1 #include <stdint.h>
2 
3 static const uint32_t to_ucs4[256] = {
4   [0x01] = 0x0001,
5   [0x02] = 0x0002,
6   [0x03] = 0x0003,
7   [0x04] = 0x0004,
8   [0x05] = 0x0005,
9   [0x06] = 0x0006,
10   [0x07] = 0x0007,
11   [0x08] = 0x0008,
12   [0x09] = 0x0009,
13   [0x0a] = 0x000A,
14   [0x0b] = 0x000B,
15   [0x0c] = 0x000C,
16   [0x0d] = 0x000D,
17   [0x0e] = 0x000E,
18   [0x0f] = 0x000F,
19   [0x10] = 0x0010,
20   [0x11] = 0x0011,
21   [0x12] = 0x0012,
22   [0x13] = 0x0013,
23   [0x14] = 0x0014,
24   [0x15] = 0x0015,
25   [0x16] = 0x0016,
26   [0x17] = 0x0017,
27   [0x18] = 0x0018,
28   [0x19] = 0x0019,
29   [0x1a] = 0x001A,
30   [0x1b] = 0x001B,
31   [0x1c] = 0x001C,
32   [0x1d] = 0x001D,
33   [0x1e] = 0x001E,
34   [0x1f] = 0x001F,
35   [0x20] = 0x0020,
36   [0x21] = 0x0021,
37   [0x22] = 0x0022,
38   [0x23] = 0x0023,
39   [0x24] = 0x0024,
40   [0x25] = 0x0025,
41   [0x26] = 0x0026,
42   [0x27] = 0x0027,
43   [0x28] = 0x0028,
44   [0x29] = 0x0029,
45   [0x2a] = 0x002A,
46   [0x2b] = 0x002B,
47   [0x2c] = 0x002C,
48   [0x2d] = 0x002D,
49   [0x2e] = 0x002E,
50   [0x2f] = 0x002F,
51   [0x30] = 0x0030,
52   [0x31] = 0x0031,
53   [0x32] = 0x0032,
54   [0x33] = 0x0033,
55   [0x34] = 0x0034,
56   [0x35] = 0x0035,
57   [0x36] = 0x0036,
58   [0x37] = 0x0037,
59   [0x38] = 0x0038,
60   [0x39] = 0x0039,
61   [0x3a] = 0x003A,
62   [0x3b] = 0x003B,
63   [0x3c] = 0x003C,
64   [0x3d] = 0x003D,
65   [0x3e] = 0x003E,
66   [0x3f] = 0x003F,
67   [0x40] = 0x0040,
68   [0x41] = 0x0041,
69   [0x42] = 0x0042,
70   [0x43] = 0x0043,
71   [0x44] = 0x0044,
72   [0x45] = 0x0045,
73   [0x46] = 0x0046,
74   [0x47] = 0x0047,
75   [0x48] = 0x0048,
76   [0x49] = 0x0049,
77   [0x4a] = 0x004A,
78   [0x4b] = 0x004B,
79   [0x4c] = 0x004C,
80   [0x4d] = 0x004D,
81   [0x4e] = 0x004E,
82   [0x4f] = 0x004F,
83   [0x50] = 0x0050,
84   [0x51] = 0x0051,
85   [0x52] = 0x0052,
86   [0x53] = 0x0053,
87   [0x54] = 0x0054,
88   [0x55] = 0x0055,
89   [0x56] = 0x0056,
90   [0x57] = 0x0057,
91   [0x58] = 0x0058,
92   [0x59] = 0x0059,
93   [0x5a] = 0x005A,
94   [0x5b] = 0x005B,
95   [0x5c] = 0x005C,
96   [0x5d] = 0x005D,
97   [0x5e] = 0x005E,
98   [0x5f] = 0x005F,
99   [0x60] = 0x0060,
100   [0x61] = 0x0061,
101   [0x62] = 0x0062,
102   [0x63] = 0x0063,
103   [0x64] = 0x0064,
104   [0x65] = 0x0065,
105   [0x66] = 0x0066,
106   [0x67] = 0x0067,
107   [0x68] = 0x0068,
108   [0x69] = 0x0069,
109   [0x6a] = 0x006A,
110   [0x6b] = 0x006B,
111   [0x6c] = 0x006C,
112   [0x6d] = 0x006D,
113   [0x6e] = 0x006E,
114   [0x6f] = 0x006F,
115   [0x70] = 0x0070,
116   [0x71] = 0x0071,
117   [0x72] = 0x0072,
118   [0x73] = 0x0073,
119   [0x74] = 0x0074,
120   [0x75] = 0x0075,
121   [0x76] = 0x0076,
122   [0x77] = 0x0077,
123   [0x78] = 0x0078,
124   [0x79] = 0x0079,
125   [0x7a] = 0x007A,
126   [0x7b] = 0x007B,
127   [0x7c] = 0x007C,
128   [0x7d] = 0x007D,
129   [0x7e] = 0x007E,
130   [0x7f] = 0x007F,
131   [0x80] = 0x0000,
132   [0x81] = 0x0001,
133   [0x82] = 0x0002,
134   [0x83] = 0x0003,
135   [0x84] = 0x0004,
136   [0x85] = 0x0005,
137   [0x86] = 0x0006,
138   [0x87] = 0x0007,
139   [0x88] = 0x0008,
140   [0x89] = 0x0009,
141   [0x8a] = 0x000A,
142   [0x8b] = 0x000B,
143   [0x8c] = 0x000C,
144   [0x8d] = 0x000D,
145   [0x8e] = 0x000E,
146   [0x8f] = 0x000F,
147   [0x90] = 0x0010,
148   [0x91] = 0x0011,
149   [0x92] = 0x0012,
150   [0x93] = 0x0013,
151   [0x94] = 0x0014,
152   [0x95] = 0x0015,
153   [0x96] = 0x0016,
154   [0x97] = 0x0017,
155   [0x98] = 0x0018,
156   [0x99] = 0x0019,
157   [0x9a] = 0x001A,
158   [0x9b] = 0x001B,
159   [0x9c] = 0x001C,
160   [0x9d] = 0x001D,
161   [0x9e] = 0x001E,
162   [0x9f] = 0x001F,
163   [0xa0] = 0x0020,
164   [0xa1] = 0x200C,
165   [0xa2] = 0x200D,
166   [0xa3] = 0x0021,
167   [0xa4] = 0x00A4,
168   [0xa5] = 0x066A,
169   [0xa6] = 0x002E,
170   [0xa7] = 0x066C,
171   [0xa8] = 0x0029,
172   [0xa9] = 0x0028,
173   [0xaa] = 0x00D7,
174   [0xab] = 0x002B,
175   [0xac] = 0x060C,
176   [0xad] = 0x002D,
177   [0xae] = 0x066B,
178   [0xaf] = 0x002F,
179   [0xb0] = 0x06F0,
180   [0xb1] = 0x06F1,
181   [0xb2] = 0x06F2,
182   [0xb3] = 0x06F3,
183   [0xb4] = 0x06F4,
184   [0xb5] = 0x06F5,
185   [0xb6] = 0x06F6,
186   [0xb7] = 0x06F7,
187   [0xb8] = 0x06F8,
188   [0xb9] = 0x06F9,
189   [0xba] = 0x003A,
190   [0xbb] = 0x061B,
191   [0xbc] = 0x003C,
192   [0xbd] = 0x003D,
193   [0xbe] = 0x003E,
194   [0xbf] = 0x061F,
195   [0xc0] = 0x0622,
196   [0xc1] = 0x0627,
197   [0xc2] = 0x0621,
198   [0xc3] = 0x0628,
199   [0xc4] = 0x067E,
200   [0xc5] = 0x062A,
201   [0xc6] = 0x062B,
202   [0xc7] = 0x062C,
203   [0xc8] = 0x0686,
204   [0xc9] = 0x062D,
205   [0xca] = 0x062E,
206   [0xcb] = 0x062F,
207   [0xcc] = 0x0630,
208   [0xcd] = 0x0631,
209   [0xce] = 0x0632,
210   [0xcf] = 0x0698,
211   [0xd0] = 0x0633,
212   [0xd1] = 0x0634,
213   [0xd2] = 0x0635,
214   [0xd3] = 0x0636,
215   [0xd4] = 0x0637,
216   [0xd5] = 0x0638,
217   [0xd6] = 0x0639,
218   [0xd7] = 0x063A,
219   [0xd8] = 0x0641,
220   [0xd9] = 0x0642,
221   [0xda] = 0x06A9,
222   [0xdb] = 0x06AF,
223   [0xdc] = 0x0644,
224   [0xdd] = 0x0645,
225   [0xde] = 0x0646,
226   [0xdf] = 0x0648,
227   [0xe0] = 0x0647,
228   [0xe1] = 0x06CC,
229   [0xe2] = 0x005D,
230   [0xe3] = 0x005B,
231   [0xe4] = 0x007D,
232   [0xe5] = 0x007B,
233   [0xe6] = 0x00AB,
234   [0xe7] = 0x00BB,
235   [0xe8] = 0x002A,
236   [0xe9] = 0x0640,
237   [0xea] = 0x007C,
238   [0xeb] = 0x005C,
239   [0xf0] = 0x064E,
240   [0xf1] = 0x0650,
241   [0xf2] = 0x064F,
242   [0xf3] = 0x064B,
243   [0xf4] = 0x064D,
244   [0xf5] = 0x064C,
245   [0xf6] = 0x0651,
246   [0xf7] = 0x0652,
247   [0xf8] = 0x0623,
248   [0xf9] = 0x0624,
249   [0xfa] = 0x0625,
250   [0xfb] = 0x0626,
251   [0xfc] = 0x0629,
252   [0xfd] = 0x0643,
253   [0xfe] = 0x064A,
254   [0xff] = 0x007F,
255 };
256 static const struct gap from_idx[] = {
257   { .start = 0x0000, .end = 0x007f, .idx =     0 },
258   { .start = 0x00a4, .end = 0x00a4, .idx =   -36 },
259   { .start = 0x00ab, .end = 0x00ab, .idx =   -42 },
260   { .start = 0x00bb, .end = 0x00bb, .idx =   -57 },
261   { .start = 0x00d7, .end = 0x00d7, .idx =   -84 },
262   { .start = 0x060c, .end = 0x060c, .idx = -1416 },
263   { .start = 0x061b, .end = 0x0652, .idx = -1430 },
264   { .start = 0x066a, .end = 0x066c, .idx = -1453 },
265   { .start = 0x067e, .end = 0x067e, .idx = -1470 },
266   { .start = 0x0686, .end = 0x0686, .idx = -1477 },
267   { .start = 0x0698, .end = 0x0698, .idx = -1494 },
268   { .start = 0x06a9, .end = 0x06af, .idx = -1510 },
269   { .start = 0x06cc, .end = 0x06cc, .idx = -1538 },
270   { .start = 0x06f0, .end = 0x06f9, .idx = -1573 },
271   { .start = 0x200c, .end = 0x200d, .idx = -7991 },
272   { .start = 0xffff, .end = 0xffff, .idx =     0 }
273 };
274 static const char from_ucs4[] = {
275   /* 0x0000..0x007f */
276   '\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07',
277   '\x08', '\x09', '\x0a', '\x0b', '\x0c', '\x0d', '\x0e', '\x0f',
278   '\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17',
279   '\x18', '\x19', '\x1a', '\x1b', '\x1c', '\x1d', '\x1e', '\x1f',
280   '\x20', '\x21', '\x22', '\x23', '\x24', '\x25', '\x26', '\x27',
281   '\x28', '\x29', '\x2a', '\x2b', '\x2c', '\x2d', '\x2e', '\x2f',
282   '\x30', '\x31', '\x32', '\x33', '\x34', '\x35', '\x36', '\x37',
283   '\x38', '\x39', '\x3a', '\x3b', '\x3c', '\x3d', '\x3e', '\x3f',
284   '\x40', '\x41', '\x42', '\x43', '\x44', '\x45', '\x46', '\x47',
285   '\x48', '\x49', '\x4a', '\x4b', '\x4c', '\x4d', '\x4e', '\x4f',
286   '\x50', '\x51', '\x52', '\x53', '\x54', '\x55', '\x56', '\x57',
287   '\x58', '\x59', '\x5a', '\x5b', '\x5c', '\x5d', '\x5e', '\x5f',
288   '\x60', '\x61', '\x62', '\x63', '\x64', '\x65', '\x66', '\x67',
289   '\x68', '\x69', '\x6a', '\x6b', '\x6c', '\x6d', '\x6e', '\x6f',
290   '\x70', '\x71', '\x72', '\x73', '\x74', '\x75', '\x76', '\x77',
291   '\x78', '\x79', '\x7a', '\x7b', '\x7c', '\x7d', '\x7e', '\x7f',
292   /* 0x00a4..0x00a4 */
293   '\xa4',
294   /* 0x00ab..0x00ab */
295   '\xe6',
296   /* 0x00bb..0x00bb */
297   '\xe7',
298   /* 0x00d7..0x00d7 */
299   '\xaa',
300   /* 0x060c..0x060c */
301   '\xac',
302   /* 0x061b..0x0652 */
303   '\xbb', '\x00', '\x00', '\x00', '\xbf', '\x00', '\xc2', '\xc0',
304   '\xf8', '\xf9', '\xfa', '\xfb', '\xc1', '\xc3', '\xfc', '\xc5',
305   '\xc6', '\xc7', '\xc9', '\xca', '\xcb', '\xcc', '\xcd', '\xce',
306   '\xd0', '\xd1', '\xd2', '\xd3', '\xd4', '\xd5', '\xd6', '\xd7',
307   '\x00', '\x00', '\x00', '\x00', '\x00', '\xe9', '\xd8', '\xd9',
308   '\xfd', '\xdc', '\xdd', '\xde', '\xe0', '\xdf', '\x00', '\xfe',
309   '\xf3', '\xf5', '\xf4', '\xf0', '\xf2', '\xf1', '\xf6', '\xf7',
310   /* 0x066a..0x066c */
311   '\xa5', '\xae', '\xa7',
312   /* 0x067e..0x067e */
313   '\xc4',
314   /* 0x0686..0x0686 */
315   '\xc8',
316   /* 0x0698..0x0698 */
317   '\xcf',
318   /* 0x06a9..0x06af */
319   '\xda', '\x00', '\x00', '\x00', '\x00', '\x00', '\xdb',
320   /* 0x06cc..0x06cc */
321   '\xe1',
322   /* 0x06f0..0x06f9 */
323   '\xb0', '\xb1', '\xb2', '\xb3', '\xb4', '\xb5', '\xb6', '\xb7',
324   '\xb8', '\xb9',
325   /* 0x200c..0x200d */
326   '\xa1', '\xa2',
327 };
328