?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Update.
[glibc.git] / iconvdata / iso-2022-jp.c
blob8fc67b2b4356a3364f6df6917633ead4c8cdc43c
1 /* Conversion module for ISO-2022-JP.
2 Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.
5
6 The GNU C Library is free software; you can redistribute it and/or
7 modify it under the terms of the GNU Library General Public License as
8 published by the Free Software Foundation; either version 2 of the
9 License, or (at your option) any later version.
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
14 Library General Public License for more details.
15
16 You should have received a copy of the GNU Library General Public
17 License along with the GNU C Library; see the file COPYING.LIB. If not,
18 write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19 Boston, MA 02111-1307, USA. */
20
21 #include <gconv.h>
22 #include <stdint.h>
23 #include <stdlib.h>
24 #include <string.h>
25 #include "jis0201.h"
26 #include "jis0208.h"
27 #include "jis0212.h"
28 #include "gb2312.h"
29 #include "ksc5601.h"
30
31 struct gap
32 {
33 uint16_t start;
34 uint16_t end;
35 int32_t idx;
36 };
37
38 #include "iso8859-7jp.h"
39
40 /* This makes obvious what everybody knows: 0x1b is the Esc character. */
41 #define ESC 0x1b
42
43 /* We provide our own initialization and destructor function. */
44 #define DEFINE_INIT 0
45 #define DEFINE_FINI 0
46
47 /* Definitions used in the body of the `gconv' function. */
48 #define FROM_LOOP from_iso2022jp_loop
49 #define TO_LOOP to_iso2022jp_loop
50 #define MIN_NEEDED_FROM 1
51 #define MAX_NEEDED_FROM 4
52 #define MIN_NEEDED_TO 4
53 #define MAX_NEEDED_TO 4
54 #define FROM_DIRECTION (dir == from_iso2022jp)
55 #define PREPARE_LOOP \
56 enum direction dir = ((struct iso2022jp_data *) step->__data)->dir; \
57 enum variant var = ((struct iso2022jp_data *) step->__data)->var; \
58 int save_set; \
59 int *setp = &data->__statep->__count;
60 #define EXTRA_LOOP_ARGS , var, setp
61
62
63 /* Direction of the transformation. */
64 enum direction
65 {
66 illegal_dir,
67 to_iso2022jp,
68 from_iso2022jp
69 };
70
71 /* We handle ISO-2022-jp and ISO-2022-JP-2 here. */
72 enum variant
73 {
74 illegal_var,
75 iso2022jp,
76 iso2022jp2
77 };
78
79
80 struct iso2022jp_data
81 {
82 enum direction dir;
83 enum variant var;
84 };
85
86
87 /* The COUNT element of the state keeps track of the currently selected
88 character set. The possible values are: */
89 enum
90 {
91 ASCII_set = 0,
92 JISX0208_1978_set = 8,
93 JISX0208_1983_set = 16,
94 JISX0201_Roman_set = 24,
95 JISX0201_Kana_set = 32,
96 GB2312_set = 40,
97 KSC5601_set = 48,
98 JISX0212_set = 56,
99 CURRENT_SEL_MASK = 56
100 };
101
102 /* The second value stored is the designation of the G2 set. The following
103 values are possible: */
104 enum
105 {
106 UNSPECIFIED_set = 0,
107 ISO88591_set = 64,
108 ISO88597_set = 128,
109 CURRENT_ASSIGN_MASK = 192
110 };
111
112
113 int
114 gconv_init (struct __gconv_step *step)
115 {
116 /* Determine which direction. */
117 struct iso2022jp_data *new_data;
118 enum direction dir = illegal_dir;
119 enum variant var = illegal_var;
120 int result;
121
122 if (__strcasecmp (step->__from_name, "ISO-2022-JP//") == 0)
123 {
124 dir = from_iso2022jp;
125 var = iso2022jp;
126 }
127 else if (__strcasecmp (step->__to_name, "ISO-2022-JP//") == 0)
128 {
129 dir = to_iso2022jp;
130 var = iso2022jp;
131 }
132 else if (__strcasecmp (step->__from_name, "ISO-2022-JP-2//") == 0)
133 {
134 dir = from_iso2022jp;
135 var = iso2022jp2;
136 }
137 else if (__strcasecmp (step->__to_name, "ISO-2022-JP-2//") == 0)
138 {
139 dir = to_iso2022jp;
140 var = iso2022jp2;
141 }
142
143 result = __GCONV_NOCONV;
144 if (__builtin_expect (dir, from_iso2022jp) != illegal_dir)
145 {
146 new_data
147 = (struct iso2022jp_data *) malloc (sizeof (struct iso2022jp_data));
148
149 result = __GCONV_NOMEM;
150 if (new_data != NULL)
151 {
152 new_data->dir = dir;
153 new_data->var = var;
154 step->__data = new_data;
155
156 if (dir == from_iso2022jp)
157 {
158 step->__min_needed_from = MIN_NEEDED_FROM;
159 step->__max_needed_from = MAX_NEEDED_FROM;
160 step->__min_needed_to = MIN_NEEDED_TO;
161 step->__max_needed_to = MAX_NEEDED_TO;
162 }
163 else
164 {
165 step->__min_needed_from = MIN_NEEDED_TO;
166 step->__max_needed_from = MAX_NEEDED_TO;
167 step->__min_needed_to = MIN_NEEDED_FROM;
168 step->__max_needed_to = MAX_NEEDED_FROM + 2;
169 }
170
171 /* Yes, this is a stateful encoding. */
172 step->__stateful = 1;
173
174 result = __GCONV_OK;
175 }
176 }
177
178 return result;
179 }
180
181
182 void
183 gconv_end (struct __gconv_step *data)
184 {
185 free (data->__data);
186 }
187
188
189 /* Since this is a stateful encoding we have to provide code which resets
190 the output state to the initial state. This has to be done during the
191 flushing. */
192 #define EMIT_SHIFT_TO_INIT \
193 if ((data->__statep->__count & ~7) != ASCII_set) \
194 { \
195 enum direction dir = ((struct iso2022jp_data *) step->__data)->dir; \
196 \
197 if (dir == from_iso2022jp) \
198 { \
199 /* It's easy, we don't have to emit anything, we just reset the \
200 state for the input. Note that this also clears the G2 \
201 designation. */ \
202 data->__statep->__count &= 7; \
203 data->__statep->__count |= ASCII_set; \
204 } \
205 else \
206 { \
207 unsigned char *outbuf = data->__outbuf; \
208 \
209 /* We are not in the initial state. To switch back we have \
210 to emit the sequence `Esc ( B'. */ \
211 if (__builtin_expect (outbuf + 3 > data->__outbufend, 0)) \
212 /* We don't have enough room in the output buffer. */ \
213 status = __GCONV_FULL_OUTPUT; \
214 else \
215 { \
216 /* Write out the shift sequence. */ \
217 *outbuf++ = ESC; \
218 *outbuf++ = '('; \
219 *outbuf++ = 'B'; \
220 data->__outbuf = outbuf; \
221 /* Note that this also clears the G2 designation. */ \
222 data->__statep->__count &= ~7; \
223 data->__statep->__count |= ASCII_set; \
224 } \
225 } \
226 }
227
228
229 /* Since we might have to reset input pointer we must be able to save
230 and retore the state. */
231 #define SAVE_RESET_STATE(Save) \
232 if (Save) \
233 save_set = *setp; \
234 else \
235 *setp = save_set
236
237
238 /* First define the conversion function from ISO-2022-JP to UCS4. */
239 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
240 #define MAX_NEEDED_INPUT MAX_NEEDED_FROM
241 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
242 #define LOOPFCT FROM_LOOP
243 #define BODY \
244 { \
245 uint32_t ch = *inptr; \
246 \
247 /* Recognize escape sequences. */ \
248 if (__builtin_expect (ch, 0) == ESC) \
249 { \
250 /* We now must be prepared to read two to three more \
251 chracters. If we have a match in the first character but \
252 then the input buffer ends we terminate with an error since \
253 we must not risk missing an escape sequence just because it \
254 is not entirely in the current input buffer. */ \
255 if (__builtin_expect (inptr + 2 >= inend, 0) \
256 || (var == iso2022jp2 && inptr[1] == '$' && inptr[2] == '(' \
257 && __builtin_expect (inptr + 3 >= inend, 0))) \
258 { \
259 /* Not enough input available. */ \
260 result = __GCONV_EMPTY_INPUT; \
261 break; \
262 } \
263 \
264 if (inptr[1] == '(') \
265 { \
266 if (inptr[2] == 'B') \
267 { \
268 /* ASCII selected. */ \
269 set = ASCII_set; \
270 inptr += 3; \
271 continue; \
272 } \
273 else if (inptr[2] == 'J') \
274 { \
275 /* JIS X 0201 selected. */ \
276 set = JISX0201_Roman_set; \
277 inptr += 3; \
278 continue; \
279 } \
280 else if (var == iso2022jp2 && inptr[2] == 'I') \
281 { \
282 /* JIS X 0201 selected. */ \
283 set = JISX0201_Kana_set; \
284 inptr += 3; \
285 continue; \
286 } \
287 } \
288 else if (inptr[1] == '$') \
289 { \
290 if (inptr[2] == '@') \
291 { \
292 /* JIS X 0208-1978 selected. */ \
293 set = JISX0208_1978_set; \
294 inptr += 3; \
295 continue; \
296 } \
297 else if (inptr[2] == 'B') \
298 { \
299 /* JIS X 0208-1983 selected. */ \
300 set = JISX0208_1983_set; \
301 inptr += 3; \
302 continue; \
303 } \
304 else if (var == iso2022jp2) \
305 { \
306 if (inptr[2] == 'A') \
307 { \
308 /* GB 2312-1980 selected. */ \
309 set = GB2312_set; \
310 inptr += 3; \
311 continue; \
312 } \
313 else if (inptr[2] == '(') \
314 { \
315 if (inptr[3] == 'C') \
316 { \
317 /* KSC 5601-1987 selected. */ \
318 set = KSC5601_set; \
319 inptr += 4; \
320 continue; \
321 } \
322 else if (inptr[3] == 'D') \
323 { \
324 /* JIS X 0212-1990 selected. */ \
325 set = JISX0212_set; \
326 inptr += 4; \
327 continue; \
328 } \
329 } \
330 } \
331 } \
332 else if (var == iso2022jp2 && inptr[1] == '.') \
333 { \
334 if (inptr[2] == 'A') \
335 { \
336 /* ISO 8859-1-GR selected. */ \
337 set2 = ISO88591_set; \
338 inptr += 3; \
339 continue; \
340 } \
341 else if (inptr[2] == 'F') \
342 { \
343 /* ISO 8859-7-GR selected. */ \
344 set2 = ISO88597_set; \
345 inptr += 3; \
346 continue; \
347 } \
348 } \
349 } \
350 \
351 if (ch == ESC && var == iso2022jp2 && inptr[1] == 'N') \
352 { \
353 if (set2 == ISO88591_set) \
354 { \
355 ch = inptr[2] | 0x80; \
356 inptr += 3; \
357 } \
358 else if (__builtin_expect (set2, ISO88597_set) == ISO88597_set) \
359 { \
360 /* We use the table from the ISO 8859-7 module. */ \
361 if (inptr[2] < 0x20 || inptr[2] > 0x80) \
362 { \
363 if (! ignore_errors_p ()) \
364 { \
365 result = __GCONV_ILLEGAL_INPUT; \
366 break; \
367 } \
368 \
369 ++inptr; \
370 ++*converted; \
371 continue; \
372 } \
373 ch = iso88597_to_ucs4[inptr[2] - 0x20]; \
374 if (ch == 0) \
375 { \
376 if (! ignore_errors_p ()) \
377 { \
378 result = __GCONV_ILLEGAL_INPUT; \
379 break; \
380 } \
381 \
382 inptr += 3; \
383 ++*converted; \
384 continue; \
385 } \
386 inptr += 3; \
387 } \
388 else \
389 { \
390 if (! ignore_errors_p ()) \
391 { \
392 result = __GCONV_ILLEGAL_INPUT; \
393 break; \
394 } \
395 \
396 ++inptr; \
397 ++*converted; \
398 continue; \
399 } \
400 } \
401 else if (set == ASCII_set || (ch < 0x21 || ch == 0x7f)) \
402 /* Almost done, just advance the input pointer. */ \
403 ++inptr; \
404 else if (set == JISX0201_Roman_set) \
405 { \
406 /* Use the JIS X 0201 table. */ \
407 ch = jisx0201_to_ucs4 (ch); \
408 if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR) \
409 { \
410 if (! ignore_errors_p ()) \
411 { \
412 result = __GCONV_ILLEGAL_INPUT; \
413 break; \
414 } \
415 \
416 ++inptr; \
417 ++*converted; \
418 continue; \
419 } \
420 ++inptr; \
421 } \
422 else if (set == JISX0201_Kana_set) \
423 { \
424 /* Use the JIS X 0201 table. */ \
425 ch = jisx0201_to_ucs4 (ch + 0x80); \
426 if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR) \
427 { \
428 if (! ignore_errors_p ()) \
429 { \
430 result = __GCONV_ILLEGAL_INPUT; \
431 break; \
432 } \
433 \
434 ++inptr; \
435 ++*converted; \
436 continue; \
437 } \
438 ++inptr; \
439 } \
440 else \
441 { \
442 if (set == JISX0208_1978_set || set == JISX0208_1983_set) \
443 /* XXX I don't have the tables for these two old variants of \
444 JIS X 0208. Therefore I'm using the tables for JIS X \
445 0208-1990. If somebody has problems with this please \
446 provide the appropriate tables. */ \
447 ch = jisx0208_to_ucs4 (&inptr, \
448 NEED_LENGTH_TEST ? inend - inptr : 2, 0); \
449 else if (set == JISX0212_set) \
450 /* Use the JIS X 0212 table. */ \
451 ch = jisx0212_to_ucs4 (&inptr, \
452 NEED_LENGTH_TEST ? inend - inptr : 2, 0); \
453 else if (set == GB2312_set) \
454 /* Use the GB 2312 table. */ \
455 ch = gb2312_to_ucs4 (&inptr, \
456 NEED_LENGTH_TEST ? inend - inptr : 2, 0); \
457 else \
458 { \
459 assert (set == KSC5601_set); \
460 \
461 /* Use the KSC 5601 table. */ \
462 ch = ksc5601_to_ucs4 (&inptr, \
463 NEED_LENGTH_TEST ? inend - inptr : 2, 0); \
464 } \
465 \
466 if (NEED_LENGTH_TEST && __builtin_expect (ch, 1) == 0) \
467 { \
468 result = __GCONV_EMPTY_INPUT; \
469 break; \
470 } \
471 else if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR) \
472 { \
473 if (! ignore_errors_p ()) \
474 { \
475 result = __GCONV_ILLEGAL_INPUT; \
476 break; \
477 } \
478 \
479 ++inptr; \
480 ++*converted; \
481 continue; \
482 } \
483 } \
484 \
485 put32 (outptr, ch); \
486 outptr += 4; \
487 }
488 #define EXTRA_LOOP_DECLS , enum variant var, int *setp
489 #define INIT_PARAMS int set = *setp & CURRENT_SEL_MASK; \
490 int set2 = *setp & CURRENT_ASSIGN_MASK
491 #define UPDATE_PARAMS *setp = set | set2
492 #include <iconv/loop.c>
493
494
495 /* Next, define the other direction. */
496 #define MIN_NEEDED_INPUT MIN_NEEDED_TO
497 #define MIN_NEEDED_OUTPUT MIN_NEEDED_FROM
498 #define MAX_NEEDED_OUTPUT (MAX_NEEDED_FROM + 2)
499 #define LOOPFCT TO_LOOP
500 #define BODY \
501 { \
502 uint32_t ch; \
503 size_t written = 0; \
504 \
505 ch = get32 (inptr); \
506 \
507 /* First see whether we can write the character using the currently \
508 selected character set. */ \
509 if (set == ASCII_set) \
510 { \
511 /* Please note that the NUL byte is *not* matched if we are not \
512 currently using the ASCII charset. This is because we must \
513 switch to the initial state whenever a NUL byte is written. */ \
514 if (ch <= 0x7f) \
515 { \
516 *outptr++ = ch; \
517 written = 1; \
518 } \
519 /* At the beginning of a line, G2 designation is cleared. */ \
520 if (var == iso2022jp2 && ch == 0x0a) \
521 set2 = UNSPECIFIED_set; \
522 } \
523 /* ISO-2022-JP recommends to encode the newline character always in \
524 ASCII since this allows a context-free interpretation of the \
525 characters at the beginning of the next line. Otherwise it would \
526 have to be known whether the last line ended using ASCII or \
527 JIS X 0201. */ \
528 else if (set == JISX0201_Roman_set) \
529 { \
530 unsigned char buf[2]; \
531 written = ucs4_to_jisx0201 (ch, buf); \
532 if (written != __UNKNOWN_10646_CHAR && buf[0] > 0x20 \
533 && buf[0] < 0x80) \
534 { \
535 *outptr++ = buf[0]; \
536 written = 1; \
537 } \
538 else \
539 written = __UNKNOWN_10646_CHAR; \
540 } \
541 else if (set == JISX0201_Kana_set) \
542 { \
543 unsigned char buf[2]; \
544 written = ucs4_to_jisx0201 (ch, buf); \
545 if (written != __UNKNOWN_10646_CHAR && buf[0] > 0xa0 \
546 && buf[0] < 0xe0) \
547 { \
548 *outptr++ = buf[0] - 0x80; \
549 written = 1; \
550 } \
551 else \
552 written = __UNKNOWN_10646_CHAR; \
553 } \
554 else \
555 { \
556 if (set == JISX0208_1978_set || set == JISX0208_1983_set) \
557 written = ucs4_to_jisx0208 (ch, outptr, \
558 (NEED_LENGTH_TEST \
559 ? outend - outptr : 2)); \
560 else if (set == JISX0212_set) \
561 written = ucs4_to_jisx0212 (ch, outptr, \
562 (NEED_LENGTH_TEST \
563 ? outend - outptr : 2)); \
564 else if (set == GB2312_set) \
565 written = ucs4_to_gb2312 (ch, outptr, (NEED_LENGTH_TEST \
566 ? outend - outptr : 2)); \
567 else \
568 { \
569 assert (set == KSC5601_set); \
570 \
571 written = ucs4_to_ksc5601 (ch, outptr, \
572 (NEED_LENGTH_TEST \
573 ? outend - outptr : 2)); \
574 } \
575 \
576 if (NEED_LENGTH_TEST && __builtin_expect (written, 1) == 0) \
577 { \
578 result = __GCONV_FULL_OUTPUT; \
579 break; \
580 } \
581 else if (written != __UNKNOWN_10646_CHAR) \
582 outptr += written; \
583 } \
584 \
585 if (written == __UNKNOWN_10646_CHAR || written == 0) \
586 { \
587 if (set2 == ISO88591_set) \
588 { \
589 if (ch >= 0x80 && ch <= 0xff) \
590 { \
591 *outptr++ = ESC; \
592 *outptr++ = 'N'; \
593 *outptr++ = ch & 0x7f; \
594 written = 3; \
595 } \
596 } \
597 else if (set2 == ISO88597_set) \
598 { \
599 const struct gap *rp = from_idx; \
600 \
601 while (ch > rp->end) \
602 ++rp; \
603 if (ch >= rp->start) \
604 { \
605 unsigned char res = iso88597_from_ucs4[ch - 0xa0 + rp->idx]; \
606 if (res != '\0') \
607 { \
608 *outptr++ = ESC; \
609 *outptr++ = 'N'; \
610 *outptr++ = res; \
611 written = 3; \
612 } \
613 } \
614 } \
615 } \
616 \
617 if (written == __UNKNOWN_10646_CHAR || written == 0) \
618 { \
619 /* Either this is an unknown character or we have to switch \
620 the currently selected character set. The character sets \
621 do not code entirely separate parts of ISO 10646 and \
622 therefore there is no single correct result. If we choose \
623 the character set to use wrong we might be end up with \
624 using yet another character set for the next character \
625 though the current and the next could be encoded with one \
626 character set. We leave this kind of optimization for \
627 later and now simply use a fixed order in which we test for \
628 availability */ \
629 \
630 if (ch <= 0x7f) \
631 { \
632 /* We must encode using ASCII. First write out the \
633 escape sequence. */ \
634 if (NEED_LENGTH_TEST && __builtin_expect (outptr + 3 > outend, 0))\
635 { \
636 result = __GCONV_FULL_OUTPUT; \
637 break; \
638 } \
639 \
640 *outptr++ = ESC; \
641 *outptr++ = '('; \
642 *outptr++ = 'B'; \
643 set = ASCII_set; \
644 \
645 if (NEED_LENGTH_TEST && __builtin_expect (outptr + 1 > outend, 0))\
646 { \
647 result = __GCONV_FULL_OUTPUT; \
648 break; \
649 } \
650 *outptr++ = ch; \
651 \
652 /* At the beginning of a line, G2 designation is cleared. */ \
653 if (var == iso2022jp2 && ch == 0x0a) \
654 set2 = UNSPECIFIED_set; \
655 } \
656 else \
657 { \
658 /* Now it becomes difficult. We must search the other \
659 character sets one by one and we cannot use simple \
660 arithmetic to determine whether the character can be \
661 encoded using this set. */ \
662 size_t written; \
663 unsigned char buf[2]; \
664 \
665 written = ucs4_to_jisx0201 (ch, buf); \
666 if (written != __UNKNOWN_10646_CHAR && buf[0] < 0x80) \
667 { \
668 /* We use JIS X 0201. */ \
669 if (NEED_LENGTH_TEST \
670 && __builtin_expect (outptr + 3 > outend, 0)) \
671 { \
672 result = __GCONV_FULL_OUTPUT; \
673 break; \
674 } \
675 \
676 *outptr++ = ESC; \
677 *outptr++ = '('; \
678 *outptr++ = 'J'; \
679 set = JISX0201_Roman_set; \
680 \
681 if (NEED_LENGTH_TEST \
682 && __builtin_expect (outptr + 1 > outend, 0)) \
683 { \
684 result = __GCONV_FULL_OUTPUT; \
685 break; \
686 } \
687 *outptr++ = buf[0]; \
688 } \
689 else \
690 { \
691 written = ucs4_to_jisx0208 (ch, buf, 2); \
692 if (written != __UNKNOWN_10646_CHAR) \
693 { \
694 /* We use JIS X 0208. */ \
695 if (NEED_LENGTH_TEST \
696 && __builtin_expect (outptr + 3 > outend, 0)) \
697 { \
698 result = __GCONV_FULL_OUTPUT; \
699 break; \
700 } \
701 \
702 *outptr++ = ESC; \
703 *outptr++ = '$'; \
704 *outptr++ = 'B'; \
705 set = JISX0208_1983_set; \
706 \
707 if (NEED_LENGTH_TEST \
708 && __builtin_expect (outptr + 2 > outend, 0)) \
709 { \
710 result = __GCONV_FULL_OUTPUT; \
711 break; \
712 } \
713 *outptr++ = buf[0]; \
714 *outptr++ = buf[1]; \
715 } \
716 else if (__builtin_expect (var, iso2022jp2) == iso2022jp) \
717 { \
718 /* We have no other choice. */ \
719 if (! ignore_errors_p ()) \
720 { \
721 result = __GCONV_ILLEGAL_INPUT; \
722 break; \
723 } \
724 \
725 ++*converted; \
726 } \
727 else \
728 { \
729 written = ucs4_to_jisx0212 (ch, buf, 2); \
730 if (written != __UNKNOWN_10646_CHAR) \
731 { \
732 /* We use JIS X 0212. */ \
733 if (NEED_LENGTH_TEST \
734 && __builtin_expect (outptr + 4 > outend, 0)) \
735 { \
736 result = __GCONV_FULL_OUTPUT; \
737 break; \
738 } \
739 *outptr++ = ESC; \
740 *outptr++ = '$'; \
741 *outptr++ = '('; \
742 *outptr++ = 'D'; \
743 set = JISX0212_set; \
744 \
745 if (NEED_LENGTH_TEST \
746 && __builtin_expect (outptr + 2 > outend, 0)) \
747 { \
748 result = __GCONV_FULL_OUTPUT; \
749 break; \
750 } \
751 *outptr++ = buf[0]; \
752 *outptr++ = buf[1]; \
753 } \
754 else \
755 { \
756 written = ucs4_to_jisx0201 (ch, buf); \
757 if (written != __UNKNOWN_10646_CHAR \
758 && buf[0] >= 0x80) \
759 { \
760 /* We use JIS X 0201. */ \
761 if (NEED_LENGTH_TEST \
762 && __builtin_expect (outptr + 3 > outend, 0)) \
763 { \
764 result = __GCONV_FULL_OUTPUT; \
765 break; \
766 } \
767 \
768 *outptr++ = ESC; \
769 *outptr++ = '('; \
770 *outptr++ = 'I'; \
771 set = JISX0201_Kana_set; \
772 \
773 if (NEED_LENGTH_TEST \
774 && __builtin_expect (outptr + 1 > outend, 0)) \
775 { \
776 result = __GCONV_FULL_OUTPUT; \
777 break; \
778 } \
779 *outptr++ = buf[0] - 0x80; \
780 } \
781 else if (ch != 0xa5 && ch >= 0x80 && ch <= 0xff) \
782 { \
783 /* ISO 8859-1 upper half. */ \
784 if (NEED_LENGTH_TEST \
785 && __builtin_expect (outptr + 3 > outend, 0)) \
786 { \
787 result = __GCONV_FULL_OUTPUT; \
788 break; \
789 } \
790 \
791 *outptr++ = ESC; \
792 *outptr++ = '.'; \
793 *outptr++ = 'A'; \
794 set2 = ISO88591_set; \
795 \
796 if (NEED_LENGTH_TEST \
797 && __builtin_expect (outptr + 3 > outend, 0)) \
798 { \
799 result = __GCONV_FULL_OUTPUT; \
800 break; \
801 } \
802 *outptr++ = ESC; \
803 *outptr++ = 'N'; \
804 *outptr++ = ch; \
805 } \
806 else \
807 { \
808 written = ucs4_to_gb2312 (ch, buf, 2); \
809 if (written != __UNKNOWN_10646_CHAR) \
810 { \
811 /* We use GB 2312. */ \
812 if (NEED_LENGTH_TEST \
813 && __builtin_expect (outptr + 3 > outend, \
814 0)) \
815 { \
816 result = __GCONV_FULL_OUTPUT; \
817 break; \
818 } \
819 \
820 *outptr++ = ESC; \
821 *outptr++ = '$'; \
822 *outptr++ = 'A'; \
823 set = GB2312_set; \
824 \
825 if (NEED_LENGTH_TEST \
826 && __builtin_expect (outptr + 2 > outend, \
827 0)) \
828 { \
829 result = __GCONV_FULL_OUTPUT; \
830 break; \
831 } \
832 *outptr++ = buf[0]; \
833 *outptr++ = buf[1]; \
834 } \
835 else \
836 { \
837 written = ucs4_to_ksc5601 (ch, buf, 2); \
838 if (written != __UNKNOWN_10646_CHAR) \
839 { \
840 /* We use KSC 5601. */ \
841 if (NEED_LENGTH_TEST \
842 && __builtin_expect (outptr + 4 \
843 > outend, 0)) \
844 { \
845 result = __GCONV_FULL_OUTPUT; \
846 break; \
847 } \
848 *outptr++ = ESC; \
849 *outptr++ = '$'; \
850 *outptr++ = '('; \
851 *outptr++ = 'C'; \
852 set = KSC5601_set; \
853 \
854 if (NEED_LENGTH_TEST \
855 && __builtin_expect (outptr + 2 \
856 > outend, 0)) \
857 { \
858 result = __GCONV_FULL_OUTPUT; \
859 break; \
860 } \
861 *outptr++ = buf[0]; \
862 *outptr++ = buf[1]; \
863 } \
864 else \
865 { \
866 const struct gap *rp = from_idx; \
867 unsigned char gch = 0; \
868 \
869 while (ch > rp->end) \
870 ++rp; \
871 if (ch >= rp->start) \
872 { \
873 ch = ch - 0xa0 + rp->idx; \
874 gch = iso88597_from_ucs4[ch]; \
875 } \
876 \
877 if (__builtin_expect (gch, 1) != 0) \
878 { \
879 /* We use ISO 8859-7 greek. */ \
880 if (NEED_LENGTH_TEST \
881 && __builtin_expect (outptr + 3 \
882 > outend, 0))\
883 { \
884 result = __GCONV_FULL_OUTPUT; \
885 break; \
886 } \
887 *outptr++ = ESC; \
888 *outptr++ = '.'; \
889 *outptr++ = 'F'; \
890 set2 = ISO88597_set; \
891 \
892 if (NEED_LENGTH_TEST \
893 && __builtin_expect (outptr + 3 \
894 > outend, 0))\
895 { \
896 result = __GCONV_FULL_OUTPUT; \
897 break; \
898 } \
899 *outptr++ = ESC; \
900 *outptr++ = 'N'; \
901 *outptr++ = gch; \
902 } \
903 else \
904 { \
905 if (! ignore_errors_p ()) \
906 { \
907 result = __GCONV_ILLEGAL_INPUT; \
908 break; \
909 } \
910 \
911 ++*converted; \
912 } \
913 } \
914 } \
915 } \
916 } \
917 } \
918 } \
919 } \
920 } \
921 \
922 /* Now that we wrote the output increment the input pointer. */ \
923 inptr += 4; \
924 }
925 #define EXTRA_LOOP_DECLS , enum variant var, int *setp
926 #define INIT_PARAMS int set = *setp & CURRENT_SEL_MASK; \
927 int set2 = *setp & CURRENT_ASSIGN_MASK
928 #define UPDATE_PARAMS *setp = set | set2
929 #include <iconv/loop.c>
930
931
932 /* Now define the toplevel functions. */
933 #include <iconv/skeleton.c>
sources.redhat.com git mirror of glibc CVS