]> git.ipfire.org Git - thirdparty/glibc.git/blame - iconvdata/iso-2022-cn-ext.c
Update copyright notices with scripts/update-copyrights.
[thirdparty/glibc.git] / iconvdata / iso-2022-cn-ext.c
CommitLineData
c0d99958 1/* Conversion module for ISO-2022-CN-EXT.
568035b7 2 Copyright (C) 2000-2013 Free Software Foundation, Inc.
c0d99958
UD
3 This file is part of the GNU C Library.
4 Contributed by Ulrich Drepper <drepper@cygnus.com>, 2000.
5
6 The GNU C Library is free software; you can redistribute it and/or
41bdb6e2
AJ
7 modify it under the terms of the GNU Lesser General Public
8 License as published by the Free Software Foundation; either
9 version 2.1 of the License, or (at your option) any later version.
c0d99958
UD
10
11 The GNU C Library is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
41bdb6e2 14 Lesser General Public License for more details.
c0d99958 15
41bdb6e2 16 You should have received a copy of the GNU Lesser General Public
59ba27a6
PE
17 License along with the GNU C Library; if not, see
18 <http://www.gnu.org/licenses/>. */
c0d99958
UD
19
20#include <dlfcn.h>
21#include <gconv.h>
22#include <stdint.h>
466dc74c 23#include <stdlib.h>
c0d99958
UD
24#include <string.h>
25#include "gb2312.h"
26#include "iso-ir-165.h"
755104ed 27#include "cns11643.h"
c0d99958
UD
28#include "cns11643l1.h"
29#include "cns11643l2.h"
30
31#include <assert.h>
32
33/* This makes obvious what everybody knows: 0x1b is the Esc character. */
34#define ESC 0x1b
35
36/* We have single-byte shift-in and shift-out sequences, and the single
37 shift sequences SS2 and SS3 which replaces the SS2/SS3 designation for
38 the next two bytes. */
39#define SI 0x0f
40#define SO 0x0e
41#define SS2_0 ESC
42#define SS2_1 0x4e
43#define SS3_0 ESC
44#define SS3_1 0x4f
45
46/* Definitions used in the body of the `gconv' function. */
47#define CHARSET_NAME "ISO-2022-CN-EXT//"
48#define DEFINE_INIT 1
49#define DEFINE_FINI 1
50#define FROM_LOOP from_iso2022cn_ext_loop
51#define TO_LOOP to_iso2022cn_ext_loop
faaa6f62
UD
52#define FROM_LOOP_MIN_NEEDED_FROM 1
53#define FROM_LOOP_MAX_NEEDED_FROM 4
54#define FROM_LOOP_MIN_NEEDED_TO 4
55#define FROM_LOOP_MAX_NEEDED_TO 4
56#define TO_LOOP_MIN_NEEDED_FROM 4
57#define TO_LOOP_MAX_NEEDED_FROM 4
58#define TO_LOOP_MIN_NEEDED_TO 1
59#define TO_LOOP_MAX_NEEDED_TO 6
c0d99958
UD
60#define PREPARE_LOOP \
61 int save_set; \
62 int *setp = &data->__statep->__count;
63#define EXTRA_LOOP_ARGS , setp
64
65
66/* The charsets GB/T 12345-90, GB 7589-87, GB/T 13131-9X, GB 7590-87,
67 and GB/T 13132-9X are not registered to the best of my knowledge and
68 therefore have no escape sequence assigned. We cannot handle them
69 for this reason. Tell the implementation about this. */
70#define X12345 '\0'
71#define X7589 '\0'
72#define X13131 '\0'
73#define X7590 '\0'
74#define X13132 '\0'
75
76
77/* The COUNT element of the state keeps track of the currently selected
78 character set. The possible values are: */
79enum
80{
81 ASCII_set = 0,
82 GB2312_set,
83 GB12345_set,
84 CNS11643_1_set,
85 ISO_IR_165_set,
86 SO_mask = 7,
87
755104ed
UD
88 GB7589_set = 1 << 3,
89 GB13131_set = 2 << 3,
90 CNS11643_2_set = 3 << 3,
91 SS2_mask = 3 << 3,
c0d99958 92
755104ed
UD
93 GB7590_set = 1 << 5,
94 GB13132_set = 2 << 5,
95 CNS11643_3_set = 3 << 5,
96 CNS11643_4_set = 4 << 5,
97 CNS11643_5_set = 5 << 5,
98 CNS11643_6_set = 6 << 5,
99 CNS11643_7_set = 7 << 5,
100 SS3_mask = 7 << 5,
c0d99958
UD
101
102#define CURRENT_MASK (SO_mask | SS2_mask | SS3_mask)
103
755104ed
UD
104 GB2312_ann = 1 << 8,
105 GB12345_ann = 2 << 8,
106 CNS11643_1_ann = 3 << 8,
107 ISO_IR_165_ann = 4 << 8,
108 SO_ann = 7 << 8,
c0d99958 109
755104ed
UD
110 GB7589_ann = 1 << 11,
111 GB13131_ann = 2 << 11,
112 CNS11643_2_ann = 3 << 11,
113 SS2_ann = 3 << 11,
c0d99958 114
755104ed
UD
115 GB7590_ann = 1 << 13,
116 GB13132_ann = 2 << 13,
117 CNS11643_3_ann = 3 << 13,
118 CNS11643_4_ann = 4 << 13,
119 CNS11643_5_ann = 5 << 13,
120 CNS11643_6_ann = 6 << 13,
121 CNS11643_7_ann = 7 << 13,
122 SS3_ann = 7 << 13
c0d99958
UD
123};
124
125
126/* Since this is a stateful encoding we have to provide code which resets
127 the output state to the initial state. This has to be done during the
128 flushing. */
129#define EMIT_SHIFT_TO_INIT \
655de5fd 130 if (data->__statep->__count >> 3 != ASCII_set) \
c0d99958
UD
131 { \
132 if (FROM_DIRECTION) \
133 /* It's easy, we don't have to emit anything, we just reset the \
134 state for the input. */ \
655de5fd 135 data->__statep->__count = ASCII_set << 3; \
c0d99958
UD
136 else \
137 { \
c0d99958
UD
138 /* We are not in the initial state. To switch back we have \
139 to emit `SI'. */ \
2373b30e 140 if (__builtin_expect (outbuf == outend, 0)) \
c0d99958
UD
141 /* We don't have enough room in the output buffer. */ \
142 status = __GCONV_FULL_OUTPUT; \
143 else \
144 { \
145 /* Write out the shift sequence. */ \
146 *outbuf++ = SI; \
147 if (data->__flags & __GCONV_IS_LAST) \
148 *irreversible += 1; \
655de5fd 149 data->__statep->__count = ASCII_set << 3; \
c0d99958
UD
150 } \
151 } \
152 }
153
154
155/* Since we might have to reset input pointer we must be able to save
156 and retore the state. */
157#define SAVE_RESET_STATE(Save) \
158 if (Save) \
159 save_set = *setp; \
160 else \
161 *setp = save_set
162
163
164/* First define the conversion function from ISO-2022-CN to UCS4. */
faaa6f62
UD
165#define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM
166#define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM
167#define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO
168#define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO
c0d99958
UD
169#define LOOPFCT FROM_LOOP
170#define BODY \
171 { \
172 uint32_t ch = *inptr; \
173 \
174 /* This is a 7bit character set, disallow all 8bit characters. */ \
175 if (ch > 0x7f) \
e438a468 176 STANDARD_FROM_LOOP_ERR_HANDLER (1); \
c0d99958
UD
177 \
178 /* Recognize escape sequences. */ \
179 if (ch == ESC) \
180 { \
181 /* There are three kinds of escape sequences we have to handle: \
182 - those announcing the use of GB and CNS characters on the \
183 line; we can simply ignore them \
184 - the initial byte of the SS2 sequence. \
185 - the initial byte of the SS3 sequence. \
186 */ \
755104ed 187 if (inptr + 2 > inend \
c0d99958 188 || (inptr[1] == '$' \
755104ed
UD
189 && (inptr + 3 > inend \
190 || (inptr[2] == ')' && inptr + 4 > inend) \
191 || (inptr[2] == '*' && inptr + 4 > inend) \
192 || (inptr[2] == '+' && inptr + 4 > inend))) \
193 || (inptr[1] == SS2_1 && inptr + 4 > inend) \
194 || (inptr[1] == SS3_1 && inptr + 4 > inend)) \
c0d99958 195 { \
755104ed 196 result = __GCONV_INCOMPLETE_INPUT; \
c0d99958
UD
197 break; \
198 } \
199 if (inptr[1] == '$' \
200 && ((inptr[2] == ')' \
201 && (inptr[3] == 'A' \
202 || (X12345 != '\0' && inptr[3] == X12345) \
203 || inptr[3] == 'E' || inptr[3] == 'G')) \
204 || (inptr[2] == '*' \
205 && ((X7589 != '\0' && inptr[3] == X7589) \
206 || (X13131 != '\0' && inptr[3] == X13131) \
207 || inptr[3] == 'H')) \
208 || (inptr[2] == '+' \
209 && ((X7590 != '\0' && inptr[3] == X7590) \
210 || (X13132 != '\0' && inptr[3] == X13132) \
211 || inptr[3] == 'I' || inptr[3] == 'J' \
212 || inptr[3] == 'K' || inptr[3] == 'L' \
213 || inptr[3] == 'M')))) \
214 { \
215 /* OK, we accept those character sets. */ \
216 if (inptr[3] == 'A') \
217 ann = (ann & ~SO_ann) | GB2312_ann; \
218 else if (inptr[3] == 'G') \
219 ann = (ann & ~SO_ann) | CNS11643_1_ann; \
220 else if (inptr[3] == 'E') \
221 ann = (ann & ~SO_ann) | ISO_IR_165_ann; \
222 else if (X12345 != '\0' && inptr[3] == X12345) \
223 ann = (ann & ~SO_ann) | GB12345_ann; \
224 else if (inptr[3] == 'H') \
225 ann = (ann & ~SS2_ann) | CNS11643_2_ann; \
226 else if (X7589 != '\0' && inptr[3] == X7589) \
227 ann = (ann & ~SS2_ann) | GB7589_ann; \
228 else if (X13131 != '\0' && inptr[3] == X13131) \
229 ann = (ann & ~SS2_ann) | GB13131_ann; \
230 else if (inptr[3] == 'I') \
231 ann = (ann & ~SS3_ann) | CNS11643_3_ann; \
232 else if (inptr[3] == 'J') \
233 ann = (ann & ~SS3_ann) | CNS11643_4_ann; \
234 else if (inptr[3] == 'K') \
235 ann = (ann & ~SS3_ann) | CNS11643_5_ann; \
236 else if (inptr[3] == 'L') \
237 ann = (ann & ~SS3_ann) | CNS11643_6_ann; \
238 else if (inptr[3] == 'M') \
239 ann = (ann & ~SS3_ann) | CNS11643_7_ann; \
240 else if (X7590 != '\0' && inptr[3] == X7590) \
241 ann = (ann & ~SS3_ann) | GB7590_ann; \
242 else if (X13132 != '\0' && inptr[3] == X13132) \
243 ann = (ann & ~SS3_ann) | GB13132_ann; \
244 inptr += 4; \
245 continue; \
246 } \
247 } \
248 else if (ch == SO) \
249 { \
250 /* Switch to use GB2312, GB12345, CNS 11643 plane 1, or ISO-IR-165, \
251 depending on which S0 designation came last. The only problem \
252 is what to do with faulty input files where no designator came. \
253 XXX For now I'll default to use GB2312. If this is not the \
254 best behavior (e.g., we should flag an error) let me know. */ \
255 ++inptr; \
304d7abf
UD
256 if ((ann & SO_ann) != 0) \
257 switch (ann & SO_ann) \
258 { \
259 case GB2312_ann: \
260 set = GB2312_set; \
261 break; \
262 case GB12345_ann: \
263 set = GB12345_set; \
264 break; \
265 case CNS11643_1_ann: \
266 set = CNS11643_1_set; \
267 break; \
268 case ISO_IR_165_ann: \
269 set = ISO_IR_165_set; \
270 break; \
271 default: \
272 abort (); \
273 } \
274 else \
c0d99958 275 { \
e438a468 276 STANDARD_FROM_LOOP_ERR_HANDLER (1); \
c0d99958
UD
277 } \
278 continue; \
279 } \
280 else if (ch == SI) \
281 { \
282 /* Switch to use ASCII. */ \
283 ++inptr; \
284 set = ASCII_set; \
285 continue; \
286 } \
287 \
755104ed 288 if (ch == ESC && inptr[1] == SS2_1) \
c0d99958
UD
289 { \
290 /* This is a character from CNS 11643 plane 2. \
291 XXX We could test here whether the use of this character \
292 set was announced. \
289ac9dd 293 XXX Currently GB7589 and GB13131 are not supported. */ \
c0d99958
UD
294 inptr += 2; \
295 ch = cns11643l2_to_ucs4 (&inptr, 2, 0); \
296 if (ch == __UNKNOWN_10646_CHAR) \
e438a468 297 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
c0d99958 298 } \
755104ed 299 /* Note that we can assume here that at least 4 bytes are available if \
c0d99958
UD
300 the first byte is ESC since otherwise the first if would have been \
301 true. */ \
302 else if (ch == ESC && inptr[1] == SS3_1) \
303 { \
304 /* This is a character from CNS 11643 plane 3 or higher. \
755104ed 305 XXX Currently GB7590 and GB13132 are not supported. */ \
085a4412
UD
306 unsigned char buf[3]; \
307 const unsigned char *tmp = buf; \
755104ed
UD
308 \
309 buf[1] = inptr[2]; \
310 buf[2] = inptr[3]; \
311 switch (ann & SS3_ann) \
c0d99958 312 { \
755104ed 313 case CNS11643_3_ann: \
289ac9dd
UD
314 buf[0] = 0x23; \
315 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
316 break; \
317 case CNS11643_4_ann: \
318 buf[0] = 0x24; \
319 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
320 break; \
321 case CNS11643_5_ann: \
322 buf[0] = 0x25; \
323 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
324 break; \
325 case CNS11643_6_ann: \
326 buf[0] = 0x26; \
327 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
328 break; \
329 case CNS11643_7_ann: \
330 buf[0] = 0x27; \
331 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
755104ed
UD
332 break; \
333 default: \
289ac9dd 334 /* XXX Currently GB7590 and GB13132 are not supported. */ \
755104ed 335 ch = __UNKNOWN_10646_CHAR; \
c0d99958
UD
336 break; \
337 } \
c0d99958
UD
338 if (ch == __UNKNOWN_10646_CHAR) \
339 { \
e438a468 340 STANDARD_FROM_LOOP_ERR_HANDLER (4); \
c0d99958 341 } \
755104ed
UD
342 assert (tmp == buf + 3); \
343 inptr += 4; \
c0d99958
UD
344 } \
345 else if (set == ASCII_set) \
346 { \
347 /* Almost done, just advance the input pointer. */ \
348 ++inptr; \
349 } \
350 else \
351 { \
352 /* That's pretty easy, we have a dedicated functions for this. */ \
353 if (inend - inptr < 2) \
354 { \
355 result = __GCONV_INCOMPLETE_INPUT; \
356 break; \
357 } \
358 if (set == GB2312_set) \
359 ch = gb2312_to_ucs4 (&inptr, inend - inptr, 0); \
360 else if (set == ISO_IR_165_set) \
361 ch = isoir165_to_ucs4 (&inptr, inend - inptr); \
362 else \
363 { \
364 assert (set == CNS11643_1_set); \
365 ch = cns11643l1_to_ucs4 (&inptr, inend - inptr, 0); \
366 } \
367 \
368 if (ch == 0) \
369 { \
755104ed 370 result = __GCONV_INCOMPLETE_INPUT; \
c0d99958
UD
371 break; \
372 } \
373 else if (ch == __UNKNOWN_10646_CHAR) \
374 { \
e438a468 375 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
c0d99958
UD
376 } \
377 } \
378 \
b0df6f1a
AJ
379 *((uint32_t *) outptr) = ch; \
380 outptr += sizeof (uint32_t); \
c0d99958
UD
381 }
382#define EXTRA_LOOP_DECLS , int *setp
655de5fd
UD
383#define INIT_PARAMS int set = (*setp >> 3) & CURRENT_MASK; \
384 int ann = (*setp >> 3) & ~CURRENT_MASK
385#define UPDATE_PARAMS *setp = (set | ann) << 3
c0d99958
UD
386#define LOOP_NEED_FLAGS
387#include <iconv/loop.c>
388
389
390/* Next, define the other direction. */
faaa6f62
UD
391#define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM
392#define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM
393#define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO
394#define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO
c0d99958
UD
395#define LOOPFCT TO_LOOP
396#define BODY \
397 { \
398 uint32_t ch; \
399 size_t written = 0; \
400 \
8c0b7170 401 ch = *((const uint32_t *) inptr); \
c0d99958
UD
402 \
403 /* First see whether we can write the character using the currently \
404 selected character set. */ \
405 if (ch < 0x80) \
406 { \
407 if (set != ASCII_set) \
408 { \
409 *outptr++ = SI; \
410 set = ASCII_set; \
411 if (outptr == outend) \
412 { \
413 result = __GCONV_FULL_OUTPUT; \
414 break; \
415 } \
416 } \
417 \
418 *outptr++ = ch; \
419 written = 1; \
420 \
421 /* At the end of the line we have to clear the `ann' flags since \
422 every line must contain this information again. */ \
423 if (ch == L'\n') \
424 ann = 0; \
425 } \
426 else \
427 { \
085a4412 428 unsigned char buf[2]; \
c0d99958
UD
429 int used; \
430 \
755104ed
UD
431 if (set == GB2312_set || ((ann & SO_ann) != CNS11643_1_ann \
432 && (ann & SO_ann) != ISO_IR_165_ann)) \
c0d99958
UD
433 { \
434 written = ucs4_to_gb2312 (ch, buf, 2); \
435 used = GB2312_set; \
436 } \
755104ed 437 else if (set == ISO_IR_165_set || (ann & SO_ann) == ISO_IR_165_set) \
c0d99958 438 { \
755104ed
UD
439 written = ucs4_to_isoir165 (ch, buf, 2); \
440 used = ISO_IR_165_set; \
c0d99958
UD
441 } \
442 else \
443 { \
444 written = ucs4_to_cns11643l1 (ch, buf, 2); \
445 used = CNS11643_1_set; \
446 } \
447 \
448 if (written == __UNKNOWN_10646_CHAR) \
449 { \
450 /* Cannot convert it using the currently selected SO set. \
451 Next try the SS2 set. */ \
452 written = ucs4_to_cns11643l2 (ch, buf, 2); \
453 if (written != __UNKNOWN_10646_CHAR) \
454 /* Yep, that worked. */ \
455 used = CNS11643_2_set; \
456 else \
457 { \
085a4412 458 unsigned char tmpbuf[3]; \
c0d99958 459 \
755104ed 460 switch (0) \
c0d99958 461 { \
755104ed
UD
462 default: \
463 /* Well, see whether we have to change the SO set. */ \
464 \
465 if (used != GB2312_set) \
466 { \
467 written = ucs4_to_gb2312 (ch, buf, 2); \
468 if (written != __UNKNOWN_10646_CHAR) \
469 { \
470 used = GB2312_set; \
471 break; \
472 } \
473 } \
474 \
475 if (used != ISO_IR_165_set) \
476 { \
477 written = ucs4_to_isoir165 (ch, buf, 2); \
478 if (written != __UNKNOWN_10646_CHAR) \
479 { \
480 used = ISO_IR_165_set; \
481 break; \
482 } \
483 } \
484 \
485 if (used != CNS11643_1_set) \
486 { \
487 written = ucs4_to_cns11643l1 (ch, buf, 2); \
488 if (written != __UNKNOWN_10646_CHAR) \
489 { \
490 used = CNS11643_1_set; \
491 break; \
492 } \
493 } \
494 \
495 written = ucs4_to_cns11643 (ch, tmpbuf, 3); \
289ac9dd 496 if (written == 3 && tmpbuf[0] >= 3 && tmpbuf[0] <= 7) \
755104ed
UD
497 { \
498 buf[0] = tmpbuf[1]; \
499 buf[1] = tmpbuf[2]; \
289ac9dd 500 switch (tmpbuf[0]) \
755104ed 501 { \
289ac9dd 502 case 3: \
755104ed
UD
503 used = CNS11643_3_set; \
504 break; \
289ac9dd
UD
505 case 4: \
506 used = CNS11643_4_set; \
507 break; \
508 case 5: \
509 used = CNS11643_5_set; \
510 break; \
511 case 6: \
512 used = CNS11643_6_set; \
513 break; \
514 case 7: \
515 used = CNS11643_7_set; \
516 break; \
517 default: \
518 abort (); \
755104ed 519 } \
289ac9dd
UD
520 written = 2; \
521 break; \
755104ed
UD
522 } \
523 \
289ac9dd
UD
524 /* XXX Currently GB7590 and GB13132 are not supported. */\
525 \
c0d99958 526 /* Even this does not work. Error. */ \
755104ed
UD
527 used = ASCII_set; \
528 } \
529 if (used == ASCII_set) \
530 { \
601d2942 531 UNICODE_TAG_HANDLER (ch, 4); \
e438a468 532 STANDARD_TO_LOOP_ERR_HANDLER (4); \
c0d99958
UD
533 } \
534 } \
535 } \
536 assert (written == 2); \
537 \
538 /* See whether we have to emit an escape sequence. */ \
539 if (set != used) \
540 { \
541 /* First see whether we announced that we use this \
542 character set. */ \
755104ed 543 if ((used & SO_mask) != 0 && (ann & SO_ann) != (used << 8)) \
c0d99958
UD
544 { \
545 const char *escseq; \
546 \
547 if (outptr + 4 > outend) \
548 { \
549 result = __GCONV_FULL_OUTPUT; \
550 break; \
551 } \
552 \
553 assert (used >= 1 && used <= 4); \
755104ed
UD
554 escseq = ")A\0\0)G)E" + (used - 1) * 2; \
555 *outptr++ = ESC; \
556 *outptr++ = '$'; \
557 *outptr++ = *escseq++; \
558 *outptr++ = *escseq++; \
559 \
560 ann = (ann & ~SO_ann) | (used << 8); \
561 } \
562 else if ((used & SS2_mask) != 0 && (ann & SS2_ann) != (used << 8))\
563 { \
564 const char *escseq; \
565 \
566 assert (used == CNS11643_2_set); /* XXX */ \
567 escseq = "*H"; \
568 *outptr++ = ESC; \
569 *outptr++ = '$'; \
c0d99958
UD
570 *outptr++ = *escseq++; \
571 *outptr++ = *escseq++; \
755104ed
UD
572 \
573 ann = (ann & ~SS2_ann) | (used << 8); \
574 } \
575 else if ((used & SS3_mask) != 0 && (ann & SS3_ann) != (used << 8))\
576 { \
577 const char *escseq; \
578 \
579 assert ((used >> 5) >= 3 && (used >> 5) <= 7); \
580 escseq = "+I+J+K+L+M" + ((used >> 5) - 3) * 2; \
581 *outptr++ = ESC; \
582 *outptr++ = '$'; \
c0d99958
UD
583 *outptr++ = *escseq++; \
584 *outptr++ = *escseq++; \
585 \
755104ed 586 ann = (ann & ~SS3_ann) | (used << 8); \
c0d99958
UD
587 } \
588 \
589 if (used == CNS11643_2_set) \
590 { \
591 if (outptr + 2 > outend) \
592 { \
593 result = __GCONV_FULL_OUTPUT; \
594 break; \
595 } \
596 *outptr++ = SS2_0; \
597 *outptr++ = SS2_1; \
598 } \
755104ed
UD
599 else if (used >= CNS11643_3_set && used <= CNS11643_7_set) \
600 { \
601 if (outptr + 2 > outend) \
602 { \
603 result = __GCONV_FULL_OUTPUT; \
604 break; \
605 } \
606 *outptr++ = SS3_0; \
607 *outptr++ = SS3_1; \
608 } \
c0d99958
UD
609 else \
610 { \
611 /* We only have to emit something if currently ASCII is \
612 selected. Otherwise we are switching within the \
613 SO charset. */ \
614 if (set == ASCII_set) \
615 { \
616 if (outptr + 1 > outend) \
617 { \
618 result = __GCONV_FULL_OUTPUT; \
619 break; \
620 } \
621 *outptr++ = SO; \
622 } \
623 } \
624 \
625 /* Always test the length here since we have used up all the \
626 guaranteed output buffer slots. */ \
627 if (outptr + 2 > outend) \
628 { \
629 result = __GCONV_FULL_OUTPUT; \
630 break; \
631 } \
632 } \
633 else if (outptr + 2 > outend) \
634 { \
635 result = __GCONV_FULL_OUTPUT; \
636 break; \
637 } \
638 \
639 *outptr++ = buf[0]; \
640 *outptr++ = buf[1]; \
755104ed 641 set = used; \
c0d99958
UD
642 } \
643 \
644 /* Now that we wrote the output increment the input pointer. */ \
645 inptr += 4; \
646 }
647#define EXTRA_LOOP_DECLS , int *setp
655de5fd
UD
648#define INIT_PARAMS int set = (*setp >> 3) & CURRENT_MASK; \
649 int ann = (*setp >> 3) & ~CURRENT_MASK
4b1b449d
UD
650#define REINIT_PARAMS do \
651 { \
652 set = (*setp >> 3) & CURRENT_MASK; \
653 ann = (*setp >> 3) & ~CURRENT_MASK; \
654 } \
655 while (0)
655de5fd 656#define UPDATE_PARAMS *setp = (set | ann) << 3
c0d99958
UD
657#define LOOP_NEED_FLAGS
658#include <iconv/loop.c>
659
660
661/* Now define the toplevel functions. */
662#include <iconv/skeleton.c>