1/* Conversion from and to CP1258.
2 Copyright (C) 1998-2021 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998,
5 and Bruno Haible <haible@clisp.cons.org>, 2001.
6
7 The GNU C Library is free software; you can redistribute it and/or
8 modify it under the terms of the GNU Lesser General Public
9 License as published by the Free Software Foundation; either
10 version 2.1 of the License, or (at your option) any later version.
11
12 The GNU C Library is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 Lesser General Public License for more details.
16
17 You should have received a copy of the GNU Lesser General Public
18 License along with the GNU C Library; if not, see
19 <https://www.gnu.org/licenses/>. */
20
21#include <dlfcn.h>
22#include <stdint.h>
23#include <assert.h>
24#include <stdlib.h>
25
26#define NELEMS(arr) (sizeof (arr) / sizeof (arr[0]))
27
28/* Definitions used in the body of the `gconv' function. */
29#define CHARSET_NAME "CP1258//"
30#define FROM_LOOP from_cp1258
31#define TO_LOOP to_cp1258
32#define DEFINE_INIT 1
33#define DEFINE_FINI 1
34#define ONE_DIRECTION 0
35#define FROM_LOOP_MIN_NEEDED_FROM 1
36#define FROM_LOOP_MAX_NEEDED_FROM 1
37#define FROM_LOOP_MIN_NEEDED_TO 4
38#define FROM_LOOP_MAX_NEEDED_TO 4
39#define TO_LOOP_MIN_NEEDED_FROM 4
40#define TO_LOOP_MAX_NEEDED_FROM 4
41#define TO_LOOP_MIN_NEEDED_TO 1
42#define TO_LOOP_MAX_NEEDED_TO 2
43#define PREPARE_LOOP \
44 int saved_state; \
45 int *statep = &data->__statep->__count;
46#define EXTRA_LOOP_ARGS , statep
47
48
49/* Since we might have to reset input pointer we must be able to save
50 and restore the state. */
51#define SAVE_RESET_STATE(Save) \
52 if (Save) \
53 saved_state = *statep; \
54 else \
55 *statep = saved_state
56
57
58/* During CP1258 to UCS4 conversion, the COUNT element of the state
59 contains the last UCS4 character, shifted by 3 bits. */
60
61
62/* Since this is a stateful encoding we have to provide code which resets
63 the output state to the initial state. This has to be done during the
64 flushing. */
65#define EMIT_SHIFT_TO_INIT \
66 if (data->__statep->__count != 0) \
67 { \
68 if (FROM_DIRECTION) \
69 { \
70 if (__glibc_likely (outbuf + 4 <= outend)) \
71 { \
72 /* Write out the last character. */ \
73 *((uint32_t *) outbuf) = data->__statep->__count >> 3; \
74 outbuf += sizeof (uint32_t); \
75 data->__statep->__count = 0; \
76 } \
77 else \
78 /* We don't have enough room in the output buffer. */ \
79 status = __GCONV_FULL_OUTPUT; \
80 } \
81 else \
82 /* We don't use shift states in the TO_DIRECTION. */ \
83 data->__statep->__count = 0; \
84 }
85
86
87/* First define the conversion function from CP1258 to UCS4. */
88
89static const uint16_t to_ucs4[128] =
90 {
91 /* 0x80 */
92 0x20AC, 0, 0x201A, 0x0192, 0x201E, 0x2026, 0x2020, 0x2021,
93 0x02C6, 0x2030, 0, 0x2039, 0x0152, 0, 0, 0,
94 /* 0x90 */
95 0, 0x2018, 0x2019, 0x201C, 0x201D, 0x2022, 0x2013, 0x2014,
96 0x02DC, 0x2122, 0, 0x203A, 0x0153, 0, 0, 0x0178,
97 /* 0xA0 */
98 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
99 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
100 /* 0xB0 */
101 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7,
102 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
103 /* 0xC0 */
104 0x00C0, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x00C5, 0x00C6, 0x00C7,
105 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x0300, 0x00CD, 0x00CE, 0x00CF,
106 /* 0xD0 */
107 0x0110, 0x00D1, 0x0309, 0x00D3, 0x00D4, 0x01A0, 0x00D6, 0x00D7,
108 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x01AF, 0x0303, 0x00DF,
109 /* 0xE0 */
110 0x00E0, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
111 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x0301, 0x00ED, 0x00EE, 0x00EF,
112 /* 0xF0 */
113 0x0111, 0x00F1, 0x0323, 0x00F3, 0x00F4, 0x01A1, 0x00F6, 0x00F7,
114 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x01B0, 0x20AB, 0x00FF,
115 };
116
117/* CP1258 contains five combining characters:
118 0x0300, 0x0301, 0x0303, 0x0309, 0x0323. */
119
120/* Composition tables for each of the relevant combining characters. */
121static const struct
122{
123 uint16_t base;
124 uint16_t composed;
125} comp_table_data[] =
126 {
127#define COMP_TABLE_IDX_0300 0
128#define COMP_TABLE_LEN_0300 31
129 { 0x0041, 0x00C0 },
130 { 0x0045, 0x00C8 },
131 { 0x0049, 0x00CC },
132 { 0x004E, 0x01F8 },
133 { 0x004F, 0x00D2 },
134 { 0x0055, 0x00D9 },
135 { 0x0057, 0x1E80 },
136 { 0x0059, 0x1EF2 },
137 { 0x0061, 0x00E0 },
138 { 0x0065, 0x00E8 },
139 { 0x0069, 0x00EC },
140 { 0x006E, 0x01F9 },
141 { 0x006F, 0x00F2 },
142 { 0x0075, 0x00F9 },
143 { 0x0077, 0x1E81 },
144 { 0x0079, 0x1EF3 },
145 { 0x00A8, 0x1FED },
146 { 0x00C2, 0x1EA6 },
147 { 0x00CA, 0x1EC0 },
148 { 0x00D4, 0x1ED2 },
149 { 0x00DC, 0x01DB },
150 { 0x00E2, 0x1EA7 },
151 { 0x00EA, 0x1EC1 },
152 { 0x00F4, 0x1ED3 },
153 { 0x00FC, 0x01DC },
154 { 0x0102, 0x1EB0 },
155 { 0x0103, 0x1EB1 },
156 /*{ 0x0112, 0x1E14 },*/
157 /*{ 0x0113, 0x1E15 },*/
158 /*{ 0x014C, 0x1E50 },*/
159 /*{ 0x014D, 0x1E51 },*/
160 { 0x01A0, 0x1EDC },
161 { 0x01A1, 0x1EDD },
162 { 0x01AF, 0x1EEA },
163 { 0x01B0, 0x1EEB },
164#define COMP_TABLE_IDX_0301 (COMP_TABLE_IDX_0300 + COMP_TABLE_LEN_0300)
165#define COMP_TABLE_LEN_0301 59
166 { 0x0041, 0x00C1 },
167 { 0x0043, 0x0106 },
168 { 0x0045, 0x00C9 },
169 { 0x0047, 0x01F4 },
170 { 0x0049, 0x00CD },
171 { 0x004B, 0x1E30 },
172 { 0x004C, 0x0139 },
173 { 0x004D, 0x1E3E },
174 { 0x004E, 0x0143 },
175 { 0x004F, 0x00D3 },
176 { 0x0050, 0x1E54 },
177 { 0x0052, 0x0154 },
178 { 0x0053, 0x015A },
179 { 0x0055, 0x00DA },
180 { 0x0057, 0x1E82 },
181 { 0x0059, 0x00DD },
182 { 0x005A, 0x0179 },
183 { 0x0061, 0x00E1 },
184 { 0x0063, 0x0107 },
185 { 0x0065, 0x00E9 },
186 { 0x0067, 0x01F5 },
187 { 0x0069, 0x00ED },
188 { 0x006B, 0x1E31 },
189 { 0x006C, 0x013A },
190 { 0x006D, 0x1E3F },
191 { 0x006E, 0x0144 },
192 { 0x006F, 0x00F3 },
193 { 0x0070, 0x1E55 },
194 { 0x0072, 0x0155 },
195 { 0x0073, 0x015B },
196 { 0x0075, 0x00FA },
197 { 0x0077, 0x1E83 },
198 { 0x0079, 0x00FD },
199 { 0x007A, 0x017A },
200 { 0x00A8, 0x0385 }, /* prefer U+0385 over U+1FEE */
201 { 0x00C2, 0x1EA4 },
202 { 0x00C5, 0x01FA },
203 { 0x00C6, 0x01FC },
204 { 0x00C7, 0x1E08 },
205 { 0x00CA, 0x1EBE },
206 { 0x00CF, 0x1E2E },
207 { 0x00D4, 0x1ED0 },
208 /*{ 0x00D5, 0x1E4C },*/
209 { 0x00D8, 0x01FE },
210 { 0x00DC, 0x01D7 },
211 { 0x00E2, 0x1EA5 },
212 { 0x00E5, 0x01FB },
213 { 0x00E6, 0x01FD },
214 { 0x00E7, 0x1E09 },
215 { 0x00EA, 0x1EBF },
216 { 0x00EF, 0x1E2F },
217 { 0x00F4, 0x1ED1 },
218 /*{ 0x00F5, 0x1E4D },*/
219 { 0x00F8, 0x01FF },
220 { 0x00FC, 0x01D8 },
221 { 0x0102, 0x1EAE },
222 { 0x0103, 0x1EAF },
223 /*{ 0x0112, 0x1E16 },*/
224 /*{ 0x0113, 0x1E17 },*/
225 /*{ 0x014C, 0x1E52 },*/
226 /*{ 0x014D, 0x1E53 },*/
227 /*{ 0x0168, 0x1E78 },*/
228 /*{ 0x0169, 0x1E79 },*/
229 { 0x01A0, 0x1EDA },
230 { 0x01A1, 0x1EDB },
231 { 0x01AF, 0x1EE8 },
232 { 0x01B0, 0x1EE9 },
233#define COMP_TABLE_IDX_0303 (COMP_TABLE_IDX_0301 + COMP_TABLE_LEN_0301)
234#define COMP_TABLE_LEN_0303 34
235 { 0x0041, 0x00C3 },
236 { 0x0045, 0x1EBC },
237 { 0x0049, 0x0128 },
238 { 0x004E, 0x00D1 },
239 { 0x004F, 0x00D5 },
240 { 0x0055, 0x0168 },
241 { 0x0056, 0x1E7C },
242 { 0x0059, 0x1EF8 },
243 { 0x0061, 0x00E3 },
244 { 0x0065, 0x1EBD },
245 { 0x0069, 0x0129 },
246 { 0x006E, 0x00F1 },
247 { 0x006F, 0x00F5 },
248 { 0x0075, 0x0169 },
249 { 0x0076, 0x1E7D },
250 { 0x0079, 0x1EF9 },
251 { 0x00C2, 0x1EAA },
252 { 0x00CA, 0x1EC4 },
253 { 0x00D3, 0x1E4C },
254 { 0x00D4, 0x1ED6 },
255 { 0x00D6, 0x1E4E },
256 { 0x00DA, 0x1E78 },
257 { 0x00E2, 0x1EAB },
258 { 0x00EA, 0x1EC5 },
259 { 0x00F3, 0x1E4D },
260 { 0x00F4, 0x1ED7 },
261 { 0x00F6, 0x1E4F },
262 { 0x00FA, 0x1E79 },
263 { 0x0102, 0x1EB4 },
264 { 0x0103, 0x1EB5 },
265 { 0x01A0, 0x1EE0 },
266 { 0x01A1, 0x1EE1 },
267 { 0x01AF, 0x1EEE },
268 { 0x01B0, 0x1EEF },
269#define COMP_TABLE_IDX_0309 (COMP_TABLE_IDX_0303 + COMP_TABLE_LEN_0303)
270#define COMP_TABLE_LEN_0309 24
271 { 0x0041, 0x1EA2 },
272 { 0x0045, 0x1EBA },
273 { 0x0049, 0x1EC8 },
274 { 0x004F, 0x1ECE },
275 { 0x0055, 0x1EE6 },
276 { 0x0059, 0x1EF6 },
277 { 0x0061, 0x1EA3 },
278 { 0x0065, 0x1EBB },
279 { 0x0069, 0x1EC9 },
280 { 0x006F, 0x1ECF },
281 { 0x0075, 0x1EE7 },
282 { 0x0079, 0x1EF7 },
283 { 0x00C2, 0x1EA8 },
284 { 0x00CA, 0x1EC2 },
285 { 0x00D4, 0x1ED4 },
286 { 0x00E2, 0x1EA9 },
287 { 0x00EA, 0x1EC3 },
288 { 0x00F4, 0x1ED5 },
289 { 0x0102, 0x1EB2 },
290 { 0x0103, 0x1EB3 },
291 { 0x01A0, 0x1EDE },
292 { 0x01A1, 0x1EDF },
293 { 0x01AF, 0x1EEC },
294 { 0x01B0, 0x1EED },
295#define COMP_TABLE_IDX_0323 (COMP_TABLE_IDX_0309 + COMP_TABLE_LEN_0309)
296#define COMP_TABLE_LEN_0323 50
297 { 0x0041, 0x1EA0 },
298 { 0x0042, 0x1E04 },
299 { 0x0044, 0x1E0C },
300 { 0x0045, 0x1EB8 },
301 { 0x0048, 0x1E24 },
302 { 0x0049, 0x1ECA },
303 { 0x004B, 0x1E32 },
304 { 0x004C, 0x1E36 },
305 { 0x004D, 0x1E42 },
306 { 0x004E, 0x1E46 },
307 { 0x004F, 0x1ECC },
308 { 0x0052, 0x1E5A },
309 { 0x0053, 0x1E62 },
310 { 0x0054, 0x1E6C },
311 { 0x0055, 0x1EE4 },
312 { 0x0056, 0x1E7E },
313 { 0x0057, 0x1E88 },
314 { 0x0059, 0x1EF4 },
315 { 0x005A, 0x1E92 },
316 { 0x0061, 0x1EA1 },
317 { 0x0062, 0x1E05 },
318 { 0x0064, 0x1E0D },
319 { 0x0065, 0x1EB9 },
320 { 0x0068, 0x1E25 },
321 { 0x0069, 0x1ECB },
322 { 0x006B, 0x1E33 },
323 { 0x006C, 0x1E37 },
324 { 0x006D, 0x1E43 },
325 { 0x006E, 0x1E47 },
326 { 0x006F, 0x1ECD },
327 { 0x0072, 0x1E5B },
328 { 0x0073, 0x1E63 },
329 { 0x0074, 0x1E6D },
330 { 0x0075, 0x1EE5 },
331 { 0x0076, 0x1E7F },
332 { 0x0077, 0x1E89 },
333 { 0x0079, 0x1EF5 },
334 { 0x007A, 0x1E93 },
335 { 0x00C2, 0x1EAC },
336 { 0x00CA, 0x1EC6 },
337 { 0x00D4, 0x1ED8 },
338 { 0x00E2, 0x1EAD },
339 { 0x00EA, 0x1EC7 },
340 { 0x00F4, 0x1ED9 },
341 { 0x0102, 0x1EB6 },
342 { 0x0103, 0x1EB7 },
343 { 0x01A0, 0x1EE2 },
344 { 0x01A1, 0x1EE3 },
345 { 0x01AF, 0x1EF0 },
346 { 0x01B0, 0x1EF1 },
347#define COMP_TABLE_IDX_END (COMP_TABLE_IDX_0323 + COMP_TABLE_LEN_0323)
348 };
349/* Compile-time verification of table size. */
350typedef int verify1[(NELEMS (comp_table_data) == COMP_TABLE_IDX_END) - 1];
351
352static const struct
353{
354 unsigned int idx;
355 unsigned int len;
356} comp_table[5] =
357 {
358 { COMP_TABLE_IDX_0300, COMP_TABLE_LEN_0300 },
359 { COMP_TABLE_IDX_0301, COMP_TABLE_LEN_0301 },
360 { COMP_TABLE_IDX_0303, COMP_TABLE_LEN_0303 },
361 { COMP_TABLE_IDX_0309, COMP_TABLE_LEN_0309 },
362 { COMP_TABLE_IDX_0323, COMP_TABLE_LEN_0323 }
363 };
364
365#define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM
366#define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM
367#define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO
368#define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO
369#define LOOPFCT FROM_LOOP
370#define BODY \
371 { \
372 uint32_t ch = *inptr; \
373 uint32_t last_ch; \
374 int must_buffer_ch; \
375 \
376 if (ch >= 0x80) \
377 { \
378 ch = to_ucs4[ch - 0x80]; \
379 if (__glibc_unlikely (ch == L'\0')) \
380 { \
381 /* This is an illegal character. */ \
382 STANDARD_FROM_LOOP_ERR_HANDLER (1); \
383 } \
384 } \
385 \
386 /* Determine whether there is a buffered character pending. */ \
387 last_ch = *statep >> 3; \
388 \
389 /* We have to buffer ch if it is a possible match in comp_table_data. */ \
390 must_buffer_ch = (ch >= 0x0041 && ch <= 0x01b0); \
391 \
392 if (last_ch) \
393 { \
394 if (ch >= 0x0300 && ch < 0x0340) \
395 { \
396 /* See whether last_ch and ch can be combined. */ \
397 unsigned int i, i1, i2; \
398 \
399 switch (ch) \
400 { \
401 case 0x0300: \
402 i = 0; \
403 break; \
404 case 0x0301: \
405 i = 1; \
406 break; \
407 case 0x0303: \
408 i = 2; \
409 break; \
410 case 0x0309: \
411 i = 3; \
412 break; \
413 case 0x0323: \
414 i = 4; \
415 break; \
416 default: \
417 abort (); \
418 } \
419 \
420 i1 = comp_table[i].idx; \
421 i2 = i1 + comp_table[i].len - 1; \
422 \
423 if (last_ch >= comp_table_data[i1].base \
424 && last_ch <= comp_table_data[i2].base) \
425 { \
426 for (;;) \
427 { \
428 i = (i1 + i2) >> 1; \
429 if (last_ch == comp_table_data[i].base) \
430 break; \
431 if (last_ch < comp_table_data[i].base) \
432 { \
433 if (i1 == i) \
434 goto not_combining; \
435 i2 = i; \
436 } \
437 else \
438 { \
439 if (i1 != i) \
440 i1 = i; \
441 else \
442 { \
443 i = i2; \
444 if (last_ch == comp_table_data[i].base) \
445 break; \
446 goto not_combining; \
447 } \
448 } \
449 } \
450 last_ch = comp_table_data[i].composed; \
451 /* Output the combined character. */ \
452 put32 (outptr, last_ch); \
453 outptr += 4; \
454 *statep = 0; \
455 ++inptr; \
456 continue; \
457 } \
458 } \
459 \
460 not_combining: \
461 /* Output the buffered character. */ \
462 put32 (outptr, last_ch); \
463 outptr += 4; \
464 *statep = 0; \
465 \
466 /* If we don't have enough room to output ch as well, then deal \
467 with it in another round. */ \
468 if (!must_buffer_ch && __builtin_expect (outptr + 4 > outend, 0)) \
469 continue; \
470 } \
471 \
472 if (must_buffer_ch) \
473 *statep = ch << 3; \
474 else \
475 { \
476 put32 (outptr, ch); \
477 outptr += 4; \
478 } \
479 ++inptr; \
480 }
481#define LOOP_NEED_FLAGS
482#define EXTRA_LOOP_DECLS , int *statep
483#define ONEBYTE_BODY \
484 { \
485 uint32_t ch; \
486 \
487 if (c < 0x80) \
488 ch = c; \
489 else \
490 { \
491 ch = to_ucs4[c - 0x80]; \
492 if (ch == L'\0') \
493 return WEOF; \
494 } \
495 if (ch >= 0x0041 && ch <= 0x01b0) \
496 return WEOF; \
497 return ch; \
498 }
499#include <iconv/loop.c>
500
501
502/* Next, define the conversion function from UCS4 to CP1258. */
503
504static const unsigned char from_ucs4[] =
505 {
506#define FROM_IDX_00 0
507 0xc4, 0xc5, 0xc6, 0xc7, /* 0x00c4-0x00c7 */
508 0xc8, 0xc9, 0xca, 0xcb, 0x00, 0xcd, 0xce, 0xcf, /* 0x00c8-0x00cf */
509 0x00, 0xd1, 0x00, 0xd3, 0xd4, 0x00, 0xd6, 0xd7, /* 0x00d0-0x00d7 */
510 0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0x00, 0x00, 0xdf, /* 0x00d8-0x00df */
511 0xe0, 0xe1, 0xe2, 0x00, 0xe4, 0xe5, 0xe6, 0xe7, /* 0x00e0-0x00e7 */
512 0xe8, 0xe9, 0xea, 0xeb, 0x00, 0xed, 0xee, 0xef, /* 0x00e8-0x00ef */
513 0x00, 0xf1, 0x00, 0xf3, 0xf4, 0x00, 0xf6, 0xf7, /* 0x00f0-0x00f7 */
514 0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0x00, 0x00, 0xff, /* 0x00f8-0x00ff */
515 0x00, 0x00, 0xc3, 0xe3, 0x00, 0x00, 0x00, 0x00, /* 0x0100-0x0107 */
516 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0108-0x010f */
517 0xd0, 0xf0, /* 0x0110-0x0111 */
518#define FROM_IDX_01 (FROM_IDX_00 + 78)
519 0x8c, 0x9c, 0x00, 0x00, 0x00, 0x00, /* 0x0152-0x0157 */
520 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0158-0x015f */
521 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0160-0x0167 */
522 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0168-0x016f */
523 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0170-0x0177 */
524 0x9f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0178-0x017f */
525 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0180-0x0187 */
526 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0188-0x018f */
527 0x00, 0x00, 0x83, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0190-0x0197 */
528 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0198-0x019f */
529 0xd5, 0xf5, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x01a0-0x01a7 */
530 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xdd, /* 0x01a8-0x01af */
531 0xfd, /* 0x01b0-0x01b0 */
532#define FROM_IDX_02 (FROM_IDX_01 + 95)
533 0x88, 0x00, /* 0x02c6-0x02c7 */
534 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x02c8-0x02cf */
535 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x02d0-0x02d7 */
536 0x00, 0x00, 0x00, 0x00, 0x98, /* 0x02d8-0x02dc */
537#define FROM_IDX_03 (FROM_IDX_02 + 23)
538 0xcc, 0xec, 0x00, 0xde, 0x00, 0x00, 0x00, 0x00, /* 0x0300-0x0307 */
539 0x00, 0xd2, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0308-0x030f */
540 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0310-0x0317 */
541 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0318-0x031f */
542 0x00, 0x00, 0x00, 0xf2, /* 0x0320-0x0323 */
543#define FROM_IDX_20 (FROM_IDX_03 + 36)
544 0x96, 0x97, 0x00, 0x00, 0x00, /* 0x2013-0x2017 */
545 0x91, 0x92, 0x82, 0x00, 0x93, 0x94, 0x84, 0x00, /* 0x2018-0x201f */
546 0x86, 0x87, 0x95, 0x00, 0x00, 0x00, 0x85, 0x00, /* 0x2020-0x2027 */
547 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x2028-0x202f */
548 0x89, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x2030-0x2037 */
549 0x00, 0x8b, 0x9b, /* 0x2038-0x203a */
550#define FROM_IDX_FF (FROM_IDX_20 + 40)
551 };
552/* Compile-time verification of table size. */
553typedef int verify2[(NELEMS (from_ucs4) == FROM_IDX_FF) - 1];
554
555/* Decomposition table for the relevant Unicode characters. */
556static const struct
557{
558 uint16_t composed;
559 uint32_t base:8;
560 uint32_t comb1:8;
561} decomp_table[] =
562 {
563 { 0x00c0, 0x41, 0xcc },
564 { 0x00c1, 0x41, 0xec },
565 { 0x00c3, 0x41, 0xde },
566 { 0x00c8, 0x45, 0xcc },
567 { 0x00c9, 0x45, 0xec },
568 { 0x00cc, 0x49, 0xcc },
569 { 0x00cd, 0x49, 0xec },
570 { 0x00d1, 0x4e, 0xde },
571 { 0x00d2, 0x4f, 0xcc },
572 { 0x00d3, 0x4f, 0xec },
573 { 0x00d5, 0x4f, 0xde },
574 { 0x00d9, 0x55, 0xcc },
575 { 0x00da, 0x55, 0xec },
576 { 0x00dd, 0x59, 0xec },
577 { 0x00e0, 0x61, 0xcc },
578 { 0x00e1, 0x61, 0xec },
579 { 0x00e3, 0x61, 0xde },
580 { 0x00e8, 0x65, 0xcc },
581 { 0x00e9, 0x65, 0xec },
582 { 0x00ec, 0x69, 0xcc },
583 { 0x00ed, 0x69, 0xec },
584 { 0x00f1, 0x6e, 0xde },
585 { 0x00f2, 0x6f, 0xcc },
586 { 0x00f3, 0x6f, 0xec },
587 { 0x00f5, 0x6f, 0xde },
588 { 0x00f9, 0x75, 0xcc },
589 { 0x00fa, 0x75, 0xec },
590 { 0x00fd, 0x79, 0xec },
591 { 0x0106, 0x43, 0xec },
592 { 0x0107, 0x63, 0xec },
593 { 0x0128, 0x49, 0xde },
594 { 0x0129, 0x69, 0xde },
595 { 0x0139, 0x4c, 0xec },
596 { 0x013a, 0x6c, 0xec },
597 { 0x0143, 0x4e, 0xec },
598 { 0x0144, 0x6e, 0xec },
599 { 0x0154, 0x52, 0xec },
600 { 0x0155, 0x72, 0xec },
601 { 0x015a, 0x53, 0xec },
602 { 0x015b, 0x73, 0xec },
603 { 0x0168, 0x55, 0xde },
604 { 0x0169, 0x75, 0xde },
605 { 0x0179, 0x5a, 0xec },
606 { 0x017a, 0x7a, 0xec },
607 { 0x01d7, 0xdc, 0xec },
608 { 0x01d8, 0xfc, 0xec },
609 { 0x01db, 0xdc, 0xcc },
610 { 0x01dc, 0xfc, 0xcc },
611 { 0x01f4, 0x47, 0xec },
612 { 0x01f5, 0x67, 0xec },
613 { 0x01f8, 0x4e, 0xcc },
614 { 0x01f9, 0x6e, 0xcc },
615 { 0x01fa, 0xc5, 0xec },
616 { 0x01fb, 0xe5, 0xec },
617 { 0x01fc, 0xc6, 0xec },
618 { 0x01fd, 0xe6, 0xec },
619 { 0x01fe, 0xd8, 0xec },
620 { 0x01ff, 0xf8, 0xec },
621 { 0x0385, 0xa8, 0xec },
622 { 0x1e04, 0x42, 0xf2 },
623 { 0x1e05, 0x62, 0xf2 },
624 { 0x1e08, 0xc7, 0xec },
625 { 0x1e09, 0xe7, 0xec },
626 { 0x1e0c, 0x44, 0xf2 },
627 { 0x1e0d, 0x64, 0xf2 },
628 { 0x1e24, 0x48, 0xf2 },
629 { 0x1e25, 0x68, 0xf2 },
630 { 0x1e2e, 0xcf, 0xec },
631 { 0x1e2f, 0xef, 0xec },
632 { 0x1e30, 0x4b, 0xec },
633 { 0x1e31, 0x6b, 0xec },
634 { 0x1e32, 0x4b, 0xf2 },
635 { 0x1e33, 0x6b, 0xf2 },
636 { 0x1e36, 0x4c, 0xf2 },
637 { 0x1e37, 0x6c, 0xf2 },
638 { 0x1e3e, 0x4d, 0xec },
639 { 0x1e3f, 0x6d, 0xec },
640 { 0x1e42, 0x4d, 0xf2 },
641 { 0x1e43, 0x6d, 0xf2 },
642 { 0x1e46, 0x4e, 0xf2 },
643 { 0x1e47, 0x6e, 0xf2 },
644 { 0x1e4c, 0xd3, 0xde }, /*{ 0x1e4c, 0x00d5, 0xec }, { 0x1e4c, 0x004f, 1, 0xde },*/
645 { 0x1e4d, 0xf3, 0xde }, /*{ 0x1e4d, 0x00f5, 0xec }, { 0x1e4d, 0x006f, 1, 0xde },*/
646 { 0x1e4e, 0xd6, 0xde },
647 { 0x1e4f, 0xf6, 0xde },
648 { 0x1e54, 0x50, 0xec },
649 { 0x1e55, 0x70, 0xec },
650 { 0x1e5a, 0x52, 0xf2 },
651 { 0x1e5b, 0x72, 0xf2 },
652 { 0x1e62, 0x53, 0xf2 },
653 { 0x1e63, 0x73, 0xf2 },
654 { 0x1e6c, 0x54, 0xf2 },
655 { 0x1e6d, 0x74, 0xf2 },
656 { 0x1e78, 0xda, 0xde }, /*{ 0x1e78, 0x0168, 0xec }, { 0x1e78, 0x0055, 1, 0xde },*/
657 { 0x1e79, 0xfa, 0xde }, /*{ 0x1e79, 0x0169, 0xec }, { 0x1e79, 0x0075, 1, 0xde },*/
658 { 0x1e7c, 0x56, 0xde },
659 { 0x1e7d, 0x76, 0xde },
660 { 0x1e7e, 0x56, 0xf2 },
661 { 0x1e7f, 0x76, 0xf2 },
662 { 0x1e80, 0x57, 0xcc },
663 { 0x1e81, 0x77, 0xcc },
664 { 0x1e82, 0x57, 0xec },
665 { 0x1e83, 0x77, 0xec },
666 { 0x1e88, 0x57, 0xf2 },
667 { 0x1e89, 0x77, 0xf2 },
668 { 0x1e92, 0x5a, 0xf2 },
669 { 0x1e93, 0x7a, 0xf2 },
670 { 0x1ea0, 0x41, 0xf2 },
671 { 0x1ea1, 0x61, 0xf2 },
672 { 0x1ea2, 0x41, 0xd2 },
673 { 0x1ea3, 0x61, 0xd2 },
674 { 0x1ea4, 0xc2, 0xec },
675 { 0x1ea5, 0xe2, 0xec },
676 { 0x1ea6, 0xc2, 0xcc },
677 { 0x1ea7, 0xe2, 0xcc },
678 { 0x1ea8, 0xc2, 0xd2 },
679 { 0x1ea9, 0xe2, 0xd2 },
680 { 0x1eaa, 0xc2, 0xde },
681 { 0x1eab, 0xe2, 0xde },
682 { 0x1eac, 0xc2, 0xf2 },
683 { 0x1ead, 0xe2, 0xf2 },
684 { 0x1eae, 0xc3, 0xec },
685 { 0x1eaf, 0xe3, 0xec },
686 { 0x1eb0, 0xc3, 0xcc },
687 { 0x1eb1, 0xe3, 0xcc },
688 { 0x1eb2, 0xc3, 0xd2 },
689 { 0x1eb3, 0xe3, 0xd2 },
690 { 0x1eb4, 0xc3, 0xde },
691 { 0x1eb5, 0xe3, 0xde },
692 { 0x1eb6, 0xc3, 0xf2 },
693 { 0x1eb7, 0xe3, 0xf2 },
694 { 0x1eb8, 0x45, 0xf2 },
695 { 0x1eb9, 0x65, 0xf2 },
696 { 0x1eba, 0x45, 0xd2 },
697 { 0x1ebb, 0x65, 0xd2 },
698 { 0x1ebc, 0x45, 0xde },
699 { 0x1ebd, 0x65, 0xde },
700 { 0x1ebe, 0xca, 0xec },
701 { 0x1ebf, 0xea, 0xec },
702 { 0x1ec0, 0xca, 0xcc },
703 { 0x1ec1, 0xea, 0xcc },
704 { 0x1ec2, 0xca, 0xd2 },
705 { 0x1ec3, 0xea, 0xd2 },
706 { 0x1ec4, 0xca, 0xde },
707 { 0x1ec5, 0xea, 0xde },
708 { 0x1ec6, 0xca, 0xf2 },
709 { 0x1ec7, 0xea, 0xf2 },
710 { 0x1ec8, 0x49, 0xd2 },
711 { 0x1ec9, 0x69, 0xd2 },
712 { 0x1eca, 0x49, 0xf2 },
713 { 0x1ecb, 0x69, 0xf2 },
714 { 0x1ecc, 0x4f, 0xf2 },
715 { 0x1ecd, 0x6f, 0xf2 },
716 { 0x1ece, 0x4f, 0xd2 },
717 { 0x1ecf, 0x6f, 0xd2 },
718 { 0x1ed0, 0xd4, 0xec },
719 { 0x1ed1, 0xf4, 0xec },
720 { 0x1ed2, 0xd4, 0xcc },
721 { 0x1ed3, 0xf4, 0xcc },
722 { 0x1ed4, 0xd4, 0xd2 },
723 { 0x1ed5, 0xf4, 0xd2 },
724 { 0x1ed6, 0xd4, 0xde },
725 { 0x1ed7, 0xf4, 0xde },
726 { 0x1ed8, 0xd4, 0xf2 },
727 { 0x1ed9, 0xf4, 0xf2 },
728 { 0x1eda, 0xd5, 0xec },
729 { 0x1edb, 0xf5, 0xec },
730 { 0x1edc, 0xd5, 0xcc },
731 { 0x1edd, 0xf5, 0xcc },
732 { 0x1ede, 0xd5, 0xd2 },
733 { 0x1edf, 0xf5, 0xd2 },
734 { 0x1ee0, 0xd5, 0xde },
735 { 0x1ee1, 0xf5, 0xde },
736 { 0x1ee2, 0xd5, 0xf2 },
737 { 0x1ee3, 0xf5, 0xf2 },
738 { 0x1ee4, 0x55, 0xf2 },
739 { 0x1ee5, 0x75, 0xf2 },
740 { 0x1ee6, 0x55, 0xd2 },
741 { 0x1ee7, 0x75, 0xd2 },
742 { 0x1ee8, 0xdd, 0xec },
743 { 0x1ee9, 0xfd, 0xec },
744 { 0x1eea, 0xdd, 0xcc },
745 { 0x1eeb, 0xfd, 0xcc },
746 { 0x1eec, 0xdd, 0xd2 },
747 { 0x1eed, 0xfd, 0xd2 },
748 { 0x1eee, 0xdd, 0xde },
749 { 0x1eef, 0xfd, 0xde },
750 { 0x1ef0, 0xdd, 0xf2 },
751 { 0x1ef1, 0xfd, 0xf2 },
752 { 0x1ef2, 0x59, 0xcc },
753 { 0x1ef3, 0x79, 0xcc },
754 { 0x1ef4, 0x59, 0xf2 },
755 { 0x1ef5, 0x79, 0xf2 },
756 { 0x1ef6, 0x59, 0xd2 },
757 { 0x1ef7, 0x79, 0xd2 },
758 { 0x1ef8, 0x59, 0xde },
759 { 0x1ef9, 0x79, 0xde },
760 { 0x1fed, 0xa8, 0xcc },
761 { 0x1fee, 0xa8, 0xec },
762 };
763
764#define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM
765#define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM
766#define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO
767#define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO
768#define LOOPFCT TO_LOOP
769#define BODY \
770 { \
771 uint32_t ch = get32 (inptr); \
772 \
773 if (ch < 0x0080 || (ch >= 0x00a0 && ch < 0x00c3)) \
774 { \
775 *outptr++ = ch; \
776 inptr += 4; \
777 } \
778 else \
779 { \
780 unsigned char res; \
781 \
782 if (ch >= 0x00c4 && ch < 0x0112) \
783 res = from_ucs4[ch - 0x00c4 + FROM_IDX_00]; \
784 else if (ch >= 0x0152 && ch < 0x01b1) \
785 res = from_ucs4[ch - 0x0152 + FROM_IDX_01]; \
786 else if (ch >= 0x02c6 && ch < 0x02dd) \
787 res = from_ucs4[ch - 0x02c6 + FROM_IDX_02]; \
788 else if (ch >= 0x0300 && ch < 0x0324) \
789 res = from_ucs4[ch - 0x0300 + FROM_IDX_03]; \
790 else if (ch >= 0x0340 && ch < 0x0342) /* Vietnamese tone marks */ \
791 res = from_ucs4[ch - 0x0340 + FROM_IDX_03]; \
792 else if (ch >= 0x2013 && ch < 0x203b) \
793 res = from_ucs4[ch - 0x2013 + FROM_IDX_20]; \
794 else if (ch == 0x20ab) \
795 res = 0xfe; \
796 else if (ch == 0x20ac) \
797 res = 0x80; \
798 else if (ch == 0x2122) \
799 res = 0x99; \
800 else \
801 { \
802 UNICODE_TAG_HANDLER (ch, 4); \
803 res = 0; \
804 } \
805 \
806 if (__glibc_likely (res != 0)) \
807 { \
808 *outptr++ = res; \
809 inptr += 4; \
810 } \
811 else \
812 { \
813 /* Try canonical decomposition. */ \
814 unsigned int i1, i2; \
815 \
816 i1 = 0; \
817 i2 = sizeof (decomp_table) / sizeof (decomp_table[0]) - 1; \
818 if (ch >= decomp_table[i1].composed \
819 && ch <= decomp_table[i2].composed) \
820 { \
821 unsigned int i; \
822 \
823 for (;;) \
824 { \
825 i = (i1 + i2) >> 1; \
826 if (ch == decomp_table[i].composed) \
827 break; \
828 if (ch < decomp_table[i].composed) \
829 { \
830 if (i1 == i) \
831 goto failed; \
832 i2 = i; \
833 } \
834 else \
835 { \
836 if (i1 != i) \
837 i1 = i; \
838 else \
839 { \
840 i = i2; \
841 if (ch == decomp_table[i].composed) \
842 break; \
843 goto failed; \
844 } \
845 } \
846 } \
847 \
848 /* See whether we have room for two bytes. */ \
849 if (__glibc_unlikely (outptr + 1 >= outend)) \
850 { \
851 result = __GCONV_FULL_OUTPUT; \
852 break; \
853 } \
854 \
855 /* Found a canonical decomposition. */ \
856 *outptr++ = decomp_table[i].base; \
857 *outptr++ = decomp_table[i].comb1; \
858 inptr += 4; \
859 continue; \
860 } \
861 \
862 failed: \
863 /* This is an illegal character. */ \
864 STANDARD_TO_LOOP_ERR_HANDLER (4); \
865 } \
866 } \
867 }
868#define LOOP_NEED_FLAGS
869#define EXTRA_LOOP_DECLS , int *statep
870#include <iconv/loop.c>
871
872
873/* Now define the toplevel functions. */
874#include <iconv/skeleton.c>
875