]>
Commit | Line | Data |
---|---|---|
faaa6f62 | 1 | /* Conversion module for ISO-2022-JP-3. |
d4697bc9 | 2 | Copyright (C) 1998-2014 Free Software Foundation, Inc. |
faaa6f62 UD |
3 | This file is part of the GNU C Library. |
4 | Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998, | |
5 | and Bruno Haible <bruno@clisp.org>, 2002. | |
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 | |
59ba27a6 PE |
18 | License along with the GNU C Library; if not, see |
19 | <http://www.gnu.org/licenses/>. */ | |
faaa6f62 UD |
20 | |
21 | #include <assert.h> | |
22 | #include <dlfcn.h> | |
23 | #include <gconv.h> | |
24 | #include <stdint.h> | |
25 | #include <string.h> | |
26 | ||
27 | #include "jis0201.h" | |
28 | #include "jis0208.h" | |
29 | #include "jisx0213.h" | |
30 | ||
31 | /* This makes obvious what everybody knows: 0x1b is the Esc character. */ | |
32 | #define ESC 0x1b | |
33 | ||
34 | /* Definitions used in the body of the `gconv' function. */ | |
35 | #define CHARSET_NAME "ISO-2022-JP-3//" | |
36 | #define FROM_LOOP from_iso2022jp3_loop | |
37 | #define TO_LOOP to_iso2022jp3_loop | |
38 | #define DEFINE_INIT 1 | |
39 | #define DEFINE_FINI 1 | |
40 | #define FROM_LOOP_MIN_NEEDED_FROM 1 | |
41 | #define FROM_LOOP_MAX_NEEDED_FROM 4 | |
42 | #define FROM_LOOP_MIN_NEEDED_TO 4 | |
43 | #define FROM_LOOP_MAX_NEEDED_TO 8 | |
44 | #define TO_LOOP_MIN_NEEDED_FROM 4 | |
45 | #define TO_LOOP_MAX_NEEDED_FROM 4 | |
46 | #define TO_LOOP_MIN_NEEDED_TO 1 | |
47 | #define TO_LOOP_MAX_NEEDED_TO 6 | |
48 | #define PREPARE_LOOP \ | |
49 | int saved_state; \ | |
50 | int *statep = &data->__statep->__count; | |
51 | #define EXTRA_LOOP_ARGS , statep | |
52 | ||
53 | ||
54 | /* The COUNT element of the state keeps track of the currently selected | |
55 | character set. The possible values are: */ | |
56 | enum | |
57 | { | |
58 | ASCII_set = 0, /* Esc ( B */ | |
59 | JISX0208_1978_set = 1 << 3, /* Esc $ @ */ | |
60 | JISX0208_1983_set = 2 << 3, /* Esc $ B */ | |
61 | JISX0201_Roman_set = 3 << 3, /* Esc ( J */ | |
62 | JISX0201_Kana_set = 4 << 3, /* Esc ( I */ | |
5d32be9a | 63 | JISX0213_1_2000_set = 5 << 3, /* Esc $ ( O */ |
faaa6f62 | 64 | JISX0213_2_set = 6 << 3, /* Esc $ ( P */ |
5d32be9a | 65 | JISX0213_1_2004_set = 7 << 3, /* Esc $ ( Q */ |
faaa6f62 UD |
66 | CURRENT_SEL_MASK = 7 << 3 |
67 | }; | |
68 | ||
69 | /* During UCS-4 to ISO-2022-JP-3 conversion, the COUNT element of the state | |
70 | also contains the last two bytes to be output, shifted by 6 bits, and a | |
71 | one-bit indicator whether they must be preceded by the shift sequence, | |
72 | in bit 22. */ | |
73 | ||
74 | /* Since this is a stateful encoding we have to provide code which resets | |
75 | the output state to the initial state. This has to be done during the | |
76 | flushing. */ | |
77 | #define EMIT_SHIFT_TO_INIT \ | |
78 | if ((data->__statep->__count & ~7) != ASCII_set) \ | |
79 | { \ | |
80 | if (FROM_DIRECTION) \ | |
81 | { \ | |
82 | /* It's easy, we don't have to emit anything, we just reset the \ | |
83 | state for the input. */ \ | |
84 | data->__statep->__count &= 7; \ | |
85 | data->__statep->__count |= ASCII_set; \ | |
86 | } \ | |
87 | else \ | |
88 | { \ | |
89 | /* We are not in the initial state. To switch back we have \ | |
90 | to write out the buffered character and/or emit the sequence \ | |
91 | `Esc ( B'. */ \ | |
92 | size_t need = \ | |
93 | (data->__statep->__count >> 6 \ | |
94 | ? (data->__statep->__count >> 22 ? 3 : 0) + 2 \ | |
95 | : 0) \ | |
96 | + ((data->__statep->__count & CURRENT_SEL_MASK) != ASCII_set \ | |
97 | ? 3 : 0); \ | |
98 | \ | |
99 | if (__builtin_expect (outbuf + need > outend, 0)) \ | |
100 | /* We don't have enough room in the output buffer. */ \ | |
101 | status = __GCONV_FULL_OUTPUT; \ | |
102 | else \ | |
103 | { \ | |
104 | if (data->__statep->__count >> 6) \ | |
105 | { \ | |
106 | uint32_t lasttwo = data->__statep->__count >> 6; \ | |
107 | \ | |
108 | if (lasttwo >> 16) \ | |
109 | { \ | |
110 | /* Write out the shift sequence before the last \ | |
111 | character. */ \ | |
b664d723 UD |
112 | assert ((data->__statep->__count & CURRENT_SEL_MASK) \ |
113 | == JISX0208_1983_set); \ | |
faaa6f62 UD |
114 | *outbuf++ = ESC; \ |
115 | *outbuf++ = '$'; \ | |
116 | *outbuf++ = 'B'; \ | |
117 | } \ | |
118 | /* Write out the last character. */ \ | |
119 | *outbuf++ = (lasttwo >> 8) & 0xff; \ | |
120 | *outbuf++ = lasttwo & 0xff; \ | |
121 | } \ | |
122 | if ((data->__statep->__count & CURRENT_SEL_MASK) != ASCII_set) \ | |
123 | { \ | |
124 | /* Write out the shift sequence. */ \ | |
125 | *outbuf++ = ESC; \ | |
126 | *outbuf++ = '('; \ | |
127 | *outbuf++ = 'B'; \ | |
128 | } \ | |
129 | data->__statep->__count &= 7; \ | |
130 | data->__statep->__count |= ASCII_set; \ | |
131 | } \ | |
132 | } \ | |
133 | } | |
134 | ||
135 | ||
136 | /* Since we might have to reset input pointer we must be able to save | |
137 | and retore the state. */ | |
138 | #define SAVE_RESET_STATE(Save) \ | |
139 | if (Save) \ | |
140 | saved_state = *statep; \ | |
141 | else \ | |
142 | *statep = saved_state | |
143 | ||
144 | ||
145 | /* First define the conversion function from ISO-2022-JP-3 to UCS-4. */ | |
146 | #define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM | |
147 | #define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM | |
148 | #define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO | |
149 | #define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO | |
150 | #define LOOPFCT FROM_LOOP | |
151 | #define BODY \ | |
152 | { \ | |
153 | uint32_t ch = *inptr; \ | |
154 | \ | |
155 | /* Recognize escape sequences. */ \ | |
156 | if (__builtin_expect (ch == ESC, 0)) \ | |
157 | { \ | |
158 | /* We now must be prepared to read two to three more bytes. \ | |
159 | If we have a match in the first byte but then the input buffer \ | |
160 | ends we terminate with an error since we must not risk missing \ | |
161 | an escape sequence just because it is not entirely in the \ | |
162 | current input buffer. */ \ | |
163 | if (__builtin_expect (inptr + 2 >= inend, 0) \ | |
164 | || (inptr[1] == '$' && inptr[2] == '(' \ | |
165 | && __builtin_expect (inptr + 3 >= inend, 0))) \ | |
166 | { \ | |
167 | /* Not enough input available. */ \ | |
168 | result = __GCONV_INCOMPLETE_INPUT; \ | |
169 | break; \ | |
170 | } \ | |
171 | \ | |
172 | if (inptr[1] == '(') \ | |
173 | { \ | |
174 | if (inptr[2] == 'B') \ | |
175 | { \ | |
176 | /* ASCII selected. */ \ | |
177 | set = ASCII_set; \ | |
178 | inptr += 3; \ | |
179 | continue; \ | |
180 | } \ | |
181 | else if (inptr[2] == 'J') \ | |
182 | { \ | |
183 | /* JIS X 0201 selected. */ \ | |
184 | set = JISX0201_Roman_set; \ | |
185 | inptr += 3; \ | |
186 | continue; \ | |
187 | } \ | |
188 | else if (inptr[2] == 'I') \ | |
189 | { \ | |
190 | /* JIS X 0201 selected. */ \ | |
191 | set = JISX0201_Kana_set; \ | |
192 | inptr += 3; \ | |
193 | continue; \ | |
194 | } \ | |
195 | } \ | |
196 | else if (inptr[1] == '$') \ | |
197 | { \ | |
198 | if (inptr[2] == '@') \ | |
199 | { \ | |
200 | /* JIS X 0208-1978 selected. */ \ | |
201 | set = JISX0208_1978_set; \ | |
202 | inptr += 3; \ | |
203 | continue; \ | |
204 | } \ | |
205 | else if (inptr[2] == 'B') \ | |
206 | { \ | |
207 | /* JIS X 0208-1983 selected. */ \ | |
208 | set = JISX0208_1983_set; \ | |
209 | inptr += 3; \ | |
210 | continue; \ | |
211 | } \ | |
212 | else if (inptr[2] == '(') \ | |
213 | { \ | |
5d32be9a | 214 | if (inptr[3] == 'O' || inptr[3] == 'Q') \ |
faaa6f62 UD |
215 | { \ |
216 | /* JIS X 0213 plane 1 selected. */ \ | |
5d32be9a UD |
217 | /* In this direction we don't need to distinguish the \ |
218 | versions from 2000 and 2004. */ \ | |
219 | set = JISX0213_1_2004_set; \ | |
faaa6f62 UD |
220 | inptr += 4; \ |
221 | continue; \ | |
222 | } \ | |
223 | else if (inptr[3] == 'P') \ | |
224 | { \ | |
225 | /* JIS X 0213 plane 2 selected. */ \ | |
226 | set = JISX0213_2_set; \ | |
227 | inptr += 4; \ | |
228 | continue; \ | |
229 | } \ | |
230 | } \ | |
231 | } \ | |
232 | } \ | |
233 | \ | |
234 | if (ch >= 0x80) \ | |
235 | { \ | |
e438a468 | 236 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
faaa6f62 UD |
237 | } \ |
238 | else if (set == ASCII_set || (ch < 0x21 || ch == 0x7f)) \ | |
239 | /* Almost done, just advance the input pointer. */ \ | |
240 | ++inptr; \ | |
241 | else if (set == JISX0201_Roman_set) \ | |
242 | { \ | |
243 | /* Use the JIS X 0201 table. */ \ | |
244 | ch = jisx0201_to_ucs4 (ch); \ | |
245 | if (__builtin_expect (ch == __UNKNOWN_10646_CHAR, 0)) \ | |
246 | { \ | |
e438a468 | 247 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
faaa6f62 UD |
248 | } \ |
249 | ++inptr; \ | |
250 | } \ | |
251 | else if (set == JISX0201_Kana_set) \ | |
252 | { \ | |
253 | /* Use the JIS X 0201 table. */ \ | |
254 | ch = jisx0201_to_ucs4 (ch + 0x80); \ | |
255 | if (__builtin_expect (ch == __UNKNOWN_10646_CHAR, 0)) \ | |
256 | { \ | |
e438a468 | 257 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
faaa6f62 UD |
258 | } \ |
259 | ++inptr; \ | |
260 | } \ | |
261 | else if (set == JISX0208_1978_set || set == JISX0208_1983_set) \ | |
262 | { \ | |
263 | /* XXX I don't have the tables for these two old variants of \ | |
264 | JIS X 0208. Therefore I'm using the tables for JIS X \ | |
265 | 0208-1990. If somebody has problems with this please \ | |
266 | provide the appropriate tables. */ \ | |
267 | ch = jisx0208_to_ucs4 (&inptr, inend - inptr, 0); \ | |
268 | \ | |
269 | if (__builtin_expect (ch == 0, 0)) \ | |
270 | { \ | |
271 | result = __GCONV_INCOMPLETE_INPUT; \ | |
272 | break; \ | |
273 | } \ | |
274 | else if (__builtin_expect (ch == __UNKNOWN_10646_CHAR, 0)) \ | |
275 | { \ | |
e438a468 | 276 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
faaa6f62 UD |
277 | } \ |
278 | } \ | |
5d32be9a | 279 | else /* (set == JISX0213_1_2004_set || set == JISX0213_2_set) */ \ |
faaa6f62 UD |
280 | { \ |
281 | if (__builtin_expect (inptr + 1 >= inend, 0)) \ | |
282 | { \ | |
283 | result = __GCONV_INCOMPLETE_INPUT; \ | |
284 | break; \ | |
285 | } \ | |
286 | \ | |
5d32be9a UD |
287 | ch = jisx0213_to_ucs4 ( \ |
288 | ((JISX0213_1_2004_set - set + (1 << 3)) << 5) + ch, \ | |
289 | inptr[1]); \ | |
faaa6f62 | 290 | if (ch == 0) \ |
e438a468 | 291 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
faaa6f62 UD |
292 | \ |
293 | if (ch < 0x80) \ | |
294 | { \ | |
295 | /* It's a combining character. */ \ | |
296 | uint32_t u1 = __jisx0213_to_ucs_combining[ch - 1][0]; \ | |
297 | uint32_t u2 = __jisx0213_to_ucs_combining[ch - 1][1]; \ | |
298 | \ | |
299 | /* See whether we have room for two characters. */ \ | |
300 | if (outptr + 8 <= outend) \ | |
301 | { \ | |
302 | inptr += 2; \ | |
303 | put32 (outptr, u1); \ | |
304 | outptr += 4; \ | |
305 | put32 (outptr, u2); \ | |
306 | outptr += 4; \ | |
307 | continue; \ | |
308 | } \ | |
309 | else \ | |
310 | { \ | |
311 | result = __GCONV_FULL_OUTPUT; \ | |
312 | break; \ | |
313 | } \ | |
314 | } \ | |
315 | \ | |
316 | inptr += 2; \ | |
317 | } \ | |
318 | \ | |
319 | put32 (outptr, ch); \ | |
320 | outptr += 4; \ | |
321 | } | |
322 | #define LOOP_NEED_FLAGS | |
323 | #define EXTRA_LOOP_DECLS , int *statep | |
324 | #define INIT_PARAMS int set = *statep | |
325 | #define UPDATE_PARAMS *statep = set | |
326 | #include <iconv/loop.c> | |
327 | ||
328 | ||
329 | /* Next, define the other direction, from UCS-4 to ISO-2022-JP-3. */ | |
330 | ||
331 | /* Composition tables for each of the relevant combining characters. */ | |
332 | static const struct | |
333 | { | |
334 | uint16_t base; | |
335 | uint16_t composed; | |
336 | } comp_table_data[] = | |
337 | { | |
338 | #define COMP_TABLE_IDX_02E5 0 | |
339 | #define COMP_TABLE_LEN_02E5 1 | |
340 | { 0x2b64, 0x2b65 }, /* 0x12B65 = 0x12B64 U+02E5 */ | |
341 | #define COMP_TABLE_IDX_02E9 (COMP_TABLE_IDX_02E5 + COMP_TABLE_LEN_02E5) | |
342 | #define COMP_TABLE_LEN_02E9 1 | |
343 | { 0x2b60, 0x2b66 }, /* 0x12B66 = 0x12B60 U+02E9 */ | |
344 | #define COMP_TABLE_IDX_0300 (COMP_TABLE_IDX_02E9 + COMP_TABLE_LEN_02E9) | |
345 | #define COMP_TABLE_LEN_0300 5 | |
346 | { 0x295c, 0x2b44 }, /* 0x12B44 = 0x1295C U+0300 */ | |
347 | { 0x2b38, 0x2b48 }, /* 0x12B48 = 0x12B38 U+0300 */ | |
348 | { 0x2b37, 0x2b4a }, /* 0x12B4A = 0x12B37 U+0300 */ | |
349 | { 0x2b30, 0x2b4c }, /* 0x12B4C = 0x12B30 U+0300 */ | |
350 | { 0x2b43, 0x2b4e }, /* 0x12B4E = 0x12B43 U+0300 */ | |
351 | #define COMP_TABLE_IDX_0301 (COMP_TABLE_IDX_0300 + COMP_TABLE_LEN_0300) | |
352 | #define COMP_TABLE_LEN_0301 4 | |
353 | { 0x2b38, 0x2b49 }, /* 0x12B49 = 0x12B38 U+0301 */ | |
354 | { 0x2b37, 0x2b4b }, /* 0x12B4B = 0x12B37 U+0301 */ | |
355 | { 0x2b30, 0x2b4d }, /* 0x12B4D = 0x12B30 U+0301 */ | |
356 | { 0x2b43, 0x2b4f }, /* 0x12B4F = 0x12B43 U+0301 */ | |
357 | #define COMP_TABLE_IDX_309A (COMP_TABLE_IDX_0301 + COMP_TABLE_LEN_0301) | |
358 | #define COMP_TABLE_LEN_309A 14 | |
359 | { 0x242b, 0x2477 }, /* 0x12477 = 0x1242B U+309A */ | |
360 | { 0x242d, 0x2478 }, /* 0x12478 = 0x1242D U+309A */ | |
361 | { 0x242f, 0x2479 }, /* 0x12479 = 0x1242F U+309A */ | |
362 | { 0x2431, 0x247a }, /* 0x1247A = 0x12431 U+309A */ | |
363 | { 0x2433, 0x247b }, /* 0x1247B = 0x12433 U+309A */ | |
364 | { 0x252b, 0x2577 }, /* 0x12577 = 0x1252B U+309A */ | |
365 | { 0x252d, 0x2578 }, /* 0x12578 = 0x1252D U+309A */ | |
366 | { 0x252f, 0x2579 }, /* 0x12579 = 0x1252F U+309A */ | |
367 | { 0x2531, 0x257a }, /* 0x1257A = 0x12531 U+309A */ | |
368 | { 0x2533, 0x257b }, /* 0x1257B = 0x12533 U+309A */ | |
369 | { 0x253b, 0x257c }, /* 0x1257C = 0x1253B U+309A */ | |
370 | { 0x2544, 0x257d }, /* 0x1257D = 0x12544 U+309A */ | |
371 | { 0x2548, 0x257e }, /* 0x1257E = 0x12548 U+309A */ | |
372 | { 0x2675, 0x2678 }, /* 0x12678 = 0x12675 U+309A */ | |
373 | }; | |
374 | ||
375 | #define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM | |
376 | #define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM | |
377 | #define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO | |
378 | #define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO | |
379 | #define LOOPFCT TO_LOOP | |
380 | #define BODY \ | |
381 | { \ | |
382 | uint32_t ch = get32 (inptr); \ | |
383 | \ | |
384 | if (lasttwo != 0) \ | |
385 | { \ | |
386 | /* Attempt to combine the last character with this one. */ \ | |
387 | unsigned int idx; \ | |
388 | unsigned int len; \ | |
389 | \ | |
390 | if (ch == 0x02e5) \ | |
391 | idx = COMP_TABLE_IDX_02E5, len = COMP_TABLE_LEN_02E5; \ | |
392 | else if (ch == 0x02e9) \ | |
393 | idx = COMP_TABLE_IDX_02E9, len = COMP_TABLE_LEN_02E9; \ | |
394 | else if (ch == 0x0300) \ | |
395 | idx = COMP_TABLE_IDX_0300, len = COMP_TABLE_LEN_0300; \ | |
396 | else if (ch == 0x0301) \ | |
397 | idx = COMP_TABLE_IDX_0301, len = COMP_TABLE_LEN_0301; \ | |
398 | else if (ch == 0x309a) \ | |
399 | idx = COMP_TABLE_IDX_309A, len = COMP_TABLE_LEN_309A; \ | |
400 | else \ | |
401 | goto not_combining; \ | |
402 | \ | |
403 | do \ | |
404 | if (comp_table_data[idx].base == (uint16_t) lasttwo) \ | |
405 | break; \ | |
406 | while (++idx, --len > 0); \ | |
407 | \ | |
408 | if (len > 0) \ | |
409 | { \ | |
410 | /* Output the combined character. */ \ | |
411 | /* We know the combined character is in JISX0213 plane 1, \ | |
412 | but the buffered character may have been in JISX0208 or in \ | |
413 | JISX0213 plane 1. */ \ | |
5d32be9a UD |
414 | size_t need = \ |
415 | (lasttwo >> 16 \ | |
416 | || (set != JISX0213_1_2000_set && set != JISX0213_1_2004_set) \ | |
417 | ? 4 : 0); \ | |
faaa6f62 UD |
418 | \ |
419 | if (__builtin_expect (outptr + need + 2 > outend, 0)) \ | |
420 | { \ | |
421 | result = __GCONV_FULL_OUTPUT; \ | |
422 | break; \ | |
423 | } \ | |
424 | if (need) \ | |
425 | { \ | |
426 | /* But first, output the escape sequence. */ \ | |
427 | *outptr++ = ESC; \ | |
428 | *outptr++ = '$'; \ | |
429 | *outptr++ = '('; \ | |
430 | *outptr++ = 'O'; \ | |
5d32be9a | 431 | set = JISX0213_1_2000_set; \ |
faaa6f62 UD |
432 | } \ |
433 | lasttwo = comp_table_data[idx].composed; \ | |
434 | *outptr++ = (lasttwo >> 8) & 0xff; \ | |
435 | *outptr++ = lasttwo & 0xff; \ | |
436 | lasttwo = 0; \ | |
437 | inptr += 4; \ | |
438 | continue; \ | |
439 | } \ | |
440 | \ | |
441 | not_combining: \ | |
442 | /* Output the buffered character. */ \ | |
443 | /* We know it is in JISX0208 or in JISX0213 plane 1. */ \ | |
444 | { \ | |
445 | size_t need = (lasttwo >> 16 ? 3 : 0); \ | |
446 | \ | |
447 | if (__builtin_expect (outptr + need + 2 > outend, 0)) \ | |
448 | { \ | |
449 | result = __GCONV_FULL_OUTPUT; \ | |
450 | break; \ | |
451 | } \ | |
452 | if (need) \ | |
453 | { \ | |
454 | /* But first, output the escape sequence. */ \ | |
455 | assert (set == JISX0208_1983_set); \ | |
456 | *outptr++ = ESC; \ | |
457 | *outptr++ = '$'; \ | |
458 | *outptr++ = 'B'; \ | |
459 | } \ | |
460 | *outptr++ = (lasttwo >> 8) & 0xff; \ | |
461 | *outptr++ = lasttwo & 0xff; \ | |
462 | lasttwo = 0; \ | |
463 | continue; \ | |
464 | } \ | |
465 | } \ | |
466 | \ | |
467 | /* First see whether we can write the character using the currently \ | |
468 | selected character set. */ \ | |
469 | if (set == ASCII_set) \ | |
470 | { \ | |
471 | /* Please note that the NUL byte is *not* matched if we are not \ | |
472 | currently using the ASCII charset. This is because we must \ | |
473 | switch to the initial state whenever a NUL byte is written. */ \ | |
474 | if (ch <= 0x7f) \ | |
475 | { \ | |
476 | *outptr++ = ch; \ | |
477 | inptr += 4; \ | |
478 | continue; \ | |
479 | } \ | |
480 | } \ | |
481 | /* ISO-2022-JP recommends to encode the newline character always in \ | |
482 | ASCII since this allows a context-free interpretation of the \ | |
483 | characters at the beginning of the next line. Otherwise it would \ | |
484 | have to be known whether the last line ended using ASCII or \ | |
485 | JIS X 0201. */ \ | |
486 | else if (set == JISX0201_Roman_set) \ | |
487 | { \ | |
488 | unsigned char buf[1]; \ | |
489 | if (ucs4_to_jisx0201 (ch, buf) != __UNKNOWN_10646_CHAR \ | |
490 | && buf[0] > 0x20 && buf[0] < 0x80) \ | |
491 | { \ | |
492 | *outptr++ = buf[0]; \ | |
493 | inptr += 4; \ | |
494 | continue; \ | |
495 | } \ | |
496 | } \ | |
497 | else if (set == JISX0201_Kana_set) \ | |
498 | { \ | |
499 | unsigned char buf[1]; \ | |
500 | if (ucs4_to_jisx0201 (ch, buf) != __UNKNOWN_10646_CHAR \ | |
501 | && buf[0] >= 0x80) \ | |
502 | { \ | |
503 | *outptr++ = buf[0] - 0x80; \ | |
504 | inptr += 4; \ | |
505 | continue; \ | |
506 | } \ | |
507 | } \ | |
508 | else if (/*set == JISX0208_1978_set || */ set == JISX0208_1983_set) \ | |
509 | { \ | |
510 | size_t written = ucs4_to_jisx0208 (ch, outptr, outend - outptr); \ | |
511 | \ | |
512 | if (written != __UNKNOWN_10646_CHAR) \ | |
513 | { \ | |
514 | uint32_t jch = ucs4_to_jisx0213 (ch); \ | |
515 | \ | |
516 | if (jch & 0x0080) \ | |
517 | { \ | |
518 | /* A possible match in comp_table_data. Buffer it. */ \ | |
519 | lasttwo = jch & 0x7f7f; \ | |
520 | inptr += 4; \ | |
521 | continue; \ | |
522 | } \ | |
523 | if (__builtin_expect (written == 0, 0)) \ | |
524 | { \ | |
525 | result = __GCONV_FULL_OUTPUT; \ | |
526 | break; \ | |
527 | } \ | |
528 | else \ | |
529 | { \ | |
530 | outptr += written; \ | |
531 | inptr += 4; \ | |
532 | continue; \ | |
533 | } \ | |
534 | } \ | |
535 | } \ | |
536 | else \ | |
537 | { \ | |
5d32be9a UD |
538 | /* (set == JISX0213_1_2000_set || set == JISX0213_1_2004_set \ |
539 | || set == JISX0213_2_set) */ \ | |
faaa6f62 UD |
540 | uint32_t jch = ucs4_to_jisx0213 (ch); \ |
541 | \ | |
542 | if (jch != 0 \ | |
5d32be9a UD |
543 | && (jch & 0x8000 \ |
544 | ? set == JISX0213_2_set \ | |
545 | : (set == JISX0213_1_2004_set \ | |
546 | || (set == JISX0213_1_2000_set \ | |
547 | && !jisx0213_added_in_2004_p (jch))))) \ | |
faaa6f62 UD |
548 | { \ |
549 | if (jch & 0x0080) \ | |
550 | { \ | |
551 | /* A possible match in comp_table_data. Buffer it. */ \ | |
552 | \ | |
553 | /* We know it's a JISX 0213 plane 1 character. */ \ | |
554 | assert ((jch & 0x8000) == 0); \ | |
555 | \ | |
556 | lasttwo = jch & 0x7f7f; \ | |
557 | inptr += 4; \ | |
558 | continue; \ | |
559 | } \ | |
560 | \ | |
561 | if (__builtin_expect (outptr + 1 >= outend, 0)) \ | |
562 | { \ | |
563 | result = __GCONV_FULL_OUTPUT; \ | |
564 | break; \ | |
565 | } \ | |
566 | *outptr++ = (jch >> 8) & 0x7f; \ | |
567 | *outptr++ = jch & 0x7f; \ | |
568 | inptr += 4; \ | |
569 | continue; \ | |
570 | } \ | |
571 | } \ | |
572 | \ | |
573 | /* The attempts to use the currently selected character set failed, \ | |
574 | either because the character requires a different character set, \ | |
575 | or because the character is unknown. */ \ | |
576 | \ | |
577 | if (ch <= 0x7f) \ | |
578 | { \ | |
579 | /* We must encode using ASCII. First write out the escape \ | |
580 | sequence. */ \ | |
581 | if (__builtin_expect (outptr + 3 > outend, 0)) \ | |
582 | { \ | |
583 | result = __GCONV_FULL_OUTPUT; \ | |
584 | break; \ | |
585 | } \ | |
586 | \ | |
587 | *outptr++ = ESC; \ | |
588 | *outptr++ = '('; \ | |
589 | *outptr++ = 'B'; \ | |
590 | set = ASCII_set; \ | |
591 | \ | |
592 | if (__builtin_expect (outptr >= outend, 0)) \ | |
593 | { \ | |
594 | result = __GCONV_FULL_OUTPUT; \ | |
595 | break; \ | |
596 | } \ | |
597 | *outptr++ = ch; \ | |
598 | } \ | |
599 | else \ | |
600 | { \ | |
601 | unsigned char buf[2]; \ | |
602 | \ | |
603 | /* Try JIS X 0201 Roman. */ \ | |
604 | if (ucs4_to_jisx0201 (ch, buf) != __UNKNOWN_10646_CHAR \ | |
605 | && buf[0] > 0x20 && buf[0] < 0x80) \ | |
606 | { \ | |
607 | if (set != JISX0201_Roman_set) \ | |
608 | { \ | |
609 | if (__builtin_expect (outptr + 3 > outend, 0)) \ | |
610 | { \ | |
611 | result = __GCONV_FULL_OUTPUT; \ | |
612 | break; \ | |
613 | } \ | |
614 | *outptr++ = ESC; \ | |
615 | *outptr++ = '('; \ | |
616 | *outptr++ = 'J'; \ | |
617 | set = JISX0201_Roman_set; \ | |
618 | } \ | |
619 | \ | |
620 | if (__builtin_expect (outptr >= outend, 0)) \ | |
621 | { \ | |
622 | result = __GCONV_FULL_OUTPUT; \ | |
623 | break; \ | |
624 | } \ | |
625 | *outptr++ = buf[0]; \ | |
626 | } \ | |
627 | else \ | |
628 | { \ | |
629 | uint32_t jch = ucs4_to_jisx0213 (ch); \ | |
630 | \ | |
631 | /* Try JIS X 0208. */ \ | |
632 | size_t written = ucs4_to_jisx0208 (ch, buf, 2); \ | |
633 | if (written != __UNKNOWN_10646_CHAR) \ | |
634 | { \ | |
635 | if (jch & 0x0080) \ | |
636 | { \ | |
637 | /* A possible match in comp_table_data. Buffer it. */ \ | |
638 | lasttwo = ((set != JISX0208_1983_set ? 1 : 0) << 16) \ | |
639 | | (jch & 0x7f7f); \ | |
640 | set = JISX0208_1983_set; \ | |
641 | inptr += 4; \ | |
642 | continue; \ | |
643 | } \ | |
644 | \ | |
645 | if (set != JISX0208_1983_set) \ | |
646 | { \ | |
647 | if (__builtin_expect (outptr + 3 > outend, 0)) \ | |
648 | { \ | |
649 | result = __GCONV_FULL_OUTPUT; \ | |
650 | break; \ | |
651 | } \ | |
652 | *outptr++ = ESC; \ | |
653 | *outptr++ = '$'; \ | |
654 | *outptr++ = 'B'; \ | |
655 | set = JISX0208_1983_set; \ | |
656 | } \ | |
657 | \ | |
658 | if (__builtin_expect (outptr + 2 > outend, 0)) \ | |
659 | { \ | |
660 | result = __GCONV_FULL_OUTPUT; \ | |
661 | break; \ | |
662 | } \ | |
663 | *outptr++ = buf[0]; \ | |
664 | *outptr++ = buf[1]; \ | |
665 | } \ | |
666 | else \ | |
667 | { \ | |
668 | /* Try JIS X 0213. */ \ | |
669 | if (jch != 0) \ | |
670 | { \ | |
671 | int new_set = \ | |
5d32be9a UD |
672 | (jch & 0x8000 \ |
673 | ? JISX0213_2_set \ | |
674 | : jisx0213_added_in_2004_p (jch) \ | |
675 | ? JISX0213_1_2004_set \ | |
676 | : JISX0213_1_2000_set); \ | |
faaa6f62 UD |
677 | \ |
678 | if (set != new_set) \ | |
679 | { \ | |
680 | if (__builtin_expect (outptr + 4 > outend, 0)) \ | |
681 | { \ | |
682 | result = __GCONV_FULL_OUTPUT; \ | |
683 | break; \ | |
684 | } \ | |
685 | *outptr++ = ESC; \ | |
686 | *outptr++ = '$'; \ | |
687 | *outptr++ = '('; \ | |
5d32be9a UD |
688 | *outptr++ = \ |
689 | ((new_set - JISX0213_1_2000_set) >> 3) + 'O'; \ | |
faaa6f62 UD |
690 | set = new_set; \ |
691 | } \ | |
692 | \ | |
693 | if (jch & 0x0080) \ | |
694 | { \ | |
695 | /* A possible match in comp_table_data. \ | |
696 | Buffer it. */ \ | |
697 | \ | |
698 | /* We know it's a JIS X 0213 plane 1 character. */ \ | |
699 | assert ((jch & 0x8000) == 0); \ | |
700 | \ | |
701 | lasttwo = jch & 0x7f7f; \ | |
702 | inptr += 4; \ | |
703 | continue; \ | |
704 | } \ | |
705 | \ | |
706 | if (__builtin_expect (outptr + 1 >= outend, 0)) \ | |
707 | { \ | |
708 | result = __GCONV_FULL_OUTPUT; \ | |
709 | break; \ | |
710 | } \ | |
711 | *outptr++ = (jch >> 8) & 0x7f; \ | |
712 | *outptr++ = jch & 0x7f; \ | |
713 | } \ | |
714 | else \ | |
715 | { \ | |
716 | /* Try JIS X 0201 Katakana. This is officially not part \ | |
717 | of ISO-2022-JP-3. Therefore we try it after all other \ | |
718 | attempts. */ \ | |
719 | if (ucs4_to_jisx0201 (ch, buf) != __UNKNOWN_10646_CHAR \ | |
720 | && buf[0] >= 0x80) \ | |
721 | { \ | |
722 | if (set != JISX0201_Kana_set) \ | |
723 | { \ | |
724 | if (__builtin_expect (outptr + 3 > outend, 0)) \ | |
725 | { \ | |
726 | result = __GCONV_FULL_OUTPUT; \ | |
727 | break; \ | |
728 | } \ | |
729 | *outptr++ = ESC; \ | |
730 | *outptr++ = '('; \ | |
731 | *outptr++ = 'I'; \ | |
732 | set = JISX0201_Kana_set; \ | |
733 | } \ | |
734 | \ | |
735 | if (__builtin_expect (outptr >= outend, 0)) \ | |
736 | { \ | |
737 | result = __GCONV_FULL_OUTPUT; \ | |
738 | break; \ | |
739 | } \ | |
740 | *outptr++ = buf[0] - 0x80; \ | |
741 | } \ | |
742 | else \ | |
743 | { \ | |
744 | UNICODE_TAG_HANDLER (ch, 4); \ | |
745 | \ | |
746 | /* Illegal character. */ \ | |
e438a468 | 747 | STANDARD_TO_LOOP_ERR_HANDLER (4); \ |
faaa6f62 UD |
748 | } \ |
749 | } \ | |
750 | } \ | |
751 | } \ | |
752 | } \ | |
753 | \ | |
754 | /* Now that we wrote the output increment the input pointer. */ \ | |
755 | inptr += 4; \ | |
756 | } | |
757 | #define LOOP_NEED_FLAGS | |
758 | #define EXTRA_LOOP_DECLS , int *statep | |
759 | #define INIT_PARAMS int set = *statep & CURRENT_SEL_MASK; \ | |
760 | uint32_t lasttwo = *statep >> 6 | |
4b1b449d UD |
761 | #define REINIT_PARAMS do \ |
762 | { \ | |
763 | set = *statep & CURRENT_SEL_MASK; \ | |
764 | lasttwo = *statep >> 6; \ | |
765 | } \ | |
766 | while (0) | |
faaa6f62 UD |
767 | #define UPDATE_PARAMS *statep = set | (lasttwo << 6) |
768 | #include <iconv/loop.c> | |
769 | ||
770 | ||
771 | /* Now define the toplevel functions. */ | |
772 | #include <iconv/skeleton.c> |