1 /* Conversion module for ISO-2022-CN-EXT.
2 Copyright (C) 2000-2024 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
5 The GNU C Library is free software; you can redistribute it and/or
6 modify it under the terms of the GNU Lesser General Public
7 License as published by the Free Software Foundation; either
8 version 2.1 of the License, or (at your option) any later version.
10 The GNU C Library is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 Lesser General Public License for more details.
15 You should have received a copy of the GNU Lesser General Public
16 License along with the GNU C Library; if not, see
17 <https://www.gnu.org/licenses/>. */
25 #include "iso-ir-165.h"
27 #include "cns11643l1.h"
28 #include "cns11643l2.h"
29 #include <libc-diag.h>
33 /* This makes obvious what everybody knows: 0x1b is the Esc character. */
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. */
46 /* Definitions used in the body of the `gconv' function. */
47 #define CHARSET_NAME "ISO-2022-CN-EXT//"
50 #define ONE_DIRECTION 0
51 #define FROM_LOOP from_iso2022cn_ext_loop
52 #define TO_LOOP to_iso2022cn_ext_loop
53 #define FROM_LOOP_MIN_NEEDED_FROM 1
54 #define FROM_LOOP_MAX_NEEDED_FROM 4
55 #define FROM_LOOP_MIN_NEEDED_TO 4
56 #define FROM_LOOP_MAX_NEEDED_TO 4
57 #define TO_LOOP_MIN_NEEDED_FROM 4
58 #define TO_LOOP_MAX_NEEDED_FROM 4
59 #define TO_LOOP_MIN_NEEDED_TO 1
60 #define TO_LOOP_MAX_NEEDED_TO 6
61 #define PREPARE_LOOP \
63 int *setp = &data->__statep->__count;
64 #define EXTRA_LOOP_ARGS , setp
67 /* The charsets GB/T 12345-90, GB 7589-87, GB/T 13131-9X, GB 7590-87,
68 and GB/T 13132-9X are not registered to the best of my knowledge and
69 therefore have no escape sequence assigned. We cannot handle them
70 for this reason. Tell the implementation about this. */
78 /* The COUNT element of the state keeps track of the currently selected
79 character set. The possible values are: */
91 CNS11643_2_set
= 3 << 3,
96 CNS11643_3_set
= 3 << 5,
97 CNS11643_4_set
= 4 << 5,
98 CNS11643_5_set
= 5 << 5,
99 CNS11643_6_set
= 6 << 5,
100 CNS11643_7_set
= 7 << 5,
103 #define CURRENT_MASK (SO_mask | SS2_mask | SS3_mask)
106 GB12345_ann
= 2 << 8,
107 CNS11643_1_ann
= 3 << 8,
108 ISO_IR_165_ann
= 4 << 8,
111 GB7589_ann
= 1 << 11,
112 GB13131_ann
= 2 << 11,
113 CNS11643_2_ann
= 3 << 11,
116 GB7590_ann
= 1 << 13,
117 GB13132_ann
= 2 << 13,
118 CNS11643_3_ann
= 3 << 13,
119 CNS11643_4_ann
= 4 << 13,
120 CNS11643_5_ann
= 5 << 13,
121 CNS11643_6_ann
= 6 << 13,
122 CNS11643_7_ann
= 7 << 13,
127 /* Since this is a stateful encoding we have to provide code which resets
128 the output state to the initial state. This has to be done during the
130 #define EMIT_SHIFT_TO_INIT \
131 if (data->__statep->__count >> 3 != ASCII_set) \
133 if (FROM_DIRECTION) \
134 /* It's easy, we don't have to emit anything, we just reset the \
135 state for the input. */ \
136 data->__statep->__count = ASCII_set << 3; \
139 /* We are not in the initial state. To switch back we have \
141 if (__glibc_unlikely (outbuf == outend)) \
142 /* We don't have enough room in the output buffer. */ \
143 status = __GCONV_FULL_OUTPUT; \
146 /* Write out the shift sequence. */ \
148 if (data->__flags & __GCONV_IS_LAST) \
149 *irreversible += 1; \
150 data->__statep->__count = ASCII_set << 3; \
156 /* Since we might have to reset input pointer we must be able to save
157 and restore the state. */
158 #define SAVE_RESET_STATE(Save) \
165 /* First define the conversion function from ISO-2022-CN to UCS4. */
166 #define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM
167 #define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM
168 #define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO
169 #define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO
170 #define LOOPFCT FROM_LOOP
173 uint32_t ch = *inptr; \
175 /* This is a 7bit character set, disallow all 8bit characters. */ \
177 STANDARD_FROM_LOOP_ERR_HANDLER (1); \
179 /* Recognize escape sequences. */ \
182 /* There are three kinds of escape sequences we have to handle: \
183 - those announcing the use of GB and CNS characters on the \
184 line; we can simply ignore them \
185 - the initial byte of the SS2 sequence. \
186 - the initial byte of the SS3 sequence. \
188 if (inptr + 2 > inend \
189 || (inptr[1] == '$' \
190 && (inptr + 3 > inend \
191 || (inptr[2] == ')' && inptr + 4 > inend) \
192 || (inptr[2] == '*' && inptr + 4 > inend) \
193 || (inptr[2] == '+' && inptr + 4 > inend))) \
194 || (inptr[1] == SS2_1 && inptr + 4 > inend) \
195 || (inptr[1] == SS3_1 && inptr + 4 > inend)) \
197 result = __GCONV_INCOMPLETE_INPUT; \
200 if (inptr[1] == '$' \
201 && ((inptr[2] == ')' \
202 && (inptr[3] == 'A' \
203 || (X12345 != '\0' && inptr[3] == X12345) \
204 || inptr[3] == 'E' || inptr[3] == 'G')) \
205 || (inptr[2] == '*' \
206 && ((X7589 != '\0' && inptr[3] == X7589) \
207 || (X13131 != '\0' && inptr[3] == X13131) \
208 || inptr[3] == 'H')) \
209 || (inptr[2] == '+' \
210 && ((X7590 != '\0' && inptr[3] == X7590) \
211 || (X13132 != '\0' && inptr[3] == X13132) \
212 || inptr[3] == 'I' || inptr[3] == 'J' \
213 || inptr[3] == 'K' || inptr[3] == 'L' \
214 || inptr[3] == 'M')))) \
216 /* OK, we accept those character sets. */ \
217 if (inptr[3] == 'A') \
218 ann = (ann & ~SO_ann) | GB2312_ann; \
219 else if (inptr[3] == 'G') \
220 ann = (ann & ~SO_ann) | CNS11643_1_ann; \
221 else if (inptr[3] == 'E') \
222 ann = (ann & ~SO_ann) | ISO_IR_165_ann; \
223 else if (X12345 != '\0' && inptr[3] == X12345) \
224 ann = (ann & ~SO_ann) | GB12345_ann; \
225 else if (inptr[3] == 'H') \
226 ann = (ann & ~SS2_ann) | CNS11643_2_ann; \
227 else if (X7589 != '\0' && inptr[3] == X7589) \
228 ann = (ann & ~SS2_ann) | GB7589_ann; \
229 else if (X13131 != '\0' && inptr[3] == X13131) \
230 ann = (ann & ~SS2_ann) | GB13131_ann; \
231 else if (inptr[3] == 'I') \
232 ann = (ann & ~SS3_ann) | CNS11643_3_ann; \
233 else if (inptr[3] == 'J') \
234 ann = (ann & ~SS3_ann) | CNS11643_4_ann; \
235 else if (inptr[3] == 'K') \
236 ann = (ann & ~SS3_ann) | CNS11643_5_ann; \
237 else if (inptr[3] == 'L') \
238 ann = (ann & ~SS3_ann) | CNS11643_6_ann; \
239 else if (inptr[3] == 'M') \
240 ann = (ann & ~SS3_ann) | CNS11643_7_ann; \
241 else if (X7590 != '\0' && inptr[3] == X7590) \
242 ann = (ann & ~SS3_ann) | GB7590_ann; \
243 else if (X13132 != '\0' && inptr[3] == X13132) \
244 ann = (ann & ~SS3_ann) | GB13132_ann; \
251 /* Switch to use GB2312, GB12345, CNS 11643 plane 1, or ISO-IR-165, \
252 depending on which S0 designation came last. The only problem \
253 is what to do with faulty input files where no designator came. \
254 XXX For now I'll default to use GB2312. If this is not the \
255 best behavior (e.g., we should flag an error) let me know. */ \
257 if ((ann & SO_ann) != 0) \
258 switch (ann & SO_ann) \
266 case CNS11643_1_ann: \
267 set = CNS11643_1_set; \
269 case ISO_IR_165_ann: \
270 set = ISO_IR_165_set; \
277 STANDARD_FROM_LOOP_ERR_HANDLER (1); \
283 /* Switch to use ASCII. */ \
289 if (ch == ESC && inptr[1] == SS2_1) \
291 /* This is a character from CNS 11643 plane 2. \
292 XXX We could test here whether the use of this character \
294 XXX Currently GB7589 and GB13131 are not supported. */ \
296 ch = cns11643l2_to_ucs4 (&inptr, 2, 0); \
297 if (ch == __UNKNOWN_10646_CHAR) \
298 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
300 /* Note that we can assume here that at least 4 bytes are available if \
301 the first byte is ESC since otherwise the first if would have been \
303 else if (ch == ESC && inptr[1] == SS3_1) \
305 /* This is a character from CNS 11643 plane 3 or higher. \
306 XXX Currently GB7590 and GB13132 are not supported. */ \
307 unsigned char buf[3]; \
308 const unsigned char *tmp = buf; \
312 switch (ann & SS3_ann) \
314 case CNS11643_3_ann: \
316 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
318 case CNS11643_4_ann: \
320 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
322 case CNS11643_5_ann: \
324 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
326 case CNS11643_6_ann: \
328 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
330 case CNS11643_7_ann: \
332 ch = cns11643_to_ucs4 (&tmp, 3, 0); \
335 /* XXX Currently GB7590 and GB13132 are not supported. */ \
336 ch = __UNKNOWN_10646_CHAR; \
339 if (ch == __UNKNOWN_10646_CHAR) \
341 STANDARD_FROM_LOOP_ERR_HANDLER (4); \
343 assert (tmp == buf + 3); \
346 else if (set == ASCII_set) \
348 /* Almost done, just advance the input pointer. */ \
353 /* That's pretty easy, we have a dedicated functions for this. */ \
354 if (inend - inptr < 2) \
356 result = __GCONV_INCOMPLETE_INPUT; \
359 if (set == GB2312_set) \
360 ch = gb2312_to_ucs4 (&inptr, inend - inptr, 0); \
361 else if (set == ISO_IR_165_set) \
362 ch = isoir165_to_ucs4 (&inptr, inend - inptr); \
365 assert (set == CNS11643_1_set); \
366 ch = cns11643l1_to_ucs4 (&inptr, inend - inptr, 0); \
371 result = __GCONV_INCOMPLETE_INPUT; \
374 else if (ch == __UNKNOWN_10646_CHAR) \
376 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
380 *((uint32_t *) outptr) = ch; \
381 outptr += sizeof (uint32_t); \
383 #define EXTRA_LOOP_DECLS , int *setp
384 #define INIT_PARAMS int set = (*setp >> 3) & CURRENT_MASK; \
385 int ann = (*setp >> 3) & ~CURRENT_MASK
386 #define UPDATE_PARAMS *setp = (set | ann) << 3
387 #define LOOP_NEED_FLAGS
388 #include <iconv/loop.c>
391 /* Next, define the other direction. */
392 #define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM
393 #define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM
394 #define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO
395 #define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO
396 #define LOOPFCT TO_LOOP
397 /* With GCC 5.3 when compiling with -Os the compiler emits a warning
398 that buf[0] and buf[1] may be used uninitialized. This can only
399 happen in the case where tmpbuf[3] is used, and in that case the
400 write to the tmpbuf[1] and tmpbuf[2] was assured because
401 ucs4_to_cns11643 would have filled in those entries. The difficulty
402 is in getting the compiler to see this logic because tmpbuf[0] is
403 involved in determining the code page and is the indicator that
404 tmpbuf[2] is initialized. */
405 DIAG_PUSH_NEEDS_COMMENT
;
406 DIAG_IGNORE_Os_NEEDS_COMMENT (5, "-Wmaybe-uninitialized");
410 size_t written = 0; \
412 ch = *((const uint32_t *) inptr); \
414 /* First see whether we can write the character using the currently \
415 selected character set. */ \
418 if (set != ASCII_set) \
422 if (outptr == outend) \
424 result = __GCONV_FULL_OUTPUT; \
432 /* At the end of the line we have to clear the `ann' flags since \
433 every line must contain this information again. */ \
439 unsigned char buf[2] = { 0, 0 }; \
442 if (set == GB2312_set || ((ann & SO_ann) != CNS11643_1_ann \
443 && (ann & SO_ann) != ISO_IR_165_ann)) \
445 written = ucs4_to_gb2312 (ch, buf, 2); \
448 else if (set == ISO_IR_165_set || (ann & SO_ann) == ISO_IR_165_set) \
450 written = ucs4_to_isoir165 (ch, buf, 2); \
451 used = ISO_IR_165_set; \
455 written = ucs4_to_cns11643l1 (ch, buf, 2); \
456 used = CNS11643_1_set; \
459 if (written == __UNKNOWN_10646_CHAR) \
461 /* Cannot convert it using the currently selected SO set. \
462 Next try the SS2 set. */ \
463 written = ucs4_to_cns11643l2 (ch, buf, 2); \
464 if (written != __UNKNOWN_10646_CHAR) \
465 /* Yep, that worked. */ \
466 used = CNS11643_2_set; \
469 unsigned char tmpbuf[3]; \
474 /* Well, see whether we have to change the SO set. */ \
476 if (used != GB2312_set) \
478 written = ucs4_to_gb2312 (ch, buf, 2); \
479 if (written != __UNKNOWN_10646_CHAR) \
486 if (used != ISO_IR_165_set) \
488 written = ucs4_to_isoir165 (ch, buf, 2); \
489 if (written != __UNKNOWN_10646_CHAR) \
491 used = ISO_IR_165_set; \
496 if (used != CNS11643_1_set) \
498 written = ucs4_to_cns11643l1 (ch, buf, 2); \
499 if (written != __UNKNOWN_10646_CHAR) \
501 used = CNS11643_1_set; \
506 written = ucs4_to_cns11643 (ch, tmpbuf, 3); \
507 if (written == 3 && tmpbuf[0] >= 3 && tmpbuf[0] <= 7) \
509 buf[0] = tmpbuf[1]; \
510 buf[1] = tmpbuf[2]; \
514 used = CNS11643_3_set; \
517 used = CNS11643_4_set; \
520 used = CNS11643_5_set; \
523 used = CNS11643_6_set; \
526 used = CNS11643_7_set; \
535 /* XXX Currently GB7590 and GB13132 are not supported. */\
537 /* Even this does not work. Error. */ \
540 if (used == ASCII_set) \
542 UNICODE_TAG_HANDLER (ch, 4); \
543 STANDARD_TO_LOOP_ERR_HANDLER (4); \
547 assert (written == 2); \
549 /* See whether we have to emit an escape sequence. */ \
552 /* First see whether we announced that we use this \
554 if ((used & SO_mask) != 0 && (ann & SO_ann) != (used << 8)) \
556 const char *escseq; \
558 if (outptr + 4 > outend) \
560 result = __GCONV_FULL_OUTPUT; \
564 assert (used >= 1 && used <= 4); \
565 escseq = ")A\0\0)G)E" + (used - 1) * 2; \
568 *outptr++ = *escseq++; \
569 *outptr++ = *escseq++; \
571 ann = (ann & ~SO_ann) | (used << 8); \
573 else if ((used & SS2_mask) != 0 && (ann & SS2_ann) != (used << 8))\
575 const char *escseq; \
577 if (outptr + 4 > outend) \
579 result = __GCONV_FULL_OUTPUT; \
583 assert (used == CNS11643_2_set); /* XXX */ \
587 *outptr++ = *escseq++; \
588 *outptr++ = *escseq++; \
590 ann = (ann & ~SS2_ann) | (used << 8); \
592 else if ((used & SS3_mask) != 0 && (ann & SS3_ann) != (used << 8))\
594 const char *escseq; \
596 if (outptr + 4 > outend) \
598 result = __GCONV_FULL_OUTPUT; \
602 assert ((used >> 5) >= 3 && (used >> 5) <= 7); \
603 escseq = "+I+J+K+L+M" + ((used >> 5) - 3) * 2; \
606 *outptr++ = *escseq++; \
607 *outptr++ = *escseq++; \
609 ann = (ann & ~SS3_ann) | (used << 8); \
612 if (used == CNS11643_2_set) \
614 if (outptr + 2 > outend) \
616 result = __GCONV_FULL_OUTPUT; \
622 else if (used >= CNS11643_3_set && used <= CNS11643_7_set) \
624 if (outptr + 2 > outend) \
626 result = __GCONV_FULL_OUTPUT; \
634 /* We only have to emit something if currently ASCII is \
635 selected. Otherwise we are switching within the \
637 if (set == ASCII_set) \
639 if (outptr + 1 > outend) \
641 result = __GCONV_FULL_OUTPUT; \
648 /* Always test the length here since we have used up all the \
649 guaranteed output buffer slots. */ \
650 if (outptr + 2 > outend) \
652 result = __GCONV_FULL_OUTPUT; \
656 else if (outptr + 2 > outend) \
658 result = __GCONV_FULL_OUTPUT; \
662 *outptr++ = buf[0]; \
663 *outptr++ = buf[1]; \
667 /* Now that we wrote the output increment the input pointer. */ \
670 DIAG_POP_NEEDS_COMMENT
;
671 #define EXTRA_LOOP_DECLS , int *setp
672 #define INIT_PARAMS int set = (*setp >> 3) & CURRENT_MASK; \
673 int ann = (*setp >> 3) & ~CURRENT_MASK
674 #define REINIT_PARAMS do \
676 set = (*setp >> 3) & CURRENT_MASK; \
677 ann = (*setp >> 3) & ~CURRENT_MASK; \
680 #define UPDATE_PARAMS *setp = (set | ann) << 3
681 #define LOOP_NEED_FLAGS
682 #include <iconv/loop.c>
685 /* Now define the toplevel functions. */
686 #include <iconv/skeleton.c>