1 | /* Conversion module for ISO-2022-CN-EXT. |
2 | Copyright (C) 2000-2021 Free Software Foundation, Inc. |
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 |
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. |
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 | Lesser General Public License for more details. |
15 | |
16 | You should have received a copy of the GNU Lesser General Public |
17 | License along with the GNU C Library; if not, see |
18 | <https://www.gnu.org/licenses/>. */ |
19 | |
20 | #include <dlfcn.h> |
21 | #include <gconv.h> |
22 | #include <stdint.h> |
23 | #include <stdlib.h> |
24 | #include <string.h> |
25 | #include "gb2312.h" |
26 | #include "iso-ir-165.h" |
27 | #include "cns11643.h" |
28 | #include "cns11643l1.h" |
29 | #include "cns11643l2.h" |
30 | #include <libc-diag.h> |
31 | |
32 | #include <assert.h> |
33 | |
34 | /* This makes obvious what everybody knows: 0x1b is the Esc character. */ |
35 | #define ESC 0x1b |
36 | |
37 | /* We have single-byte shift-in and shift-out sequences, and the single |
38 | shift sequences SS2 and SS3 which replaces the SS2/SS3 designation for |
39 | the next two bytes. */ |
40 | #define SI 0x0f |
41 | #define SO 0x0e |
42 | #define SS2_0 ESC |
43 | #define SS2_1 0x4e |
44 | #define SS3_0 ESC |
45 | #define SS3_1 0x4f |
46 | |
47 | /* Definitions used in the body of the `gconv' function. */ |
48 | #define CHARSET_NAME "ISO-2022-CN-EXT//" |
49 | #define DEFINE_INIT 1 |
50 | #define DEFINE_FINI 1 |
51 | #define ONE_DIRECTION 0 |
52 | #define FROM_LOOP from_iso2022cn_ext_loop |
53 | #define TO_LOOP to_iso2022cn_ext_loop |
54 | #define FROM_LOOP_MIN_NEEDED_FROM 1 |
55 | #define FROM_LOOP_MAX_NEEDED_FROM 4 |
56 | #define FROM_LOOP_MIN_NEEDED_TO 4 |
57 | #define FROM_LOOP_MAX_NEEDED_TO 4 |
58 | #define TO_LOOP_MIN_NEEDED_FROM 4 |
59 | #define TO_LOOP_MAX_NEEDED_FROM 4 |
60 | #define TO_LOOP_MIN_NEEDED_TO 1 |
61 | #define TO_LOOP_MAX_NEEDED_TO 6 |
62 | #define PREPARE_LOOP \ |
63 | int save_set; \ |
64 | int *setp = &data->__statep->__count; |
65 | #define , setp |
66 | |
67 | |
68 | /* The charsets GB/T 12345-90, GB 7589-87, GB/T 13131-9X, GB 7590-87, |
69 | and GB/T 13132-9X are not registered to the best of my knowledge and |
70 | therefore have no escape sequence assigned. We cannot handle them |
71 | for this reason. Tell the implementation about this. */ |
72 | #define X12345 '\0' |
73 | #define X7589 '\0' |
74 | #define X13131 '\0' |
75 | #define X7590 '\0' |
76 | #define X13132 '\0' |
77 | |
78 | |
79 | /* The COUNT element of the state keeps track of the currently selected |
80 | character set. The possible values are: */ |
81 | enum |
82 | { |
83 | ASCII_set = 0, |
84 | GB2312_set, |
85 | GB12345_set, |
86 | CNS11643_1_set, |
87 | ISO_IR_165_set, |
88 | SO_mask = 7, |
89 | |
90 | GB7589_set = 1 << 3, |
91 | GB13131_set = 2 << 3, |
92 | CNS11643_2_set = 3 << 3, |
93 | SS2_mask = 3 << 3, |
94 | |
95 | GB7590_set = 1 << 5, |
96 | GB13132_set = 2 << 5, |
97 | CNS11643_3_set = 3 << 5, |
98 | CNS11643_4_set = 4 << 5, |
99 | CNS11643_5_set = 5 << 5, |
100 | CNS11643_6_set = 6 << 5, |
101 | CNS11643_7_set = 7 << 5, |
102 | SS3_mask = 7 << 5, |
103 | |
104 | #define CURRENT_MASK (SO_mask | SS2_mask | SS3_mask) |
105 | |
106 | GB2312_ann = 1 << 8, |
107 | GB12345_ann = 2 << 8, |
108 | CNS11643_1_ann = 3 << 8, |
109 | ISO_IR_165_ann = 4 << 8, |
110 | SO_ann = 7 << 8, |
111 | |
112 | GB7589_ann = 1 << 11, |
113 | GB13131_ann = 2 << 11, |
114 | CNS11643_2_ann = 3 << 11, |
115 | SS2_ann = 3 << 11, |
116 | |
117 | GB7590_ann = 1 << 13, |
118 | GB13132_ann = 2 << 13, |
119 | CNS11643_3_ann = 3 << 13, |
120 | CNS11643_4_ann = 4 << 13, |
121 | CNS11643_5_ann = 5 << 13, |
122 | CNS11643_6_ann = 6 << 13, |
123 | CNS11643_7_ann = 7 << 13, |
124 | SS3_ann = 7 << 13 |
125 | }; |
126 | |
127 | |
128 | /* Since this is a stateful encoding we have to provide code which resets |
129 | the output state to the initial state. This has to be done during the |
130 | flushing. */ |
131 | #define EMIT_SHIFT_TO_INIT \ |
132 | if (data->__statep->__count >> 3 != ASCII_set) \ |
133 | { \ |
134 | if (FROM_DIRECTION) \ |
135 | /* It's easy, we don't have to emit anything, we just reset the \ |
136 | state for the input. */ \ |
137 | data->__statep->__count = ASCII_set << 3; \ |
138 | else \ |
139 | { \ |
140 | /* We are not in the initial state. To switch back we have \ |
141 | to emit `SI'. */ \ |
142 | if (__glibc_unlikely (outbuf == outend)) \ |
143 | /* We don't have enough room in the output buffer. */ \ |
144 | status = __GCONV_FULL_OUTPUT; \ |
145 | else \ |
146 | { \ |
147 | /* Write out the shift sequence. */ \ |
148 | *outbuf++ = SI; \ |
149 | if (data->__flags & __GCONV_IS_LAST) \ |
150 | *irreversible += 1; \ |
151 | data->__statep->__count = ASCII_set << 3; \ |
152 | } \ |
153 | } \ |
154 | } |
155 | |
156 | |
157 | /* Since we might have to reset input pointer we must be able to save |
158 | and retore the state. */ |
159 | #define SAVE_RESET_STATE(Save) \ |
160 | if (Save) \ |
161 | save_set = *setp; \ |
162 | else \ |
163 | *setp = save_set |
164 | |
165 | |
166 | /* First define the conversion function from ISO-2022-CN to UCS4. */ |
167 | #define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM |
168 | #define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM |
169 | #define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO |
170 | #define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO |
171 | #define LOOPFCT FROM_LOOP |
172 | #define BODY \ |
173 | { \ |
174 | uint32_t ch = *inptr; \ |
175 | \ |
176 | /* This is a 7bit character set, disallow all 8bit characters. */ \ |
177 | if (ch > 0x7f) \ |
178 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
179 | \ |
180 | /* Recognize escape sequences. */ \ |
181 | if (ch == ESC) \ |
182 | { \ |
183 | /* There are three kinds of escape sequences we have to handle: \ |
184 | - those announcing the use of GB and CNS characters on the \ |
185 | line; we can simply ignore them \ |
186 | - the initial byte of the SS2 sequence. \ |
187 | - the initial byte of the SS3 sequence. \ |
188 | */ \ |
189 | if (inptr + 2 > inend \ |
190 | || (inptr[1] == '$' \ |
191 | && (inptr + 3 > inend \ |
192 | || (inptr[2] == ')' && inptr + 4 > inend) \ |
193 | || (inptr[2] == '*' && inptr + 4 > inend) \ |
194 | || (inptr[2] == '+' && inptr + 4 > inend))) \ |
195 | || (inptr[1] == SS2_1 && inptr + 4 > inend) \ |
196 | || (inptr[1] == SS3_1 && inptr + 4 > inend)) \ |
197 | { \ |
198 | result = __GCONV_INCOMPLETE_INPUT; \ |
199 | break; \ |
200 | } \ |
201 | if (inptr[1] == '$' \ |
202 | && ((inptr[2] == ')' \ |
203 | && (inptr[3] == 'A' \ |
204 | || (X12345 != '\0' && inptr[3] == X12345) \ |
205 | || inptr[3] == 'E' || inptr[3] == 'G')) \ |
206 | || (inptr[2] == '*' \ |
207 | && ((X7589 != '\0' && inptr[3] == X7589) \ |
208 | || (X13131 != '\0' && inptr[3] == X13131) \ |
209 | || inptr[3] == 'H')) \ |
210 | || (inptr[2] == '+' \ |
211 | && ((X7590 != '\0' && inptr[3] == X7590) \ |
212 | || (X13132 != '\0' && inptr[3] == X13132) \ |
213 | || inptr[3] == 'I' || inptr[3] == 'J' \ |
214 | || inptr[3] == 'K' || inptr[3] == 'L' \ |
215 | || inptr[3] == 'M')))) \ |
216 | { \ |
217 | /* OK, we accept those character sets. */ \ |
218 | if (inptr[3] == 'A') \ |
219 | ann = (ann & ~SO_ann) | GB2312_ann; \ |
220 | else if (inptr[3] == 'G') \ |
221 | ann = (ann & ~SO_ann) | CNS11643_1_ann; \ |
222 | else if (inptr[3] == 'E') \ |
223 | ann = (ann & ~SO_ann) | ISO_IR_165_ann; \ |
224 | else if (X12345 != '\0' && inptr[3] == X12345) \ |
225 | ann = (ann & ~SO_ann) | GB12345_ann; \ |
226 | else if (inptr[3] == 'H') \ |
227 | ann = (ann & ~SS2_ann) | CNS11643_2_ann; \ |
228 | else if (X7589 != '\0' && inptr[3] == X7589) \ |
229 | ann = (ann & ~SS2_ann) | GB7589_ann; \ |
230 | else if (X13131 != '\0' && inptr[3] == X13131) \ |
231 | ann = (ann & ~SS2_ann) | GB13131_ann; \ |
232 | else if (inptr[3] == 'I') \ |
233 | ann = (ann & ~SS3_ann) | CNS11643_3_ann; \ |
234 | else if (inptr[3] == 'J') \ |
235 | ann = (ann & ~SS3_ann) | CNS11643_4_ann; \ |
236 | else if (inptr[3] == 'K') \ |
237 | ann = (ann & ~SS3_ann) | CNS11643_5_ann; \ |
238 | else if (inptr[3] == 'L') \ |
239 | ann = (ann & ~SS3_ann) | CNS11643_6_ann; \ |
240 | else if (inptr[3] == 'M') \ |
241 | ann = (ann & ~SS3_ann) | CNS11643_7_ann; \ |
242 | else if (X7590 != '\0' && inptr[3] == X7590) \ |
243 | ann = (ann & ~SS3_ann) | GB7590_ann; \ |
244 | else if (X13132 != '\0' && inptr[3] == X13132) \ |
245 | ann = (ann & ~SS3_ann) | GB13132_ann; \ |
246 | inptr += 4; \ |
247 | continue; \ |
248 | } \ |
249 | } \ |
250 | else if (ch == SO) \ |
251 | { \ |
252 | /* Switch to use GB2312, GB12345, CNS 11643 plane 1, or ISO-IR-165, \ |
253 | depending on which S0 designation came last. The only problem \ |
254 | is what to do with faulty input files where no designator came. \ |
255 | XXX For now I'll default to use GB2312. If this is not the \ |
256 | best behavior (e.g., we should flag an error) let me know. */ \ |
257 | ++inptr; \ |
258 | if ((ann & SO_ann) != 0) \ |
259 | switch (ann & SO_ann) \ |
260 | { \ |
261 | case GB2312_ann: \ |
262 | set = GB2312_set; \ |
263 | break; \ |
264 | case GB12345_ann: \ |
265 | set = GB12345_set; \ |
266 | break; \ |
267 | case CNS11643_1_ann: \ |
268 | set = CNS11643_1_set; \ |
269 | break; \ |
270 | case ISO_IR_165_ann: \ |
271 | set = ISO_IR_165_set; \ |
272 | break; \ |
273 | default: \ |
274 | abort (); \ |
275 | } \ |
276 | else \ |
277 | { \ |
278 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
279 | } \ |
280 | continue; \ |
281 | } \ |
282 | else if (ch == SI) \ |
283 | { \ |
284 | /* Switch to use ASCII. */ \ |
285 | ++inptr; \ |
286 | set = ASCII_set; \ |
287 | continue; \ |
288 | } \ |
289 | \ |
290 | if (ch == ESC && inptr[1] == SS2_1) \ |
291 | { \ |
292 | /* This is a character from CNS 11643 plane 2. \ |
293 | XXX We could test here whether the use of this character \ |
294 | set was announced. \ |
295 | XXX Currently GB7589 and GB13131 are not supported. */ \ |
296 | inptr += 2; \ |
297 | ch = cns11643l2_to_ucs4 (&inptr, 2, 0); \ |
298 | if (ch == __UNKNOWN_10646_CHAR) \ |
299 | STANDARD_FROM_LOOP_ERR_HANDLER (2); \ |
300 | } \ |
301 | /* Note that we can assume here that at least 4 bytes are available if \ |
302 | the first byte is ESC since otherwise the first if would have been \ |
303 | true. */ \ |
304 | else if (ch == ESC && inptr[1] == SS3_1) \ |
305 | { \ |
306 | /* This is a character from CNS 11643 plane 3 or higher. \ |
307 | XXX Currently GB7590 and GB13132 are not supported. */ \ |
308 | unsigned char buf[3]; \ |
309 | const unsigned char *tmp = buf; \ |
310 | \ |
311 | buf[1] = inptr[2]; \ |
312 | buf[2] = inptr[3]; \ |
313 | switch (ann & SS3_ann) \ |
314 | { \ |
315 | case CNS11643_3_ann: \ |
316 | buf[0] = 0x23; \ |
317 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
318 | break; \ |
319 | case CNS11643_4_ann: \ |
320 | buf[0] = 0x24; \ |
321 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
322 | break; \ |
323 | case CNS11643_5_ann: \ |
324 | buf[0] = 0x25; \ |
325 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
326 | break; \ |
327 | case CNS11643_6_ann: \ |
328 | buf[0] = 0x26; \ |
329 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
330 | break; \ |
331 | case CNS11643_7_ann: \ |
332 | buf[0] = 0x27; \ |
333 | ch = cns11643_to_ucs4 (&tmp, 3, 0); \ |
334 | break; \ |
335 | default: \ |
336 | /* XXX Currently GB7590 and GB13132 are not supported. */ \ |
337 | ch = __UNKNOWN_10646_CHAR; \ |
338 | break; \ |
339 | } \ |
340 | if (ch == __UNKNOWN_10646_CHAR) \ |
341 | { \ |
342 | STANDARD_FROM_LOOP_ERR_HANDLER (4); \ |
343 | } \ |
344 | assert (tmp == buf + 3); \ |
345 | inptr += 4; \ |
346 | } \ |
347 | else if (set == ASCII_set) \ |
348 | { \ |
349 | /* Almost done, just advance the input pointer. */ \ |
350 | ++inptr; \ |
351 | } \ |
352 | else \ |
353 | { \ |
354 | /* That's pretty easy, we have a dedicated functions for this. */ \ |
355 | if (inend - inptr < 2) \ |
356 | { \ |
357 | result = __GCONV_INCOMPLETE_INPUT; \ |
358 | break; \ |
359 | } \ |
360 | if (set == GB2312_set) \ |
361 | ch = gb2312_to_ucs4 (&inptr, inend - inptr, 0); \ |
362 | else if (set == ISO_IR_165_set) \ |
363 | ch = isoir165_to_ucs4 (&inptr, inend - inptr); \ |
364 | else \ |
365 | { \ |
366 | assert (set == CNS11643_1_set); \ |
367 | ch = cns11643l1_to_ucs4 (&inptr, inend - inptr, 0); \ |
368 | } \ |
369 | \ |
370 | if (ch == 0) \ |
371 | { \ |
372 | result = __GCONV_INCOMPLETE_INPUT; \ |
373 | break; \ |
374 | } \ |
375 | else if (ch == __UNKNOWN_10646_CHAR) \ |
376 | { \ |
377 | STANDARD_FROM_LOOP_ERR_HANDLER (2); \ |
378 | } \ |
379 | } \ |
380 | \ |
381 | *((uint32_t *) outptr) = ch; \ |
382 | outptr += sizeof (uint32_t); \ |
383 | } |
384 | #define , int *setp |
385 | #define INIT_PARAMS int set = (*setp >> 3) & CURRENT_MASK; \ |
386 | int ann = (*setp >> 3) & ~CURRENT_MASK |
387 | #define UPDATE_PARAMS *setp = (set | ann) << 3 |
388 | #define LOOP_NEED_FLAGS |
389 | #include <iconv/loop.c> |
390 | |
391 | |
392 | /* Next, define the other direction. */ |
393 | #define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM |
394 | #define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM |
395 | #define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO |
396 | #define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO |
397 | #define LOOPFCT TO_LOOP |
398 | /* With GCC 5.3 when compiling with -Os the compiler emits a warning |
399 | that buf[0] and buf[1] may be used uninitialized. This can only |
400 | happen in the case where tmpbuf[3] is used, and in that case the |
401 | write to the tmpbuf[1] and tmpbuf[2] was assured because |
402 | ucs4_to_cns11643 would have filled in those entries. The difficulty |
403 | is in getting the compiler to see this logic because tmpbuf[0] is |
404 | involved in determining the code page and is the indicator that |
405 | tmpbuf[2] is initialized. */ |
406 | DIAG_PUSH_NEEDS_COMMENT; |
407 | DIAG_IGNORE_Os_NEEDS_COMMENT (5, "-Wmaybe-uninitialized" ); |
408 | #define BODY \ |
409 | { \ |
410 | uint32_t ch; \ |
411 | size_t written = 0; \ |
412 | \ |
413 | ch = *((const uint32_t *) inptr); \ |
414 | \ |
415 | /* First see whether we can write the character using the currently \ |
416 | selected character set. */ \ |
417 | if (ch < 0x80) \ |
418 | { \ |
419 | if (set != ASCII_set) \ |
420 | { \ |
421 | *outptr++ = SI; \ |
422 | set = ASCII_set; \ |
423 | if (outptr == outend) \ |
424 | { \ |
425 | result = __GCONV_FULL_OUTPUT; \ |
426 | break; \ |
427 | } \ |
428 | } \ |
429 | \ |
430 | *outptr++ = ch; \ |
431 | written = 1; \ |
432 | \ |
433 | /* At the end of the line we have to clear the `ann' flags since \ |
434 | every line must contain this information again. */ \ |
435 | if (ch == L'\n') \ |
436 | ann = 0; \ |
437 | } \ |
438 | else \ |
439 | { \ |
440 | unsigned char buf[2] = { 0, 0 }; \ |
441 | int used; \ |
442 | \ |
443 | if (set == GB2312_set || ((ann & SO_ann) != CNS11643_1_ann \ |
444 | && (ann & SO_ann) != ISO_IR_165_ann)) \ |
445 | { \ |
446 | written = ucs4_to_gb2312 (ch, buf, 2); \ |
447 | used = GB2312_set; \ |
448 | } \ |
449 | else if (set == ISO_IR_165_set || (ann & SO_ann) == ISO_IR_165_set) \ |
450 | { \ |
451 | written = ucs4_to_isoir165 (ch, buf, 2); \ |
452 | used = ISO_IR_165_set; \ |
453 | } \ |
454 | else \ |
455 | { \ |
456 | written = ucs4_to_cns11643l1 (ch, buf, 2); \ |
457 | used = CNS11643_1_set; \ |
458 | } \ |
459 | \ |
460 | if (written == __UNKNOWN_10646_CHAR) \ |
461 | { \ |
462 | /* Cannot convert it using the currently selected SO set. \ |
463 | Next try the SS2 set. */ \ |
464 | written = ucs4_to_cns11643l2 (ch, buf, 2); \ |
465 | if (written != __UNKNOWN_10646_CHAR) \ |
466 | /* Yep, that worked. */ \ |
467 | used = CNS11643_2_set; \ |
468 | else \ |
469 | { \ |
470 | unsigned char tmpbuf[3]; \ |
471 | \ |
472 | switch (0) \ |
473 | { \ |
474 | default: \ |
475 | /* Well, see whether we have to change the SO set. */ \ |
476 | \ |
477 | if (used != GB2312_set) \ |
478 | { \ |
479 | written = ucs4_to_gb2312 (ch, buf, 2); \ |
480 | if (written != __UNKNOWN_10646_CHAR) \ |
481 | { \ |
482 | used = GB2312_set; \ |
483 | break; \ |
484 | } \ |
485 | } \ |
486 | \ |
487 | if (used != ISO_IR_165_set) \ |
488 | { \ |
489 | written = ucs4_to_isoir165 (ch, buf, 2); \ |
490 | if (written != __UNKNOWN_10646_CHAR) \ |
491 | { \ |
492 | used = ISO_IR_165_set; \ |
493 | break; \ |
494 | } \ |
495 | } \ |
496 | \ |
497 | if (used != CNS11643_1_set) \ |
498 | { \ |
499 | written = ucs4_to_cns11643l1 (ch, buf, 2); \ |
500 | if (written != __UNKNOWN_10646_CHAR) \ |
501 | { \ |
502 | used = CNS11643_1_set; \ |
503 | break; \ |
504 | } \ |
505 | } \ |
506 | \ |
507 | written = ucs4_to_cns11643 (ch, tmpbuf, 3); \ |
508 | if (written == 3 && tmpbuf[0] >= 3 && tmpbuf[0] <= 7) \ |
509 | { \ |
510 | buf[0] = tmpbuf[1]; \ |
511 | buf[1] = tmpbuf[2]; \ |
512 | switch (tmpbuf[0]) \ |
513 | { \ |
514 | case 3: \ |
515 | used = CNS11643_3_set; \ |
516 | break; \ |
517 | case 4: \ |
518 | used = CNS11643_4_set; \ |
519 | break; \ |
520 | case 5: \ |
521 | used = CNS11643_5_set; \ |
522 | break; \ |
523 | case 6: \ |
524 | used = CNS11643_6_set; \ |
525 | break; \ |
526 | case 7: \ |
527 | used = CNS11643_7_set; \ |
528 | break; \ |
529 | default: \ |
530 | abort (); \ |
531 | } \ |
532 | written = 2; \ |
533 | break; \ |
534 | } \ |
535 | \ |
536 | /* XXX Currently GB7590 and GB13132 are not supported. */\ |
537 | \ |
538 | /* Even this does not work. Error. */ \ |
539 | used = ASCII_set; \ |
540 | } \ |
541 | if (used == ASCII_set) \ |
542 | { \ |
543 | UNICODE_TAG_HANDLER (ch, 4); \ |
544 | STANDARD_TO_LOOP_ERR_HANDLER (4); \ |
545 | } \ |
546 | } \ |
547 | } \ |
548 | assert (written == 2); \ |
549 | \ |
550 | /* See whether we have to emit an escape sequence. */ \ |
551 | if (set != used) \ |
552 | { \ |
553 | /* First see whether we announced that we use this \ |
554 | character set. */ \ |
555 | if ((used & SO_mask) != 0 && (ann & SO_ann) != (used << 8)) \ |
556 | { \ |
557 | const char *escseq; \ |
558 | \ |
559 | if (outptr + 4 > outend) \ |
560 | { \ |
561 | result = __GCONV_FULL_OUTPUT; \ |
562 | break; \ |
563 | } \ |
564 | \ |
565 | assert (used >= 1 && used <= 4); \ |
566 | escseq = ")A\0\0)G)E" + (used - 1) * 2; \ |
567 | *outptr++ = ESC; \ |
568 | *outptr++ = '$'; \ |
569 | *outptr++ = *escseq++; \ |
570 | *outptr++ = *escseq++; \ |
571 | \ |
572 | ann = (ann & ~SO_ann) | (used << 8); \ |
573 | } \ |
574 | else if ((used & SS2_mask) != 0 && (ann & SS2_ann) != (used << 8))\ |
575 | { \ |
576 | const char *escseq; \ |
577 | \ |
578 | assert (used == CNS11643_2_set); /* XXX */ \ |
579 | escseq = "*H"; \ |
580 | *outptr++ = ESC; \ |
581 | *outptr++ = '$'; \ |
582 | *outptr++ = *escseq++; \ |
583 | *outptr++ = *escseq++; \ |
584 | \ |
585 | ann = (ann & ~SS2_ann) | (used << 8); \ |
586 | } \ |
587 | else if ((used & SS3_mask) != 0 && (ann & SS3_ann) != (used << 8))\ |
588 | { \ |
589 | const char *escseq; \ |
590 | \ |
591 | assert ((used >> 5) >= 3 && (used >> 5) <= 7); \ |
592 | escseq = "+I+J+K+L+M" + ((used >> 5) - 3) * 2; \ |
593 | *outptr++ = ESC; \ |
594 | *outptr++ = '$'; \ |
595 | *outptr++ = *escseq++; \ |
596 | *outptr++ = *escseq++; \ |
597 | \ |
598 | ann = (ann & ~SS3_ann) | (used << 8); \ |
599 | } \ |
600 | \ |
601 | if (used == CNS11643_2_set) \ |
602 | { \ |
603 | if (outptr + 2 > outend) \ |
604 | { \ |
605 | result = __GCONV_FULL_OUTPUT; \ |
606 | break; \ |
607 | } \ |
608 | *outptr++ = SS2_0; \ |
609 | *outptr++ = SS2_1; \ |
610 | } \ |
611 | else if (used >= CNS11643_3_set && used <= CNS11643_7_set) \ |
612 | { \ |
613 | if (outptr + 2 > outend) \ |
614 | { \ |
615 | result = __GCONV_FULL_OUTPUT; \ |
616 | break; \ |
617 | } \ |
618 | *outptr++ = SS3_0; \ |
619 | *outptr++ = SS3_1; \ |
620 | } \ |
621 | else \ |
622 | { \ |
623 | /* We only have to emit something if currently ASCII is \ |
624 | selected. Otherwise we are switching within the \ |
625 | SO charset. */ \ |
626 | if (set == ASCII_set) \ |
627 | { \ |
628 | if (outptr + 1 > outend) \ |
629 | { \ |
630 | result = __GCONV_FULL_OUTPUT; \ |
631 | break; \ |
632 | } \ |
633 | *outptr++ = SO; \ |
634 | } \ |
635 | } \ |
636 | \ |
637 | /* Always test the length here since we have used up all the \ |
638 | guaranteed output buffer slots. */ \ |
639 | if (outptr + 2 > outend) \ |
640 | { \ |
641 | result = __GCONV_FULL_OUTPUT; \ |
642 | break; \ |
643 | } \ |
644 | } \ |
645 | else if (outptr + 2 > outend) \ |
646 | { \ |
647 | result = __GCONV_FULL_OUTPUT; \ |
648 | break; \ |
649 | } \ |
650 | \ |
651 | *outptr++ = buf[0]; \ |
652 | *outptr++ = buf[1]; \ |
653 | set = used; \ |
654 | } \ |
655 | \ |
656 | /* Now that we wrote the output increment the input pointer. */ \ |
657 | inptr += 4; \ |
658 | } |
659 | DIAG_POP_NEEDS_COMMENT; |
660 | #define , int *setp |
661 | #define INIT_PARAMS int set = (*setp >> 3) & CURRENT_MASK; \ |
662 | int ann = (*setp >> 3) & ~CURRENT_MASK |
663 | #define REINIT_PARAMS do \ |
664 | { \ |
665 | set = (*setp >> 3) & CURRENT_MASK; \ |
666 | ann = (*setp >> 3) & ~CURRENT_MASK; \ |
667 | } \ |
668 | while (0) |
669 | #define UPDATE_PARAMS *setp = (set | ann) << 3 |
670 | #define LOOP_NEED_FLAGS |
671 | #include <iconv/loop.c> |
672 | |
673 | |
674 | /* Now define the toplevel functions. */ |
675 | #include <iconv/skeleton.c> |
676 | |