0x4e] = 0x004E,
80 [0x4f] = 0x004F,
81 [0x50] = 0x0050,
82 [0x51] = 0x0051,
83 [0x52] = 0x0052,
84 [0x53] = 0x0053,
85 [0x54] = 0x0054,
86 [0x55] = 0x0055,
87 [0x56] = 0x0056,
88 [0x57] = 0x0057,
89 [0x58] = 0x0058,
90 [0x59] = 0x0059,
91 [0x5a] = 0x005A,
92 [0x5b] = 0x005B,
93 [0x5c] = 0x005C,
94 [0x5d] = 0x005D,
95 [0x5e] = 0x005E,
96 [0x5f] = 0x005F,
97 [0x60] = 0x0060,
98 [0x61] = 0x0061,
99 [0x62] = 0x0062,
100 [0x63] = 0x0063,
101 [0x64] = 0x0064,
102 [0x65] = 0x0065,
103 [0x66] = 0x0066,
104 [0x67] = 0x0067,
105 [0x68] = 0x0068,
106 [0x69] = 0x0069,
107 [0x6a] = 0x006A,
108 [0x6b] = 0x006B,
109 [0x6c] = 0x006C,
110 [0x6d] = 0x006D,
111 [0x6e] = 0x006E,
112 [0x6f] = 0x006F,
113 [0x70] = 0x0070,
114 [0x71] = 0x0071,
115 [0x72] = 0x0072,
116 [0x73] = 0x0073,
117 [0x74] = 0x0074,
118 [0x75] = 0x0075,
119 [0x76] = 0x0076,
120 [0x77] = 0x0077,
121 [0x78] = 0x0078,
122 [0x79] = 0x0079,
123 [0x7a] = 0x007A,
124 [0x7b] = 0x007B,
125 [0x7c] = 0x007C,
126 [0x7d] = 0x007D,
127 [0x7e] = 0x007E,
128 [0x7f] = 0x007F,
129 [0x80] = 0x20AC,
130 [0x82] = 0x201A,
131 [0x84] = 0x201E,
132 [0x85] = 0x2026,
133 [0x86] = 0x2020,
134 [0x87] = 0x2021,
135 [0x89] = 0x2030,
136 [0x8a] = 0x0160,
137 [0x8b] = 0x2039,
138 [0x8c] = 0x015A,
139 [0x8d] = 0x0164,
140 [0x8e] = 0x017D,
141 [0x8f] = 0x0179,
142 [0x91] = 0x2018,
143 [0x92] = 0x2019,
144 [0x93] = 0x201C,
145 [0x94] = 0x201D,
146 [0x95] = 0x2022,
147 [0x96] = 0x2013,
148 [0x97] = 0x2014,
149 [0x99] = 0x2122,
150 [0x9a] = 0x0161,
151 [0x9b] = 0x203A,
152 [0x9c] = 0x015B,
153 [0x9d] = 0x0165,
154 [0x9e] = 0x017E,
155 [0x9f] = 0x017A,
156 [0xa0] = 0x00A0,
157 [0xa1] = 0x02C7,
158 [0xa2] = 0x02D8,
159 [0xa3] = 0x0141,
160 [0xa4] = 0x00A4,
161 [0xa5] = 0x0104,
162 [0xa6] = 0x00A6,
163 [0xa7] = 0x00A7,
164 [0xa8] = 0x00A8,
165 [0xa9] = 0x00A9,
166 [0xaa] = 0x015E,
167 [0xab] = 0x00AB,
168 [0xac] = 0x00AC,
169 [0xad] = 0x00AD,
170 [0xae] = 0x00AE,
171 [0xaf] = 0x017B,
172 [0xb0] = 0x00B0,
173 [0xb1] = 0x00B1,
174 [0xb2] = 0x02DB,
175 [0xb3] = 0x0142,
176 [0xb4] = 0x00B4,
177 [0xb5] = 0x00B5,
178 [0xb6] = 0x00B6,
179 [0xb7] = 0x00B7,
180 [0xb8] = 0x00B8,
181 [0xb9] = 0x0105,
182 [0xba] = 0x015F,
183 [0xbb] = 0x00BB,
184 [0xbc] = 0x013D,
185 [0xbd] = 0x02DD,
186 [0xbe] = 0x013E,
187 [0xbf] = 0x017C,
188 [0xc0] = 0x0154,
189 [0xc1] = 0x00C1,
190 [0xc2] = 0x00C2,
191 [0xc3] = 0x0102,
192 [0xc4] = 0x00C4,
193 [0xc5] = 0x0139,
194 [0xc6] = 0x0106,
195 [0xc7] = 0x00C7,
196 [0xc8] = 0x010C,
197 [0xc9] = 0x00C9,
198 [0xca] = 0x0118,
199 [0xcb] = 0x00CB,
200 [0xcc] = 0x011A,
201 [0xcd] = 0x00CD,
202 [0xce] = 0x00CE,
203 [0xcf] = 0x010E,
204 [0xd0] = 0x0110,
205 [0xd1] = 0x0143,
206 [0xd2] = 0x0147,
207 [0xd3] = 0x00D3,
208 [0xd4] = 0x00D4,
209 [0xd5] = 0x0150,
210 [0xd6] = 0x00D6,
211 [0xd7] = 0x00D7,
212 [0xd8] = 0x0158,
213 [0xd9] = 0x016E,
214 [0xda] = 0x00DA,
215 [0xdb] = 0x0170,
216 [0xdc] = 0x00DC,
217 [0xdd] = 0x00DD,
218 [0xde] = 0x0162,
219 [0xdf] = 0x00DF,
220 [0xe0] = 0x0155,
221 [0xe1] = 0x00E1,
222 [0xe2] = 0x00E2,
223 [0xe3] = 0x0103,
224 [0xe4] = 0x00E4,
225 [0xe5] = 0x013A,
226 [0xe6] = 0x0107,
227 [0xe7] = 0x00E7,
228 [0xe8] = 0x010D,
229 [0xe9] = 0x00E9,
230 [0xea] = 0x0119,
231 [0xeb] = 0x00EB,
232 [0xec] = 0x011B,
233 [0xed] = 0x00ED,
234 [0xee] = 0x00EE,
235 [0xef] = 0x010F,
236 [0xf0] = 0x0111,
237 [0xf1] = 0x0144,
238 [0xf2] = 0x0148,
239 [0xf3] = 0x00F3,
240 [0xf4] = 0x00F4,
241 [0xf5] = 0x0151,
242 [0xf6] = 0x00F6,
243 [0xf7] = 0x00F7,
244 [0xf8] = 0x0159,
245 [0xf9] = 0x016F,
246 [0xfa] = 0x00FA,
247 [0xfb] = 0x0171,
248 [0xfc] = 0x00FC,
249 [0xfd] = 0x00FD,
250 [0xfe] = 0x0163,
251 [0xff] = 0x02D9,
252};
253static const struct gap from_idx[] = {
254 { .start = 0x0000, .end = 0x007f, .idx = 0 },
255 { .start = 0x00a0, .end = 0x0111, .idx = -32 },
256 { .start = 0x0118, .end = 0x011b, .idx = -38 },
257 { .start = 0x0139, .end = 0x0148, .idx = -67 },
258 { .start = 0x0150, .end = 0x0165, .idx = -74 },
259 { .start = 0x016e, .end = 0x0171, .idx = -82 },
260 { .start = 0x0179, .end = 0x017e, .idx = -89 },
261 { .start = 0x02c7, .end = 0x02c7, .idx = -417 },
262 { .start = 0x02d8, .end = 0x02dd, .idx = -433 },
263 { .start = 0x2013, .end = 0x2026, .idx = -7910 },
264 { .start = 0x2030, .end = 0x2030, .idx = -7919 },
265 { .start = 0x2039, .end = 0x203a, .idx = -7927 },
266 { .start = 0x20ac, .end = 0x20ac, .idx = -8040 },
267 { .start = 0x2122, .end = 0x2122, .idx = -8157 },
268 { .start = 0xffff, .end = 0xffff, .idx = 0 }
269};
270static const char from_ucs4[] = {
271
272 '\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07',
273 '\x08', '\x09', '\x0a', '\x0b', '\x0c', '\x0d', '\x0e', '\x0f',
274 '\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17',
275 '\x18', '\x19', '\x1a', '\x1b', '\x1c', '\x1d', '\x1e', '\x1f',
276 '\x20', '\x21', '\x22', '\x23', '\x24', '\x25', '\x26', '\x27',
277 '\x28', '\x29', '\x2a', '\x2b', '\x2c', '\x2d', '\x2e', '\x2f',
278 '\x30', '\x31', '\x32', '\x33', '\x34', '\x35', '\x36', '\x37',
279 '\x38', '\x39', '\x3a', '\x3b', '\x3c', '\x3d', '\x3e', '\x3f',
280 '\x40', '\x41', '\x42', '\x43', '\x44', '\x45', '\x46', '\x47',
281 '\x48', '\x49', '\x4a', '\x4b', '\x4c', '\x4d', '\x4e', '\x4f',
282 '\x50', '\x51', '\x52', '\x53', '\x54', '\x55', '\x56', '\x57',
283 '\x58', '\x59', '\x5a', '\x5b', '\x5c', '\x5d', '\x5e', '\x5f',
284 '\x60', '\x61', '\x62', '\x63', '\x64', '\x65', '\x66', '\x67',
285 '\x68', '\x69', '\x6a', '\x6b', '\x6c', '\x6d', '\x6e', '\x6f',
286 '\x70', '\x71', '\x72', '\x73', '\x74', '\x75', '\x76', '\x77',
287 '\x78', '\x79', '\x7a', '\x7b', '\x7c', '\x7d', '\x7e', '\x7f',
288 '\xa0', '\x00', '\x00', '\x00', '\xa4', '\x00', '\xa6', '\xa7',
289 '\xa8', '\xa9', '\x00', '\xab', '\xac', '\xad', '\xae', '\x00',
290 '\xb0', '\xb1', '\x00', '\x00', '\xb4', '\xb5', '\xb6', '\xb7',
291 '\xb8', '\x00', '\x00', '\xbb', '\x00', '\x00', '\x00', '\x00',
292 '\x00', '\xc1', '\xc2', '\x00', '\xc4', '\x00', '\x00', '\xc7',
293 '\x00', '\xc9', '\x00', '\xcb', '\x00', '\xcd', '\xce', '\x00',
294 '\x00', '\x00', '\x00', '\xd3', '\xd4', '\x00', '\xd6', '\xd7',
295 '\x00', '\x00', '\xda', '\x00', '\xdc', '\xdd', '\x00', '\xdf',
296 '\x00', '\xe1', '\xe2', '\x00', '\xe4', '\x00', '\x00', '\xe7',
297 '\x00', '\xe9', '\x00', '\xeb', '\x00', '\xed', '\xee', '\x00',
298 '\x00', '\x00', '\x00', '\xf3', '\xf4', '\x00', '\xf6', '\xf7',
299 '\x00', '\x00', '\xfa', '\x00', '\xfc', '\xfd', '\x00', '\x00',
300 '\x00', '\x00', '\xc3', '\xe3', '\xa5', '\xb9', '\xc6', '\xe6',
301 '\x00', '\x00', '\x00', '\x00', '\xc8', '\xe8', '\xcf', '\xef',
302 '\xd0', '\xf0', '\xca', '\xea', '\xcc', '\xec', '\xc5', '\xe5',
303 '\x00', '\x00', '\xbc', '\xbe', '\x00', '\x00', '\xa3', '\xb3',
304 '\xd1', '\xf1', '\x00', '\x00', '\xd2', '\xf2', '\xd5', '\xf5',
305 '\x00', '\x00', '\xc0', '\xe0', '\x00', '\x00', '\xd8', '\xf8',
306 '\x8c', '\x9c', '\x00', '\x00', '\xaa', '\xba', '\x8a', '\x9a',
307 '\xde', '\xfe', '\x8d', '\x9d', '\xd9', '\xf9', '\xdb', '\xfb',
308 '\x8f', '\x9f', '\xaf', '\xbf', '\x8e', '\x9e', '\xa1', '\xa2',
309 '\xff', '\x00', '\xb2', '\x00', '\xbd', '\x96', '\x97', '\x00',
310 '\x00', '\x00', '\x91', '\x92', '\x82', '\x00', '\x93', '\x94',
311 '\x84', '\x00', '\x86', '\x87', '\x95', '\x00', '\x00', '\x00',
312 '\x85', '\x89', '\x8b', '\x9b', '\x80', '\x99',
313};
314