1 | /* Conversion module for ISO-2022-CN-EXT. |
2 | Copyright (C) 2000-2023 Free Software Foundation, Inc. |
3 | This file is part of the GNU C Library. |
4 | |
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. |
9 | |
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. |
14 | |
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/>. */ |
18 | |
19 | #include <dlfcn.h> |
20 | #include <gconv.h> |
21 | #include <stdint.h> |
22 | #include <stdlib.h> |
23 | #include <string.h> |
24 | #include "gb2312.h" |
25 | #include "iso-ir-165.h" |
26 | #include "cns11643.h" |
27 | #include "cns11643l1.h" |
28 | #include "cns11643l2.h" |
29 | #include <libc-diag.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 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 \ |
62 | int save_set; \ |
63 | int *setp = &data->__statep->__count; |
64 | #define , setp |
65 | |
66 | |
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. */ |
71 | #define X12345 '\0' |
72 | #define X7589 '\0' |
73 | #define X13131 '\0' |
74 | #define X7590 '\0' |
75 | #define X13132 '\0' |
76 | |
77 | |
78 | /* The COUNT element of the state keeps track of the currently selected |
79 | character set. The possible values are: */ |
80 | enum |
81 | { |
82 | ASCII_set = 0, |
83 | GB2312_set, |
84 | GB12345_set, |
85 | CNS11643_1_set, |
86 | ISO_IR_165_set, |
87 | SO_mask = 7, |
88 | |
89 | GB7589_set = 1 << 3, |
90 | GB13131_set = 2 << 3, |
91 | CNS11643_2_set = 3 << 3, |
92 | SS2_mask = 3 << 3, |
93 | |
94 | GB7590_set = 1 << 5, |
95 | GB13132_set = 2 << 5, |
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, |
101 | SS3_mask = 7 << 5, |
102 | |
103 | #define CURRENT_MASK (SO_mask | SS2_mask | SS3_mask) |
104 | |
105 | GB2312_ann = 1 << 8, |
106 | GB12345_ann = 2 << 8, |
107 | CNS11643_1_ann = 3 << 8, |
108 | ISO_IR_165_ann = 4 << 8, |
109 | SO_ann = 7 << 8, |
110 | |
111 | GB7589_ann = 1 << 11, |
112 | GB13131_ann = 2 << 11, |
113 | CNS11643_2_ann = 3 << 11, |
114 | SS2_ann = 3 << 11, |
115 | |
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, |
123 | SS3_ann = 7 << 13 |
124 | }; |
125 | |
126 | |
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 |
129 | flushing. */ |
130 | #define EMIT_SHIFT_TO_INIT \ |
131 | if (data->__statep->__count >> 3 != ASCII_set) \ |
132 | { \ |
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; \ |
137 | else \ |
138 | { \ |
139 | /* We are not in the initial state. To switch back we have \ |
140 | to emit `SI'. */ \ |
141 | if (__glibc_unlikely (outbuf == outend)) \ |
142 | /* We don't have enough room in the output buffer. */ \ |
143 | status = __GCONV_FULL_OUTPUT; \ |
144 | else \ |
145 | { \ |
146 | /* Write out the shift sequence. */ \ |
147 | *outbuf++ = SI; \ |
148 | if (data->__flags & __GCONV_IS_LAST) \ |
149 | *irreversible += 1; \ |
150 | data->__statep->__count = ASCII_set << 3; \ |
151 | } \ |
152 | } \ |
153 | } |
154 | |
155 | |
156 | /* Since we might have to reset input pointer we must be able to save |
157 | and retore the state. */ |
158 | #define SAVE_RESET_STATE(Save) \ |
159 | if (Save) \ |
160 | save_set = *setp; \ |
161 | else \ |
162 | *setp = save_set |
163 | |
164 | |
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 |
171 | #define BODY \ |
172 | { \ |
173 | uint32_t ch = *inptr; \ |
174 | \ |
175 | /* This is a 7bit character set, disallow all 8bit characters. */ \ |
176 | if (ch > 0x7f) \ |
177 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
178 | \ |
179 | /* Recognize escape sequences. */ \ |
180 | if (ch == ESC) \ |
181 | { \ |
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. \ |
187 | */ \ |
188 | if (inptr + 2 > inend \ |
189 | || (inptr[1] == '$' \ |
190 | && (inptr + 3 > inend \ |
191 | || (inptr[2] == ')' && inptr + 4 > inend) \ |
192 | || (inptr[2] == '*' && inptr + 4 > inend) \ |
193 | || (inptr[2] == '+' && inptr + 4 > inend))) \ |
194 | || (inptr[1] == SS2_1 && inptr + 4 > inend) \ |
195 | || (inptr[1] == SS3_1 && inptr + 4 > inend)) \ |
196 | { \ |
197 | result = __GCONV_INCOMPLETE_INPUT; \ |
198 | break; \ |
199 | } \ |
200 | if (inptr[1] == '$' \ |
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')))) \ |
215 | { \ |
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; \ |
245 | inptr += 4; \ |
246 | continue; \ |
247 | } \ |
248 | } \ |
249 | else if (ch == SO) \ |
250 | { \ |
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. */ \ |
256 | ++inptr; \ |
257 | if ((ann & SO_ann) != 0) \ |
258 | switch (ann & SO_ann) \ |
259 | { \ |
260 | case GB2312_ann: \ |
261 | set = GB2312_set; \ |
262 | break; \ |
263 | case GB12345_ann: \ |
264 | set = GB12345_set; \ |
265 | break; \ |
266 | case CNS11643_1_ann: \ |
267 | set = CNS11643_1_set; \ |
268 | break; \ |
269 | case ISO_IR_165_ann: \ |
270 | set = ISO_IR_165_set; \ |
271 | break; \ |
272 | default: \ |
273 | abort (); \ |
274 | } \ |
275 | else \ |
276 | { \ |
277 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
278 | } \ |
279 | continue; \ |
280 | } \ |
281 | else if (ch == SI) \ |
282 | { \ |
283 | /* Switch to use ASCII. */ \ |
284 | ++inptr; \ |
285 | set = ASCII_set; \ |
286 | continue; \ |
287 | } \ |
288 | \ |
289 | if (ch == ESC && inptr[1] == SS2_1) \ |
290 | { \ |
291 | /* This is a character from CNS 11643 plane 2. \ |
292 | XXX We could test here whether the use of this character \ |
293 | set was announced. \ |
294 | XXX Currently GB7589 and GB13131 are not supported. */ \ |
295 | inptr += 2; \ |
296 | ch = cns11643l2_to_ucs4 (&inptr, 2, 0); \ |
297 | if (ch == __UNKNOWN_10646_CHAR) \ |
298 | STANDARD_FROM_LOOP_ERR_HANDLER (2); \ |
299 | } \ |
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 \ |
302 | true. */ \ |
303 | else if (ch == ESC && inptr[1] == SS3_1) \ |
304 | { \ |
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; \ |
309 | \ |
310 | buf[1] = inptr[2]; \ |
311 | buf[2] = inptr[3]; \ |
312 | switch (ann & SS3_ann) \ |
313 | { \ |
314 | case CNS11643_3_ann: \ |
315 | buf[0] = 0x23; \ |
316 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
317 | break; \ |
318 | case CNS11643_4_ann: \ |
319 | buf[0] = 0x24; \ |
320 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
321 | break; \ |
322 | case CNS11643_5_ann: \ |
323 | buf[0] = 0x25; \ |
324 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
325 | break; \ |
326 | case CNS11643_6_ann: \ |
327 | buf[0] = 0x26; \ |
328 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
329 | break; \ |
330 | case CNS11643_7_ann: \ |
331 | buf[0] = 0x27; \ |
332 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
333 | break; \ |
334 | default: \ |
335 | /* XXX Currently GB7590 and GB13132 are not supported. */ \ |
336 | ch = __UNKNOWN_10646_CHAR; \ |
337 | break; \ |
338 | } \ |
339 | if (ch == __UNKNOWN_10646_CHAR) \ |
340 | { \ |
341 | STANDARD_FROM_LOOP_ERR_HANDLER (4); \ |
342 | } \ |
343 | assert (tmp == buf + 3); \ |
344 | inptr += 4; \ |
345 | } \ |
346 | else if (set == ASCII_set) \ |
347 | { \ |
348 | /* Almost done, just advance the input pointer. */ \ |
349 | ++inptr; \ |
350 | } \ |
351 | else \ |
352 | { \ |
353 | /* That's pretty easy, we have a dedicated functions for this. */ \ |
354 | if (inend - inptr < 2) \ |
355 | { \ |
356 | result = __GCONV_INCOMPLETE_INPUT; \ |
357 | break; \ |
358 | } \ |
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); \ |
363 | else \ |
364 | { \ |
365 | assert (set == CNS11643_1_set); \ |
366 | ch = cns11643l1_to_ucs4 (&inptr, inend - inptr, 0); \ |
367 | } \ |
368 | \ |
369 | if (ch == 0) \ |
370 | { \ |
371 | result = __GCONV_INCOMPLETE_INPUT; \ |
372 | break; \ |
373 | } \ |
374 | else if (ch == __UNKNOWN_10646_CHAR) \ |
375 | { \ |
376 | STANDARD_FROM_LOOP_ERR_HANDLER (2); \ |
377 | } \ |
378 | } \ |
379 | \ |
380 | *((uint32_t *) outptr) = ch; \ |
381 | outptr += sizeof (uint32_t); \ |
382 | } |
383 | #define , 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> |
389 | |
390 | |
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" ); |
407 | #define BODY \ |
408 | { \ |
409 | uint32_t ch; \ |
410 | size_t written = 0; \ |
411 | \ |
412 | ch = *((const uint32_t *) inptr); \ |
413 | \ |
414 | /* First see whether we can write the character using the currently \ |
415 | selected character set. */ \ |
416 | if (ch < 0x80) \ |
417 | { \ |
418 | if (set != ASCII_set) \ |
419 | { \ |
420 | *outptr++ = SI; \ |
421 | set = ASCII_set; \ |
422 | if (outptr == outend) \ |
423 | { \ |
424 | result = __GCONV_FULL_OUTPUT; \ |
425 | break; \ |
426 | } \ |
427 | } \ |
428 | \ |
429 | *outptr++ = ch; \ |
430 | written = 1; \ |
431 | \ |
432 | /* At the end of the line we have to clear the `ann' flags since \ |
433 | every line must contain this information again. */ \ |
434 | if (ch == L'\n') \ |
435 | ann = 0; \ |
436 | } \ |
437 | else \ |
438 | { \ |
439 | unsigned char buf[2] = { 0, 0 }; \ |
440 | int used; \ |
441 | \ |
442 | if (set == GB2312_set || ((ann & SO_ann) != CNS11643_1_ann \ |
443 | && (ann & SO_ann) != ISO_IR_165_ann)) \ |
444 | { \ |
445 | written = ucs4_to_gb2312 (ch, buf, 2); \ |
446 | used = GB2312_set; \ |
447 | } \ |
448 | else if (set == ISO_IR_165_set || (ann & SO_ann) == ISO_IR_165_set) \ |
449 | { \ |
450 | written = ucs4_to_isoir165 (ch, buf, 2); \ |
451 | used = ISO_IR_165_set; \ |
452 | } \ |
453 | else \ |
454 | { \ |
455 | written = ucs4_to_cns11643l1 (ch, buf, 2); \ |
456 | used = CNS11643_1_set; \ |
457 | } \ |
458 | \ |
459 | if (written == __UNKNOWN_10646_CHAR) \ |
460 | { \ |
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; \ |
467 | else \ |
468 | { \ |
469 | unsigned char tmpbuf[3]; \ |
470 | \ |
471 | switch (0) \ |
472 | { \ |
473 | default: \ |
474 | /* Well, see whether we have to change the SO set. */ \ |
475 | \ |
476 | if (used != GB2312_set) \ |
477 | { \ |
478 | written = ucs4_to_gb2312 (ch, buf, 2); \ |
479 | if (written != __UNKNOWN_10646_CHAR) \ |
480 | { \ |
481 | used = GB2312_set; \ |
482 | break; \ |
483 | } \ |
484 | } \ |
485 | \ |
486 | if (used != ISO_IR_165_set) \ |
487 | { \ |
488 | written = ucs4_to_isoir165 (ch, buf, 2); \ |
489 | if (written != __UNKNOWN_10646_CHAR) \ |
490 | { \ |
491 | used = ISO_IR_165_set; \ |
492 | break; \ |
493 | } \ |
494 | } \ |
495 | \ |
496 | if (used != CNS11643_1_set) \ |
497 | { \ |
498 | written = ucs4_to_cns11643l1 (ch, buf, 2); \ |
499 | if (written != __UNKNOWN_10646_CHAR) \ |
500 | { \ |
501 | used = CNS11643_1_set; \ |
502 | break; \ |
503 | } \ |
504 | } \ |
505 | \ |
506 | written = ucs4_to_cns11643 (ch, tmpbuf, 3); \ |
507 | if (written == 3 && tmpbuf[0] >= 3 && tmpbuf[0] <= 7) \ |
508 | { \ |
509 | buf[0] = tmpbuf[1]; \ |
510 | buf[1] = tmpbuf[2]; \ |
511 | switch (tmpbuf[0]) \ |
512 | { \ |
513 | case 3: \ |
514 | used = CNS11643_3_set; \ |
515 | break; \ |
516 | case 4: \ |
517 | used = CNS11643_4_set; \ |
518 | break; \ |
519 | case 5: \ |
520 | used = CNS11643_5_set; \ |
521 | break; \ |
522 | case 6: \ |
523 | used = CNS11643_6_set; \ |
524 | break; \ |
525 | case 7: \ |
526 | used = CNS11643_7_set; \ |
527 | break; \ |
528 | default: \ |
529 | abort (); \ |
530 | } \ |
531 | written = 2; \ |
532 | break; \ |
533 | } \ |
534 | \ |
535 | /* XXX Currently GB7590 and GB13132 are not supported. */\ |
536 | \ |
537 | /* Even this does not work. Error. */ \ |
538 | used = ASCII_set; \ |
539 | } \ |
540 | if (used == ASCII_set) \ |
541 | { \ |
542 | UNICODE_TAG_HANDLER (ch, 4); \ |
543 | STANDARD_TO_LOOP_ERR_HANDLER (4); \ |
544 | } \ |
545 | } \ |
546 | } \ |
547 | assert (written == 2); \ |
548 | \ |
549 | /* See whether we have to emit an escape sequence. */ \ |
550 | if (set != used) \ |
551 | { \ |
552 | /* First see whether we announced that we use this \ |
553 | character set. */ \ |
554 | if ((used & SO_mask) != 0 && (ann & SO_ann) != (used << 8)) \ |
555 | { \ |
556 | const char *escseq; \ |
557 | \ |
558 | if (outptr + 4 > outend) \ |
559 | { \ |
560 | result = __GCONV_FULL_OUTPUT; \ |
561 | break; \ |
562 | } \ |
563 | \ |
564 | assert (used >= 1 && used <= 4); \ |
565 | escseq = ")A\0\0)G)E" + (used - 1) * 2; \ |
566 | *outptr++ = ESC; \ |
567 | *outptr++ = '$'; \ |
568 | *outptr++ = *escseq++; \ |
569 | *outptr++ = *escseq++; \ |
570 | \ |
571 | ann = (ann & ~SO_ann) | (used << 8); \ |
572 | } \ |
573 | else if ((used & SS2_mask) != 0 && (ann & SS2_ann) != (used << 8))\ |
574 | { \ |
575 | const char *escseq; \ |
576 | \ |
577 | assert (used == CNS11643_2_set); /* XXX */ \ |
578 | escseq = "*H"; \ |
579 | *outptr++ = ESC; \ |
580 | *outptr++ = '$'; \ |
581 | *outptr++ = *escseq++; \ |
582 | *outptr++ = *escseq++; \ |
583 | \ |
584 | ann = (ann & ~SS2_ann) | (used << 8); \ |
585 | } \ |
586 | else if ((used & SS3_mask) != 0 && (ann & SS3_ann) != (used << 8))\ |
587 | { \ |
588 | const char *escseq; \ |
589 | \ |
590 | assert ((used >> 5) >= 3 && (used >> 5) <= 7); \ |
591 | escseq = "+I+J+K+L+M" + ((used >> 5) - 3) * 2; \ |
592 | *outptr++ = ESC; \ |
593 | *outptr++ = '$'; \ |
594 | *outptr++ = *escseq++; \ |
595 | *outptr++ = *escseq++; \ |
596 | \ |
597 | ann = (ann & ~SS3_ann) | (used << 8); \ |
598 | } \ |
599 | \ |
600 | if (used == CNS11643_2_set) \ |
601 | { \ |
602 | if (outptr + 2 > outend) \ |
603 | { \ |
604 | result = __GCONV_FULL_OUTPUT; \ |
605 | break; \ |
606 | } \ |
607 | *outptr++ = SS2_0; \ |
608 | *outptr++ = SS2_1; \ |
609 | } \ |
610 | else if (used >= CNS11643_3_set && used <= CNS11643_7_set) \ |
611 | { \ |
612 | if (outptr + 2 > outend) \ |
613 | { \ |
614 | result = __GCONV_FULL_OUTPUT; \ |
615 | break; \ |
616 | } \ |
617 | *outptr++ = SS3_0; \ |
618 | *outptr++ = SS3_1; \ |
619 | } \ |
620 | else \ |
621 | { \ |
622 | /* We only have to emit something if currently ASCII is \ |
623 | selected. Otherwise we are switching within the \ |
624 | SO charset. */ \ |
625 | if (set == ASCII_set) \ |
626 | { \ |
627 | if (outptr + 1 > outend) \ |
628 | { \ |
629 | result = __GCONV_FULL_OUTPUT; \ |
630 | break; \ |
631 | } \ |
632 | *outptr++ = SO; \ |
633 | } \ |
634 | } \ |
635 | \ |
636 | /* Always test the length here since we have used up all the \ |
637 | guaranteed output buffer slots. */ \ |
638 | if (outptr + 2 > outend) \ |
639 | { \ |
640 | result = __GCONV_FULL_OUTPUT; \ |
641 | break; \ |
642 | } \ |
643 | } \ |
644 | else if (outptr + 2 > outend) \ |
645 | { \ |
646 | result = __GCONV_FULL_OUTPUT; \ |
647 | break; \ |
648 | } \ |
649 | \ |
650 | *outptr++ = buf[0]; \ |
651 | *outptr++ = buf[1]; \ |
652 | set = used; \ |
653 | } \ |
654 | \ |
655 | /* Now that we wrote the output increment the input pointer. */ \ |
656 | inptr += 4; \ |
657 | } |
658 | DIAG_POP_NEEDS_COMMENT; |
659 | #define , int *setp |
660 | #define INIT_PARAMS int set = (*setp >> 3) & CURRENT_MASK; \ |
661 | int ann = (*setp >> 3) & ~CURRENT_MASK |
662 | #define REINIT_PARAMS do \ |
663 | { \ |
664 | set = (*setp >> 3) & CURRENT_MASK; \ |
665 | ann = (*setp >> 3) & ~CURRENT_MASK; \ |
666 | } \ |
667 | while (0) |
668 | #define UPDATE_PARAMS *setp = (set | ann) << 3 |
669 | #define LOOP_NEED_FLAGS |
670 | #include <iconv/loop.c> |
671 | |
672 | |
673 | /* Now define the toplevel functions. */ |
674 | #include <iconv/skeleton.c> |
675 | |