]>
Commit | Line | Data |
---|---|---|
1 | /* Skeleton for a conversion module. | |
2 | Copyright (C) 1998-2024 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 | /* This file can be included to provide definitions of several things | |
20 | many modules have in common. It can be customized using the following | |
21 | macros: | |
22 | ||
23 | DEFINE_INIT define the default initializer. This requires the | |
24 | following symbol to be defined. | |
25 | ||
26 | CHARSET_NAME string with official name of the coded character | |
27 | set (in all-caps) | |
28 | ||
29 | DEFINE_FINI define the default destructor function. | |
30 | ||
31 | MIN_NEEDED_FROM minimal number of bytes needed for the from-charset. | |
32 | MIN_NEEDED_TO likewise for the to-charset. | |
33 | ||
34 | MAX_NEEDED_FROM maximal number of bytes needed for the from-charset. | |
35 | This macro is optional, it defaults to MIN_NEEDED_FROM. | |
36 | MAX_NEEDED_TO likewise for the to-charset. | |
37 | ||
38 | FROM_LOOP_MIN_NEEDED_FROM | |
39 | FROM_LOOP_MAX_NEEDED_FROM | |
40 | minimal/maximal number of bytes needed on input | |
41 | of one round through the FROM_LOOP. Defaults | |
42 | to MIN_NEEDED_FROM and MAX_NEEDED_FROM, respectively. | |
43 | FROM_LOOP_MIN_NEEDED_TO | |
44 | FROM_LOOP_MAX_NEEDED_TO | |
45 | minimal/maximal number of bytes needed on output | |
46 | of one round through the FROM_LOOP. Defaults | |
47 | to MIN_NEEDED_TO and MAX_NEEDED_TO, respectively. | |
48 | TO_LOOP_MIN_NEEDED_FROM | |
49 | TO_LOOP_MAX_NEEDED_FROM | |
50 | minimal/maximal number of bytes needed on input | |
51 | of one round through the TO_LOOP. Defaults | |
52 | to MIN_NEEDED_TO and MAX_NEEDED_TO, respectively. | |
53 | TO_LOOP_MIN_NEEDED_TO | |
54 | TO_LOOP_MAX_NEEDED_TO | |
55 | minimal/maximal number of bytes needed on output | |
56 | of one round through the TO_LOOP. Defaults | |
57 | to MIN_NEEDED_FROM and MAX_NEEDED_FROM, respectively. | |
58 | ||
59 | FROM_DIRECTION this macro is supposed to return a value != 0 | |
60 | if we convert from the current character set, | |
61 | otherwise it return 0. | |
62 | ||
63 | EMIT_SHIFT_TO_INIT this symbol is optional. If it is defined it | |
64 | defines some code which writes out a sequence | |
65 | of bytes which bring the current state into | |
66 | the initial state. | |
67 | ||
68 | FROM_LOOP name of the function implementing the conversion | |
69 | from the current character set. | |
70 | TO_LOOP likewise for the other direction | |
71 | ||
72 | ONE_DIRECTION optional. If defined to 1, only one conversion | |
73 | direction is defined instead of two. In this | |
74 | case, FROM_DIRECTION should be defined to 1, and | |
75 | FROM_LOOP and TO_LOOP should have the same value. | |
76 | ||
77 | SAVE_RESET_STATE in case of an error we must reset the state for | |
78 | the rerun so this macro must be defined for | |
79 | stateful encodings. It takes an argument which | |
80 | is nonzero when saving. | |
81 | ||
82 | RESET_INPUT_BUFFER If the input character sets allow this the macro | |
83 | can be defined to reset the input buffer pointers | |
84 | to cover only those characters up to the error. | |
85 | Note that if the conversion has skipped over | |
86 | irreversible characters (due to | |
87 | __GCONV_IGNORE_ERRORS) there is no longer a direct | |
88 | correspondence between input and output pointers, | |
89 | and this macro is not called. | |
90 | ||
91 | FUNCTION_NAME if not set the conversion function is named `gconv'. | |
92 | ||
93 | PREPARE_LOOP optional code preparing the conversion loop. Can | |
94 | contain variable definitions. | |
95 | END_LOOP also optional, may be used to store information | |
96 | ||
97 | EXTRA_LOOP_ARGS optional macro specifying extra arguments passed | |
98 | to loop function. | |
99 | ||
100 | STORE_REST optional, needed only when MAX_NEEDED_FROM > 4. | |
101 | This macro stores the seen but unconverted input bytes | |
102 | in the state. | |
103 | ||
104 | FROM_ONEBYTE optional. If defined, should be the name of a | |
105 | specialized conversion function for a single byte | |
106 | from the current character set to INTERNAL. This | |
107 | function has prototype | |
108 | wint_t | |
109 | FROM_ONEBYTE (struct __gconv_step *, unsigned char); | |
110 | and does a special conversion: | |
111 | - The input is a single byte. | |
112 | - The output is a single uint32_t. | |
113 | - The state before the conversion is the initial state; | |
114 | the state after the conversion is irrelevant. | |
115 | - No transliteration. | |
116 | - __invocation_counter = 0. | |
117 | - __internal_use = 1. | |
118 | - do_flush = 0. | |
119 | ||
120 | Modules can use mbstate_t to store conversion state as follows: | |
121 | ||
122 | * Bits 2..0 of '__count' contain the number of lookahead input bytes | |
123 | stored in __value.__wchb. Always zero if the converter never | |
124 | returns __GCONV_INCOMPLETE_INPUT. | |
125 | ||
126 | * Bits 31..3 of '__count' are module dependent shift state. | |
127 | ||
128 | * __value: When STORE_REST/UNPACK_BYTES aren't defined and when the | |
129 | converter has returned __GCONV_INCOMPLETE_INPUT, this contains | |
130 | at most 4 lookahead bytes. Converters with an mb_cur_max > 4 | |
131 | (currently only UTF-8) must find a way to store their state | |
132 | in __value.__wch and define STORE_REST/UNPACK_BYTES appropriately. | |
133 | ||
134 | When __value contains lookahead, __count must not be zero, because | |
135 | the converter is not in the initial state then, and mbsinit() -- | |
136 | defined as a (__count == 0) test -- must reflect this. | |
137 | */ | |
138 | ||
139 | #include <assert.h> | |
140 | #include <iconv/gconv_int.h> | |
141 | #include <string.h> | |
142 | #define __need_size_t | |
143 | #define __need_NULL | |
144 | #include <stddef.h> | |
145 | ||
146 | #ifndef STATIC_GCONV | |
147 | # include <dlfcn.h> | |
148 | #endif | |
149 | ||
150 | #include <pointer_guard.h> | |
151 | #include <stdint.h> | |
152 | ||
153 | #ifndef DL_CALL_FCT | |
154 | # define DL_CALL_FCT(fct, args) fct args | |
155 | #endif | |
156 | ||
157 | /* The direction objects. */ | |
158 | #if DEFINE_INIT | |
159 | # ifndef FROM_DIRECTION | |
160 | # define FROM_DIRECTION_VAL NULL | |
161 | # define TO_DIRECTION_VAL ((void *) ~((uintptr_t) 0)) | |
162 | # define FROM_DIRECTION (step->__data == FROM_DIRECTION_VAL) | |
163 | # endif | |
164 | #else | |
165 | # ifndef FROM_DIRECTION | |
166 | # error "FROM_DIRECTION must be provided if non-default init is used" | |
167 | # endif | |
168 | #endif | |
169 | ||
170 | /* How many bytes are needed at most for the from-charset. */ | |
171 | #ifndef MAX_NEEDED_FROM | |
172 | # define MAX_NEEDED_FROM MIN_NEEDED_FROM | |
173 | #endif | |
174 | ||
175 | /* Same for the to-charset. */ | |
176 | #ifndef MAX_NEEDED_TO | |
177 | # define MAX_NEEDED_TO MIN_NEEDED_TO | |
178 | #endif | |
179 | ||
180 | /* Defaults for the per-direction min/max constants. */ | |
181 | #ifndef FROM_LOOP_MIN_NEEDED_FROM | |
182 | # define FROM_LOOP_MIN_NEEDED_FROM MIN_NEEDED_FROM | |
183 | #endif | |
184 | #ifndef FROM_LOOP_MAX_NEEDED_FROM | |
185 | # define FROM_LOOP_MAX_NEEDED_FROM MAX_NEEDED_FROM | |
186 | #endif | |
187 | #ifndef FROM_LOOP_MIN_NEEDED_TO | |
188 | # define FROM_LOOP_MIN_NEEDED_TO MIN_NEEDED_TO | |
189 | #endif | |
190 | #ifndef FROM_LOOP_MAX_NEEDED_TO | |
191 | # define FROM_LOOP_MAX_NEEDED_TO MAX_NEEDED_TO | |
192 | #endif | |
193 | #ifndef TO_LOOP_MIN_NEEDED_FROM | |
194 | # define TO_LOOP_MIN_NEEDED_FROM MIN_NEEDED_TO | |
195 | #endif | |
196 | #ifndef TO_LOOP_MAX_NEEDED_FROM | |
197 | # define TO_LOOP_MAX_NEEDED_FROM MAX_NEEDED_TO | |
198 | #endif | |
199 | #ifndef TO_LOOP_MIN_NEEDED_TO | |
200 | # define TO_LOOP_MIN_NEEDED_TO MIN_NEEDED_FROM | |
201 | #endif | |
202 | #ifndef TO_LOOP_MAX_NEEDED_TO | |
203 | # define TO_LOOP_MAX_NEEDED_TO MAX_NEEDED_FROM | |
204 | #endif | |
205 | ||
206 | ||
207 | /* For conversions from a fixed width character set to another fixed width | |
208 | character set we can define RESET_INPUT_BUFFER in a very fast way. */ | |
209 | #if !defined RESET_INPUT_BUFFER && !defined SAVE_RESET_STATE | |
210 | # if FROM_LOOP_MIN_NEEDED_FROM == FROM_LOOP_MAX_NEEDED_FROM \ | |
211 | && FROM_LOOP_MIN_NEEDED_TO == FROM_LOOP_MAX_NEEDED_TO \ | |
212 | && TO_LOOP_MIN_NEEDED_FROM == TO_LOOP_MAX_NEEDED_FROM \ | |
213 | && TO_LOOP_MIN_NEEDED_TO == TO_LOOP_MAX_NEEDED_TO | |
214 | /* We have to use these `if's here since the compiler cannot know that | |
215 | (outbuf - outerr) is always divisible by FROM/TO_LOOP_MIN_NEEDED_TO. | |
216 | The ?:1 avoids division by zero warnings that gcc 3.2 emits even for | |
217 | obviously unreachable code. */ | |
218 | # define RESET_INPUT_BUFFER \ | |
219 | if (FROM_DIRECTION) \ | |
220 | { \ | |
221 | if (FROM_LOOP_MIN_NEEDED_FROM % FROM_LOOP_MIN_NEEDED_TO == 0) \ | |
222 | *inptrp -= (outbuf - outerr) \ | |
223 | * (FROM_LOOP_MIN_NEEDED_FROM / FROM_LOOP_MIN_NEEDED_TO); \ | |
224 | else if (FROM_LOOP_MIN_NEEDED_TO % FROM_LOOP_MIN_NEEDED_FROM == 0) \ | |
225 | *inptrp -= (outbuf - outerr) \ | |
226 | / (FROM_LOOP_MIN_NEEDED_TO / FROM_LOOP_MIN_NEEDED_FROM \ | |
227 | ? : 1); \ | |
228 | else \ | |
229 | *inptrp -= ((outbuf - outerr) / FROM_LOOP_MIN_NEEDED_TO) \ | |
230 | * FROM_LOOP_MIN_NEEDED_FROM; \ | |
231 | } \ | |
232 | else \ | |
233 | { \ | |
234 | if (TO_LOOP_MIN_NEEDED_FROM % TO_LOOP_MIN_NEEDED_TO == 0) \ | |
235 | *inptrp -= (outbuf - outerr) \ | |
236 | * (TO_LOOP_MIN_NEEDED_FROM / TO_LOOP_MIN_NEEDED_TO); \ | |
237 | else if (TO_LOOP_MIN_NEEDED_TO % TO_LOOP_MIN_NEEDED_FROM == 0) \ | |
238 | *inptrp -= (outbuf - outerr) \ | |
239 | / (TO_LOOP_MIN_NEEDED_TO / TO_LOOP_MIN_NEEDED_FROM ? : 1); \ | |
240 | else \ | |
241 | *inptrp -= ((outbuf - outerr) / TO_LOOP_MIN_NEEDED_TO) \ | |
242 | * TO_LOOP_MIN_NEEDED_FROM; \ | |
243 | } | |
244 | # endif | |
245 | #endif | |
246 | ||
247 | ||
248 | /* The default init function. It simply matches the name and initializes | |
249 | the step data to point to one of the objects above. */ | |
250 | #if DEFINE_INIT | |
251 | # ifndef CHARSET_NAME | |
252 | # error "CHARSET_NAME not defined" | |
253 | # endif | |
254 | ||
255 | extern int gconv_init (struct __gconv_step *step); | |
256 | int | |
257 | gconv_init (struct __gconv_step *step) | |
258 | { | |
259 | /* Determine which direction. */ | |
260 | if (strcmp (step->__from_name, CHARSET_NAME) == 0) | |
261 | { | |
262 | step->__data = FROM_DIRECTION_VAL; | |
263 | ||
264 | step->__min_needed_from = FROM_LOOP_MIN_NEEDED_FROM; | |
265 | step->__max_needed_from = FROM_LOOP_MAX_NEEDED_FROM; | |
266 | step->__min_needed_to = FROM_LOOP_MIN_NEEDED_TO; | |
267 | step->__max_needed_to = FROM_LOOP_MAX_NEEDED_TO; | |
268 | ||
269 | #ifdef FROM_ONEBYTE | |
270 | step->__btowc_fct = FROM_ONEBYTE; | |
271 | #endif | |
272 | } | |
273 | else if (__builtin_expect (strcmp (step->__to_name, CHARSET_NAME), 0) == 0) | |
274 | { | |
275 | step->__data = TO_DIRECTION_VAL; | |
276 | ||
277 | step->__min_needed_from = TO_LOOP_MIN_NEEDED_FROM; | |
278 | step->__max_needed_from = TO_LOOP_MAX_NEEDED_FROM; | |
279 | step->__min_needed_to = TO_LOOP_MIN_NEEDED_TO; | |
280 | step->__max_needed_to = TO_LOOP_MAX_NEEDED_TO; | |
281 | } | |
282 | else | |
283 | return __GCONV_NOCONV; | |
284 | ||
285 | #ifdef SAVE_RESET_STATE | |
286 | step->__stateful = 1; | |
287 | #else | |
288 | step->__stateful = 0; | |
289 | #endif | |
290 | ||
291 | return __GCONV_OK; | |
292 | } | |
293 | #endif | |
294 | ||
295 | ||
296 | /* The default destructor function does nothing in the moment and so | |
297 | we don't define it at all. But we still provide the macro just in | |
298 | case we need it some day. */ | |
299 | #if DEFINE_FINI | |
300 | #endif | |
301 | ||
302 | ||
303 | /* If no arguments have to passed to the loop function define the macro | |
304 | as empty. */ | |
305 | #ifndef EXTRA_LOOP_ARGS | |
306 | # define EXTRA_LOOP_ARGS | |
307 | #endif | |
308 | ||
309 | ||
310 | /* This is the actual conversion function. */ | |
311 | #ifndef FUNCTION_NAME | |
312 | # define FUNCTION_NAME gconv | |
313 | #endif | |
314 | ||
315 | /* The macros are used to access the function to convert single characters. */ | |
316 | #define SINGLE(fct) SINGLE2 (fct) | |
317 | #define SINGLE2(fct) fct##_single | |
318 | ||
319 | ||
320 | extern int FUNCTION_NAME (struct __gconv_step *step, | |
321 | struct __gconv_step_data *data, | |
322 | const unsigned char **inptrp, | |
323 | const unsigned char *inend, | |
324 | unsigned char **outbufstart, size_t *irreversible, | |
325 | int do_flush, int consume_incomplete); | |
326 | int | |
327 | FUNCTION_NAME (struct __gconv_step *step, struct __gconv_step_data *data, | |
328 | const unsigned char **inptrp, const unsigned char *inend, | |
329 | unsigned char **outbufstart, size_t *irreversible, int do_flush, | |
330 | int consume_incomplete) | |
331 | { | |
332 | struct __gconv_step *next_step = step + 1; | |
333 | struct __gconv_step_data *next_data = data + 1; | |
334 | __gconv_fct fct = NULL; | |
335 | int status; | |
336 | ||
337 | if ((data->__flags & __GCONV_IS_LAST) == 0) | |
338 | { | |
339 | fct = next_step->__fct; | |
340 | if (next_step->__shlib_handle != NULL) | |
341 | PTR_DEMANGLE (fct); | |
342 | } | |
343 | ||
344 | /* If the function is called with no input this means we have to reset | |
345 | to the initial state. The possibly partly converted input is | |
346 | dropped. */ | |
347 | if (__glibc_unlikely (do_flush)) | |
348 | { | |
349 | /* This should never happen during error handling. */ | |
350 | assert (outbufstart == NULL); | |
351 | ||
352 | status = __GCONV_OK; | |
353 | ||
354 | #ifdef EMIT_SHIFT_TO_INIT | |
355 | if (do_flush == 1) | |
356 | { | |
357 | /* We preserve the initial values of the pointer variables. */ | |
358 | unsigned char *outbuf = data->__outbuf; | |
359 | unsigned char *outstart = outbuf; | |
360 | unsigned char *outend = data->__outbufend; | |
361 | ||
362 | # ifdef PREPARE_LOOP | |
363 | PREPARE_LOOP | |
364 | # endif | |
365 | ||
366 | # ifdef SAVE_RESET_STATE | |
367 | SAVE_RESET_STATE (1); | |
368 | # endif | |
369 | ||
370 | /* Emit the escape sequence to reset the state. */ | |
371 | EMIT_SHIFT_TO_INIT; | |
372 | ||
373 | /* Call the steps down the chain if there are any but only if we | |
374 | successfully emitted the escape sequence. This should only | |
375 | fail if the output buffer is full. If the input is invalid | |
376 | it should be discarded since the user wants to start from a | |
377 | clean state. */ | |
378 | if (status == __GCONV_OK) | |
379 | { | |
380 | if (data->__flags & __GCONV_IS_LAST) | |
381 | /* Store information about how many bytes are available. */ | |
382 | data->__outbuf = outbuf; | |
383 | else | |
384 | { | |
385 | /* Write out all output which was produced. */ | |
386 | if (outbuf > outstart) | |
387 | { | |
388 | const unsigned char *outerr = outstart; | |
389 | int result; | |
390 | ||
391 | result = DL_CALL_FCT (fct, (next_step, next_data, | |
392 | &outerr, outbuf, NULL, | |
393 | irreversible, 0, | |
394 | consume_incomplete)); | |
395 | ||
396 | if (result != __GCONV_EMPTY_INPUT) | |
397 | { | |
398 | if (__glibc_unlikely (outerr != outbuf)) | |
399 | { | |
400 | /* We have a problem. Undo the conversion. */ | |
401 | outbuf = outstart; | |
402 | ||
403 | /* Restore the state. */ | |
404 | # ifdef SAVE_RESET_STATE | |
405 | SAVE_RESET_STATE (0); | |
406 | # endif | |
407 | } | |
408 | ||
409 | /* Change the status. */ | |
410 | status = result; | |
411 | } | |
412 | } | |
413 | ||
414 | if (status == __GCONV_OK) | |
415 | /* Now flush the remaining steps. */ | |
416 | status = DL_CALL_FCT (fct, (next_step, next_data, NULL, | |
417 | NULL, NULL, irreversible, 1, | |
418 | consume_incomplete)); | |
419 | } | |
420 | } | |
421 | } | |
422 | else | |
423 | #endif | |
424 | { | |
425 | /* Clear the state object. There might be bytes in there from | |
426 | previous calls with CONSUME_INCOMPLETE == 1. But don't emit | |
427 | escape sequences. */ | |
428 | memset (data->__statep, '\0', sizeof (*data->__statep)); | |
429 | ||
430 | if (! (data->__flags & __GCONV_IS_LAST)) | |
431 | /* Now flush the remaining steps. */ | |
432 | status = DL_CALL_FCT (fct, (next_step, next_data, NULL, NULL, | |
433 | NULL, irreversible, do_flush, | |
434 | consume_incomplete)); | |
435 | } | |
436 | } | |
437 | else | |
438 | { | |
439 | /* We preserve the initial values of the pointer variables, | |
440 | but only some conversion modules need it. */ | |
441 | const unsigned char *inptr __attribute__ ((__unused__)) = *inptrp; | |
442 | unsigned char *outbuf = (__builtin_expect (outbufstart == NULL, 1) | |
443 | ? data->__outbuf : *outbufstart); | |
444 | unsigned char *outend = data->__outbufend; | |
445 | unsigned char *outstart; | |
446 | /* This variable is used to count the number of characters we | |
447 | actually converted. */ | |
448 | size_t lirreversible = 0; | |
449 | size_t *lirreversiblep = irreversible ? &lirreversible : NULL; | |
450 | ||
451 | #ifdef PREPARE_LOOP | |
452 | PREPARE_LOOP | |
453 | #endif | |
454 | ||
455 | #if FROM_LOOP_MAX_NEEDED_FROM > 1 || TO_LOOP_MAX_NEEDED_FROM > 1 | |
456 | /* If the function is used to implement the mb*towc*() or wc*tomb*() | |
457 | functions we must test whether any bytes from the last call are | |
458 | stored in the `state' object. */ | |
459 | if (((FROM_LOOP_MAX_NEEDED_FROM > 1 && TO_LOOP_MAX_NEEDED_FROM > 1) | |
460 | || (FROM_LOOP_MAX_NEEDED_FROM > 1 && FROM_DIRECTION) | |
461 | || (TO_LOOP_MAX_NEEDED_FROM > 1 && !FROM_DIRECTION)) | |
462 | && consume_incomplete && (data->__statep->__count & 7) != 0) | |
463 | { | |
464 | /* Yep, we have some bytes left over. Process them now. | |
465 | But this must not happen while we are called from an | |
466 | error handler. */ | |
467 | assert (outbufstart == NULL); | |
468 | ||
469 | # if FROM_LOOP_MAX_NEEDED_FROM > 1 | |
470 | if (TO_LOOP_MAX_NEEDED_FROM == 1 || FROM_DIRECTION) | |
471 | status = SINGLE(FROM_LOOP) (step, data, inptrp, inend, &outbuf, | |
472 | outend, lirreversiblep | |
473 | EXTRA_LOOP_ARGS); | |
474 | # endif | |
475 | # if !ONE_DIRECTION | |
476 | # if FROM_LOOP_MAX_NEEDED_FROM > 1 && TO_LOOP_MAX_NEEDED_FROM > 1 | |
477 | else | |
478 | # endif | |
479 | # if TO_LOOP_MAX_NEEDED_FROM > 1 | |
480 | status = SINGLE(TO_LOOP) (step, data, inptrp, inend, &outbuf, | |
481 | outend, lirreversiblep EXTRA_LOOP_ARGS); | |
482 | # endif | |
483 | # endif | |
484 | ||
485 | if (__builtin_expect (status, __GCONV_OK) != __GCONV_OK) | |
486 | return status; | |
487 | } | |
488 | #endif | |
489 | ||
490 | while (1) | |
491 | { | |
492 | /* Remember the start value for this round. */ | |
493 | inptr = *inptrp; | |
494 | /* The outbuf buffer is empty. */ | |
495 | outstart = outbuf; | |
496 | #ifdef RESET_INPUT_BUFFER | |
497 | /* Remember how many irreversible characters were skipped before | |
498 | this round. */ | |
499 | size_t loop_irreversible | |
500 | = lirreversible + (irreversible ? *irreversible : 0); | |
501 | #endif | |
502 | ||
503 | #ifdef SAVE_RESET_STATE | |
504 | SAVE_RESET_STATE (1); | |
505 | #endif | |
506 | ||
507 | if (FROM_DIRECTION) | |
508 | /* Run the conversion loop. */ | |
509 | status = FROM_LOOP (step, data, inptrp, inend, &outbuf, outend, | |
510 | lirreversiblep EXTRA_LOOP_ARGS); | |
511 | else | |
512 | /* Run the conversion loop. */ | |
513 | status = TO_LOOP (step, data, inptrp, inend, &outbuf, outend, | |
514 | lirreversiblep EXTRA_LOOP_ARGS); | |
515 | ||
516 | /* If we were called as part of an error handling module we | |
517 | don't do anything else here. */ | |
518 | if (__glibc_unlikely (outbufstart != NULL)) | |
519 | { | |
520 | *outbufstart = outbuf; | |
521 | return status; | |
522 | } | |
523 | ||
524 | /* We finished one use of the loops. */ | |
525 | ++data->__invocation_counter; | |
526 | ||
527 | /* If this is the last step leave the loop, there is nothing | |
528 | we can do. */ | |
529 | if (__glibc_unlikely (data->__flags & __GCONV_IS_LAST)) | |
530 | { | |
531 | /* Store information about how many bytes are available. */ | |
532 | data->__outbuf = outbuf; | |
533 | ||
534 | /* Remember how many non-identical characters we | |
535 | converted in an irreversible way. */ | |
536 | *irreversible += lirreversible; | |
537 | ||
538 | break; | |
539 | } | |
540 | ||
541 | /* Write out all output which was produced. */ | |
542 | if (__glibc_likely (outbuf > outstart)) | |
543 | { | |
544 | const unsigned char *outerr = data->__outbuf; | |
545 | int result; | |
546 | ||
547 | result = DL_CALL_FCT (fct, (next_step, next_data, &outerr, | |
548 | outbuf, NULL, irreversible, 0, | |
549 | consume_incomplete)); | |
550 | ||
551 | if (result != __GCONV_EMPTY_INPUT) | |
552 | { | |
553 | if (__glibc_unlikely (outerr != outbuf)) | |
554 | { | |
555 | #ifdef RESET_INPUT_BUFFER | |
556 | /* RESET_INPUT_BUFFER can only work when there were | |
557 | no new irreversible characters skipped during | |
558 | this round. */ | |
559 | if (loop_irreversible | |
560 | == lirreversible + (irreversible ? *irreversible : 0)) | |
561 | { | |
562 | RESET_INPUT_BUFFER; | |
563 | goto done_reset; | |
564 | } | |
565 | #endif | |
566 | /* We have a problem in one of the functions below. | |
567 | Undo the conversion upto the error point. */ | |
568 | size_t nstatus __attribute__ ((unused)); | |
569 | ||
570 | /* Reload the pointers. */ | |
571 | *inptrp = inptr; | |
572 | outbuf = outstart; | |
573 | ||
574 | /* Restore the state. */ | |
575 | #ifdef SAVE_RESET_STATE | |
576 | SAVE_RESET_STATE (0); | |
577 | #endif | |
578 | ||
579 | if (FROM_DIRECTION) | |
580 | /* Run the conversion loop. */ | |
581 | nstatus = FROM_LOOP (step, data, inptrp, inend, | |
582 | &outbuf, outerr, | |
583 | lirreversiblep | |
584 | EXTRA_LOOP_ARGS); | |
585 | else | |
586 | /* Run the conversion loop. */ | |
587 | nstatus = TO_LOOP (step, data, inptrp, inend, | |
588 | &outbuf, outerr, | |
589 | lirreversiblep | |
590 | EXTRA_LOOP_ARGS); | |
591 | ||
592 | /* We must run out of output buffer space in this | |
593 | rerun. */ | |
594 | assert (outbuf == outerr); | |
595 | assert (nstatus == __GCONV_FULL_OUTPUT); | |
596 | ||
597 | /* If we haven't consumed a single byte decrement | |
598 | the invocation counter. */ | |
599 | if (__glibc_unlikely (outbuf == outstart)) | |
600 | --data->__invocation_counter; | |
601 | } | |
602 | ||
603 | #ifdef RESET_INPUT_BUFFER | |
604 | done_reset: | |
605 | #endif | |
606 | /* Change the status. */ | |
607 | status = result; | |
608 | } | |
609 | else | |
610 | /* All the output is consumed, we can make another run | |
611 | if everything was ok. */ | |
612 | if (status == __GCONV_FULL_OUTPUT) | |
613 | { | |
614 | status = __GCONV_OK; | |
615 | outbuf = data->__outbuf; | |
616 | } | |
617 | } | |
618 | ||
619 | if (status != __GCONV_OK) | |
620 | break; | |
621 | ||
622 | /* Reset the output buffer pointer for the next round. */ | |
623 | outbuf = data->__outbuf; | |
624 | } | |
625 | ||
626 | #ifdef END_LOOP | |
627 | END_LOOP | |
628 | #endif | |
629 | ||
630 | /* If we are supposed to consume all character store now all of the | |
631 | remaining characters in the `state' object. */ | |
632 | #if FROM_LOOP_MAX_NEEDED_FROM > 1 || TO_LOOP_MAX_NEEDED_FROM > 1 | |
633 | if (((FROM_LOOP_MAX_NEEDED_FROM > 1 && TO_LOOP_MAX_NEEDED_FROM > 1) | |
634 | || (FROM_LOOP_MAX_NEEDED_FROM > 1 && FROM_DIRECTION) | |
635 | || (TO_LOOP_MAX_NEEDED_FROM > 1 && !FROM_DIRECTION)) | |
636 | && __builtin_expect (consume_incomplete, 0) | |
637 | && status == __GCONV_INCOMPLETE_INPUT) | |
638 | { | |
639 | # ifdef STORE_REST | |
640 | mbstate_t *state = data->__statep; | |
641 | ||
642 | STORE_REST | |
643 | # else | |
644 | /* Make sure the remaining bytes fit into the state objects | |
645 | buffer. */ | |
646 | size_t cnt_after = inend - *inptrp; | |
647 | assert (cnt_after <= sizeof (data->__statep->__value.__wchb)); | |
648 | ||
649 | size_t cnt; | |
650 | for (cnt = 0; cnt < cnt_after; ++cnt) | |
651 | data->__statep->__value.__wchb[cnt] = (*inptrp)[cnt]; | |
652 | *inptrp = inend; | |
653 | data->__statep->__count &= ~7; | |
654 | data->__statep->__count |= cnt; | |
655 | # endif | |
656 | } | |
657 | #endif | |
658 | #undef unaligned | |
659 | #undef POSSIBLY_UNALIGNED | |
660 | } | |
661 | ||
662 | return status; | |
663 | } | |
664 | ||
665 | #undef DEFINE_INIT | |
666 | #undef CHARSET_NAME | |
667 | #undef DEFINE_FINI | |
668 | #undef MIN_NEEDED_FROM | |
669 | #undef MIN_NEEDED_TO | |
670 | #undef MAX_NEEDED_FROM | |
671 | #undef MAX_NEEDED_TO | |
672 | #undef FROM_LOOP_MIN_NEEDED_FROM | |
673 | #undef FROM_LOOP_MAX_NEEDED_FROM | |
674 | #undef FROM_LOOP_MIN_NEEDED_TO | |
675 | #undef FROM_LOOP_MAX_NEEDED_TO | |
676 | #undef TO_LOOP_MIN_NEEDED_FROM | |
677 | #undef TO_LOOP_MAX_NEEDED_FROM | |
678 | #undef TO_LOOP_MIN_NEEDED_TO | |
679 | #undef TO_LOOP_MAX_NEEDED_TO | |
680 | #undef FROM_DIRECTION | |
681 | #undef EMIT_SHIFT_TO_INIT | |
682 | #undef FROM_LOOP | |
683 | #undef TO_LOOP | |
684 | #undef ONE_DIRECTION | |
685 | #undef SAVE_RESET_STATE | |
686 | #undef RESET_INPUT_BUFFER | |
687 | #undef FUNCTION_NAME | |
688 | #undef PREPARE_LOOP | |
689 | #undef END_LOOP | |
690 | #undef EXTRA_LOOP_ARGS | |
691 | #undef STORE_REST | |
692 | #undef FROM_ONEBYTE |