1 /* Simple transformations functions.
2 Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
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.
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.
16 You should have received a copy of the GNU Lesser General Public
17 License along with the GNU C Library; if not, write to the Free
18 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
30 #include <sys/param.h>
32 #define BUILTIN_ALIAS(s1, s2) /* nothing */
33 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, MinF, MaxF, \
35 extern int Fct (struct __gconv_step *, struct __gconv_step_data *, \
36 __const unsigned char **, __const unsigned char *, \
37 unsigned char **, size_t *, int, int);
38 #include "gconv_builtin.h"
42 # define EILSEQ EINVAL
46 /* Transform from the internal, UCS4-like format, to UCS4. The
47 difference between the internal ucs4 format and the real UCS4
48 format is, if any, the endianess. The Unicode/ISO 10646 says that
49 unless some higher protocol specifies it differently, the byte
50 order is big endian.*/
53 #define MIN_NEEDED_FROM 4
54 #define MIN_NEEDED_TO 4
55 #define FROM_DIRECTION 1
56 #define FROM_LOOP internal_ucs4_loop
57 #define TO_LOOP internal_ucs4_loop /* This is not used. */
58 #define FUNCTION_NAME __gconv_transform_internal_ucs4
62 internal_ucs4_loop (struct __gconv_step
*step
,
63 struct __gconv_step_data
*step_data
,
64 const unsigned char **inptrp
, const unsigned char *inend
,
65 unsigned char **outptrp
, unsigned char *outend
,
68 const unsigned char *inptr
= *inptrp
;
69 unsigned char *outptr
= *outptrp
;
70 size_t n_convert
= MIN (inend
- inptr
, outend
- outptr
) / 4;
73 #if __BYTE_ORDER == __LITTLE_ENDIAN
74 /* Sigh, we have to do some real work. */
77 for (cnt
= 0; cnt
< n_convert
; ++cnt
, inptr
+= 4)
78 *((uint32_t *) outptr
)++ = bswap_32 (*(const uint32_t *) inptr
);
82 #elif __BYTE_ORDER == __BIG_ENDIAN
83 /* Simply copy the data. */
84 *inptrp
= inptr
+ n_convert
* 4;
85 *outptrp
= __mempcpy (outptr
, inptr
, n_convert
* 4);
87 # error "This endianess is not supported."
90 /* Determine the status. */
92 result
= __GCONV_EMPTY_INPUT
;
93 else if (*outptrp
+ 4 > outend
)
94 result
= __GCONV_FULL_OUTPUT
;
96 result
= __GCONV_INCOMPLETE_INPUT
;
101 #ifndef _STRING_ARCH_unaligned
103 internal_ucs4_loop_unaligned (struct __gconv_step
*step
,
104 struct __gconv_step_data
*step_data
,
105 const unsigned char **inptrp
,
106 const unsigned char *inend
,
107 unsigned char **outptrp
, unsigned char *outend
,
108 size_t *irreversible
)
110 const unsigned char *inptr
= *inptrp
;
111 unsigned char *outptr
= *outptrp
;
112 size_t n_convert
= MIN (inend
- inptr
, outend
- outptr
) / 4;
115 # if __BYTE_ORDER == __LITTLE_ENDIAN
116 /* Sigh, we have to do some real work. */
119 for (cnt
= 0; cnt
< n_convert
; ++cnt
, inptr
+= 4, outptr
+= 4)
121 outptr
[0] = inptr
[3];
122 outptr
[1] = inptr
[2];
123 outptr
[2] = inptr
[1];
124 outptr
[3] = inptr
[0];
129 # elif __BYTE_ORDER == __BIG_ENDIAN
130 /* Simply copy the data. */
131 *inptrp
= inptr
+ n_convert
* 4;
132 *outptrp
= __mempcpy (outptr
, inptr
, n_convert
* 4);
134 # error "This endianess is not supported."
137 /* Determine the status. */
138 if (*inptrp
== inend
)
139 result
= __GCONV_EMPTY_INPUT
;
140 else if (*outptrp
+ 4 > outend
)
141 result
= __GCONV_FULL_OUTPUT
;
143 result
= __GCONV_INCOMPLETE_INPUT
;
151 internal_ucs4_loop_single (struct __gconv_step
*step
,
152 struct __gconv_step_data
*step_data
,
153 const unsigned char **inptrp
,
154 const unsigned char *inend
,
155 unsigned char **outptrp
, unsigned char *outend
,
156 size_t *irreversible
)
158 mbstate_t *state
= step_data
->__statep
;
159 size_t cnt
= state
->__count
& 7;
161 while (*inptrp
< inend
&& cnt
< 4)
162 state
->__value
.__wchb
[cnt
++] = *(*inptrp
)++;
164 if (__builtin_expect (cnt
< 4, 0))
166 /* Still not enough bytes. Store the ones in the input buffer. */
167 state
->__count
&= ~7;
168 state
->__count
|= cnt
;
170 return __GCONV_INCOMPLETE_INPUT
;
173 #if __BYTE_ORDER == __LITTLE_ENDIAN
174 (*outptrp
)[0] = state
->__value
.__wchb
[3];
175 (*outptrp
)[1] = state
->__value
.__wchb
[2];
176 (*outptrp
)[2] = state
->__value
.__wchb
[1];
177 (*outptrp
)[3] = state
->__value
.__wchb
[0];
180 #elif __BYTE_ORDER == __BIG_ENDIAN
182 *(*((uint32_t **) outptrp
)++) = state
->__value
.__wch
;
184 # error "This endianess is not supported."
187 /* Clear the state buffer. */
188 state
->__count
&= ~7;
193 #include <iconv/skeleton.c>
196 /* Transform from UCS4 to the internal, UCS4-like format. Unlike
197 for the other direction we have to check for correct values here. */
198 #define DEFINE_INIT 0
199 #define DEFINE_FINI 0
200 #define MIN_NEEDED_FROM 4
201 #define MIN_NEEDED_TO 4
202 #define FROM_DIRECTION 1
203 #define FROM_LOOP ucs4_internal_loop
204 #define TO_LOOP ucs4_internal_loop /* This is not used. */
205 #define FUNCTION_NAME __gconv_transform_ucs4_internal
209 ucs4_internal_loop (struct __gconv_step
*step
,
210 struct __gconv_step_data
*step_data
,
211 const unsigned char **inptrp
, const unsigned char *inend
,
212 unsigned char **outptrp
, unsigned char *outend
,
213 size_t *irreversible
)
215 int flags
= step_data
->__flags
;
216 const unsigned char *inptr
= *inptrp
;
217 unsigned char *outptr
= *outptrp
;
218 size_t n_convert
= MIN (inend
- inptr
, outend
- outptr
) / 4;
222 for (cnt
= 0; cnt
< n_convert
; ++cnt
, inptr
+= 4)
226 #if __BYTE_ORDER == __LITTLE_ENDIAN
227 inval
= bswap_32 (*(const uint32_t *) inptr
);
229 inval
= *(const uint32_t *) inptr
;
232 if (__builtin_expect (inval
> 0x7fffffff, 0))
234 /* The value is too large. We don't try transliteration here since
235 this is not an error because of the lack of possibilities to
236 represent the result. This is a genuine bug in the input since
237 UCS4 does not allow such values. */
238 if (irreversible
== NULL
)
239 /* We are transliterating, don't try to correct anything. */
240 return __GCONV_ILLEGAL_INPUT
;
242 if (flags
& __GCONV_IGNORE_ERRORS
)
244 /* Just ignore this character. */
251 return __GCONV_ILLEGAL_INPUT
;
254 *((uint32_t *) outptr
)++ = inval
;
260 /* Determine the status. */
261 if (*inptrp
== inend
)
262 result
= __GCONV_EMPTY_INPUT
;
263 else if (*outptrp
+ 4 > outend
)
264 result
= __GCONV_FULL_OUTPUT
;
266 result
= __GCONV_INCOMPLETE_INPUT
;
271 #ifndef _STRING_ARCH_unaligned
273 ucs4_internal_loop_unaligned (struct __gconv_step
*step
,
274 struct __gconv_step_data
*step_data
,
275 const unsigned char **inptrp
,
276 const unsigned char *inend
,
277 unsigned char **outptrp
, unsigned char *outend
,
278 size_t *irreversible
)
280 int flags
= step_data
->__flags
;
281 const unsigned char *inptr
= *inptrp
;
282 unsigned char *outptr
= *outptrp
;
283 size_t n_convert
= MIN (inend
- inptr
, outend
- outptr
) / 4;
287 for (cnt
= 0; cnt
< n_convert
; ++cnt
, inptr
+= 4)
289 if (__builtin_expect (inptr
[0] > 0x80, 0))
291 /* The value is too large. We don't try transliteration here since
292 this is not an error because of the lack of possibilities to
293 represent the result. This is a genuine bug in the input since
294 UCS4 does not allow such values. */
295 if (irreversible
== NULL
)
296 /* We are transliterating, don't try to correct anything. */
297 return __GCONV_ILLEGAL_INPUT
;
299 if (flags
& __GCONV_IGNORE_ERRORS
)
301 /* Just ignore this character. */
308 return __GCONV_ILLEGAL_INPUT
;
311 # if __BYTE_ORDER == __LITTLE_ENDIAN
312 outptr
[3] = inptr
[0];
313 outptr
[2] = inptr
[1];
314 outptr
[1] = inptr
[2];
315 outptr
[0] = inptr
[3];
317 outptr
[0] = inptr
[0];
318 outptr
[1] = inptr
[1];
319 outptr
[2] = inptr
[2];
320 outptr
[3] = inptr
[3];
328 /* Determine the status. */
329 if (*inptrp
== inend
)
330 result
= __GCONV_EMPTY_INPUT
;
331 else if (*outptrp
+ 4 > outend
)
332 result
= __GCONV_FULL_OUTPUT
;
334 result
= __GCONV_INCOMPLETE_INPUT
;
342 ucs4_internal_loop_single (struct __gconv_step
*step
,
343 struct __gconv_step_data
*step_data
,
344 const unsigned char **inptrp
,
345 const unsigned char *inend
,
346 unsigned char **outptrp
, unsigned char *outend
,
347 size_t *irreversible
)
349 mbstate_t *state
= step_data
->__statep
;
350 int flags
= step_data
->__flags
;
351 size_t cnt
= state
->__count
& 7;
353 while (*inptrp
< inend
&& cnt
< 4)
354 state
->__value
.__wchb
[cnt
++] = *(*inptrp
)++;
356 if (__builtin_expect (cnt
< 4, 0))
358 /* Still not enough bytes. Store the ones in the input buffer. */
359 state
->__count
&= ~7;
360 state
->__count
|= cnt
;
362 return __GCONV_INCOMPLETE_INPUT
;
365 if (__builtin_expect (((unsigned char *) state
->__value
.__wchb
)[0] > 0x80,
368 /* The value is too large. We don't try transliteration here since
369 this is not an error because of the lack of possibilities to
370 represent the result. This is a genuine bug in the input since
371 UCS4 does not allow such values. */
372 if (!(flags
& __GCONV_IGNORE_ERRORS
))
374 *inptrp
-= cnt
- (state
->__count
& 7);
375 return __GCONV_ILLEGAL_INPUT
;
380 #if __BYTE_ORDER == __LITTLE_ENDIAN
381 (*outptrp
)[0] = state
->__value
.__wchb
[3];
382 (*outptrp
)[1] = state
->__value
.__wchb
[2];
383 (*outptrp
)[2] = state
->__value
.__wchb
[1];
384 (*outptrp
)[3] = state
->__value
.__wchb
[0];
385 #elif __BYTE_ORDER == __BIG_ENDIAN
386 (*outptrp
)[0] = state
->__value
.__wchb
[0];
387 (*outptrp
)[1] = state
->__value
.__wchb
[1];
388 (*outptrp
)[2] = state
->__value
.__wchb
[2];
389 (*outptrp
)[3] = state
->__value
.__wchb
[3];
395 /* Clear the state buffer. */
396 state
->__count
&= ~7;
401 #include <iconv/skeleton.c>
404 /* Similarly for the little endian form. */
405 #define DEFINE_INIT 0
406 #define DEFINE_FINI 0
407 #define MIN_NEEDED_FROM 4
408 #define MIN_NEEDED_TO 4
409 #define FROM_DIRECTION 1
410 #define FROM_LOOP internal_ucs4le_loop
411 #define TO_LOOP internal_ucs4le_loop /* This is not used. */
412 #define FUNCTION_NAME __gconv_transform_internal_ucs4le
416 internal_ucs4le_loop (struct __gconv_step
*step
,
417 struct __gconv_step_data
*step_data
,
418 const unsigned char **inptrp
, const unsigned char *inend
,
419 unsigned char **outptrp
, unsigned char *outend
,
420 size_t *irreversible
)
422 const unsigned char *inptr
= *inptrp
;
423 unsigned char *outptr
= *outptrp
;
424 size_t n_convert
= MIN (inend
- inptr
, outend
- outptr
) / 4;
427 #if __BYTE_ORDER == __BIG_ENDIAN
428 /* Sigh, we have to do some real work. */
431 for (cnt
= 0; cnt
< n_convert
; ++cnt
, inptr
+= 4)
432 *((uint32_t *) outptr
)++ = bswap_32 (*(const uint32_t *) inptr
);
436 #elif __BYTE_ORDER == __LITTLE_ENDIAN
437 /* Simply copy the data. */
438 *inptrp
= inptr
+ n_convert
* 4;
439 *outptrp
= __mempcpy (outptr
, inptr
, n_convert
* 4);
441 # error "This endianess is not supported."
444 /* Determine the status. */
445 if (*inptrp
== inend
)
446 result
= __GCONV_EMPTY_INPUT
;
447 else if (*outptrp
+ 4 > outend
)
448 result
= __GCONV_FULL_OUTPUT
;
450 result
= __GCONV_INCOMPLETE_INPUT
;
455 #ifndef _STRING_ARCH_unaligned
457 internal_ucs4le_loop_unaligned (struct __gconv_step
*step
,
458 struct __gconv_step_data
*step_data
,
459 const unsigned char **inptrp
,
460 const unsigned char *inend
,
461 unsigned char **outptrp
, unsigned char *outend
,
462 size_t *irreversible
)
464 const unsigned char *inptr
= *inptrp
;
465 unsigned char *outptr
= *outptrp
;
466 size_t n_convert
= MIN (inend
- inptr
, outend
- outptr
) / 4;
469 # if __BYTE_ORDER == __BIG_ENDIAN
470 /* Sigh, we have to do some real work. */
473 for (cnt
= 0; cnt
< n_convert
; ++cnt
, inptr
+= 4, outptr
+= 4)
475 outptr
[0] = inptr
[3];
476 outptr
[1] = inptr
[2];
477 outptr
[2] = inptr
[1];
478 outptr
[3] = inptr
[0];
483 # elif __BYTE_ORDER == __LITTLE_ENDIAN
484 /* Simply copy the data. */
485 *inptrp
= inptr
+ n_convert
* 4;
486 *outptrp
= __mempcpy (outptr
, inptr
, n_convert
* 4);
488 # error "This endianess is not supported."
491 /* Determine the status. */
492 if (*inptrp
== inend
)
493 result
= __GCONV_EMPTY_INPUT
;
494 else if (*inptrp
+ 4 > inend
)
495 result
= __GCONV_INCOMPLETE_INPUT
;
498 assert (*outptrp
+ 4 > outend
);
499 result
= __GCONV_FULL_OUTPUT
;
508 internal_ucs4le_loop_single (struct __gconv_step
*step
,
509 struct __gconv_step_data
*step_data
,
510 const unsigned char **inptrp
,
511 const unsigned char *inend
,
512 unsigned char **outptrp
, unsigned char *outend
,
513 size_t *irreversible
)
515 mbstate_t *state
= step_data
->__statep
;
516 size_t cnt
= state
->__count
& 7;
518 while (*inptrp
< inend
&& cnt
< 4)
519 state
->__value
.__wchb
[cnt
++] = *(*inptrp
)++;
521 if (__builtin_expect (cnt
< 4, 0))
523 /* Still not enough bytes. Store the ones in the input buffer. */
524 state
->__count
&= ~7;
525 state
->__count
|= cnt
;
527 return __GCONV_INCOMPLETE_INPUT
;
530 #if __BYTE_ORDER == __BIG_ENDIAN
531 (*outptrp
)[0] = state
->__value
.__wchb
[3];
532 (*outptrp
)[1] = state
->__value
.__wchb
[2];
533 (*outptrp
)[2] = state
->__value
.__wchb
[1];
534 (*outptrp
)[3] = state
->__value
.__wchb
[0];
539 *(*((uint32_t **) outptrp
)++) = state
->__value
.__wch
;
542 /* Clear the state buffer. */
543 state
->__count
&= ~7;
548 #include <iconv/skeleton.c>
551 /* And finally from UCS4-LE to the internal encoding. */
552 #define DEFINE_INIT 0
553 #define DEFINE_FINI 0
554 #define MIN_NEEDED_FROM 4
555 #define MIN_NEEDED_TO 4
556 #define FROM_DIRECTION 1
557 #define FROM_LOOP ucs4le_internal_loop
558 #define TO_LOOP ucs4le_internal_loop /* This is not used. */
559 #define FUNCTION_NAME __gconv_transform_ucs4le_internal
563 ucs4le_internal_loop (struct __gconv_step
*step
,
564 struct __gconv_step_data
*step_data
,
565 const unsigned char **inptrp
, const unsigned char *inend
,
566 unsigned char **outptrp
, unsigned char *outend
,
567 size_t *irreversible
)
569 int flags
= step_data
->__flags
;
570 const unsigned char *inptr
= *inptrp
;
571 unsigned char *outptr
= *outptrp
;
572 size_t n_convert
= MIN (inend
- inptr
, outend
- outptr
) / 4;
576 for (cnt
= 0; cnt
< n_convert
; ++cnt
, inptr
+= 4)
580 #if __BYTE_ORDER == __BIG_ENDIAN
581 inval
= bswap_32 (*(const uint32_t *) inptr
);
583 inval
= *(const uint32_t *) inptr
;
586 if (__builtin_expect (inval
> 0x7fffffff, 0))
588 /* The value is too large. We don't try transliteration here since
589 this is not an error because of the lack of possibilities to
590 represent the result. This is a genuine bug in the input since
591 UCS4 does not allow such values. */
592 if (irreversible
== NULL
)
593 /* We are transliterating, don't try to correct anything. */
594 return __GCONV_ILLEGAL_INPUT
;
596 if (flags
& __GCONV_IGNORE_ERRORS
)
598 /* Just ignore this character. */
603 return __GCONV_ILLEGAL_INPUT
;
606 *((uint32_t *) outptr
)++ = inval
;
612 /* Determine the status. */
613 if (*inptrp
== inend
)
614 result
= __GCONV_EMPTY_INPUT
;
615 else if (*inptrp
+ 4 > inend
)
616 result
= __GCONV_INCOMPLETE_INPUT
;
619 assert (*outptrp
+ 4 > outend
);
620 result
= __GCONV_FULL_OUTPUT
;
626 #ifndef _STRING_ARCH_unaligned
628 ucs4le_internal_loop_unaligned (struct __gconv_step
*step
,
629 struct __gconv_step_data
*step_data
,
630 const unsigned char **inptrp
,
631 const unsigned char *inend
,
632 unsigned char **outptrp
, unsigned char *outend
,
633 size_t *irreversible
)
635 int flags
= step_data
->__flags
;
636 const unsigned char *inptr
= *inptrp
;
637 unsigned char *outptr
= *outptrp
;
638 size_t n_convert
= MIN (inend
- inptr
, outend
- outptr
) / 4;
642 for (cnt
= 0; cnt
< n_convert
; ++cnt
, inptr
+= 4)
644 if (__builtin_expect (inptr
[3] > 0x80, 0))
646 /* The value is too large. We don't try transliteration here since
647 this is not an error because of the lack of possibilities to
648 represent the result. This is a genuine bug in the input since
649 UCS4 does not allow such values. */
650 if (irreversible
== NULL
)
651 /* We are transliterating, don't try to correct anything. */
652 return __GCONV_ILLEGAL_INPUT
;
654 if (flags
& __GCONV_IGNORE_ERRORS
)
656 /* Just ignore this character. */
663 return __GCONV_ILLEGAL_INPUT
;
666 # if __BYTE_ORDER == __BIG_ENDIAN
667 outptr
[3] = inptr
[0];
668 outptr
[2] = inptr
[1];
669 outptr
[1] = inptr
[2];
670 outptr
[0] = inptr
[3];
672 outptr
[0] = inptr
[0];
673 outptr
[1] = inptr
[1];
674 outptr
[2] = inptr
[2];
675 outptr
[3] = inptr
[3];
684 /* Determine the status. */
685 if (*inptrp
== inend
)
686 result
= __GCONV_EMPTY_INPUT
;
687 else if (*inptrp
+ 4 > inend
)
688 result
= __GCONV_INCOMPLETE_INPUT
;
691 assert (*outptrp
+ 4 > outend
);
692 result
= __GCONV_FULL_OUTPUT
;
701 ucs4le_internal_loop_single (struct __gconv_step
*step
,
702 struct __gconv_step_data
*step_data
,
703 const unsigned char **inptrp
,
704 const unsigned char *inend
,
705 unsigned char **outptrp
, unsigned char *outend
,
706 size_t *irreversible
)
708 mbstate_t *state
= step_data
->__statep
;
709 int flags
= step_data
->__flags
;
710 size_t cnt
= state
->__count
& 7;
712 while (*inptrp
< inend
&& cnt
< 4)
713 state
->__value
.__wchb
[cnt
++] = *(*inptrp
)++;
715 if (__builtin_expect (cnt
< 4, 0))
717 /* Still not enough bytes. Store the ones in the input buffer. */
718 state
->__count
&= ~7;
719 state
->__count
|= cnt
;
721 return __GCONV_INCOMPLETE_INPUT
;
724 if (__builtin_expect (((unsigned char *) state
->__value
.__wchb
)[3] > 0x80,
727 /* The value is too large. We don't try transliteration here since
728 this is not an error because of the lack of possibilities to
729 represent the result. This is a genuine bug in the input since
730 UCS4 does not allow such values. */
731 if (!(flags
& __GCONV_IGNORE_ERRORS
))
732 return __GCONV_ILLEGAL_INPUT
;
736 #if __BYTE_ORDER == __BIG_ENDIAN
737 (*outptrp
)[0] = state
->__value
.__wchb
[3];
738 (*outptrp
)[1] = state
->__value
.__wchb
[2];
739 (*outptrp
)[2] = state
->__value
.__wchb
[1];
740 (*outptrp
)[3] = state
->__value
.__wchb
[0];
742 (*outptrp
)[0] = state
->__value
.__wchb
[0];
743 (*outptrp
)[1] = state
->__value
.__wchb
[1];
744 (*outptrp
)[2] = state
->__value
.__wchb
[2];
745 (*outptrp
)[3] = state
->__value
.__wchb
[3];
751 /* Clear the state buffer. */
752 state
->__count
&= ~7;
757 #include <iconv/skeleton.c>
760 /* Convert from ISO 646-IRV to the internal (UCS4-like) format. */
761 #define DEFINE_INIT 0
762 #define DEFINE_FINI 0
763 #define MIN_NEEDED_FROM 1
764 #define MIN_NEEDED_TO 4
765 #define FROM_DIRECTION 1
766 #define FROM_LOOP ascii_internal_loop
767 #define TO_LOOP ascii_internal_loop /* This is not used. */
768 #define FUNCTION_NAME __gconv_transform_ascii_internal
769 #define ONE_DIRECTION 1
771 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
772 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
773 #define LOOPFCT FROM_LOOP
776 if (__builtin_expect (*inptr > '\x7f', 0)) \
778 /* The value is too large. We don't try transliteration here since \
779 this is not an error because of the lack of possibilities to \
780 represent the result. This is a genuine bug in the input since \
781 ASCII does not allow such values. */ \
782 STANDARD_FROM_LOOP_ERR_HANDLER (1); \
785 /* It's an one byte sequence. */ \
786 *((uint32_t *) outptr)++ = *inptr++; \
788 #define LOOP_NEED_FLAGS
789 #include <iconv/loop.c>
790 #include <iconv/skeleton.c>
793 /* Convert from the internal (UCS4-like) format to ISO 646-IRV. */
794 #define DEFINE_INIT 0
795 #define DEFINE_FINI 0
796 #define MIN_NEEDED_FROM 4
797 #define MIN_NEEDED_TO 1
798 #define FROM_DIRECTION 1
799 #define FROM_LOOP internal_ascii_loop
800 #define TO_LOOP internal_ascii_loop /* This is not used. */
801 #define FUNCTION_NAME __gconv_transform_internal_ascii
802 #define ONE_DIRECTION 1
804 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
805 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
806 #define LOOPFCT FROM_LOOP
809 if (__builtin_expect (*((const uint32_t *) inptr) > 0x7f, 0)) \
811 UNICODE_TAG_HANDLER (*((const uint32_t *) inptr), 4); \
812 STANDARD_TO_LOOP_ERR_HANDLER (4); \
815 /* It's an one byte sequence. */ \
816 *outptr++ = *((const uint32_t *) inptr)++; \
818 #define LOOP_NEED_FLAGS
819 #include <iconv/loop.c>
820 #include <iconv/skeleton.c>
823 /* Convert from the internal (UCS4-like) format to UTF-8. */
824 #define DEFINE_INIT 0
825 #define DEFINE_FINI 0
826 #define MIN_NEEDED_FROM 4
827 #define MIN_NEEDED_TO 1
828 #define MAX_NEEDED_TO 6
829 #define FROM_DIRECTION 1
830 #define FROM_LOOP internal_utf8_loop
831 #define TO_LOOP internal_utf8_loop /* This is not used. */
832 #define FUNCTION_NAME __gconv_transform_internal_utf8
833 #define ONE_DIRECTION 1
835 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
836 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
837 #define MAX_NEEDED_OUTPUT MAX_NEEDED_TO
838 #define LOOPFCT FROM_LOOP
841 uint32_t wc = *((const uint32_t *) inptr); \
844 /* It's an one byte sequence. */ \
845 *outptr++ = (unsigned char) wc; \
846 else if (__builtin_expect (wc <= 0x7fffffff, 1)) \
851 for (step = 2; step < 6; ++step) \
852 if ((wc & (~(uint32_t)0 << (5 * step + 1))) == 0) \
855 if (__builtin_expect (outptr + step > outend, 0)) \
858 result = __GCONV_FULL_OUTPUT; \
863 *outptr = (unsigned char) (~0xff >> step); \
868 start[step] = 0x80 | (wc & 0x3f); \
871 while (--step > 0); \
876 STANDARD_TO_LOOP_ERR_HANDLER (4); \
881 #define LOOP_NEED_FLAGS
882 #include <iconv/loop.c>
883 #include <iconv/skeleton.c>
886 /* Convert from UTF-8 to the internal (UCS4-like) format. */
887 #define DEFINE_INIT 0
888 #define DEFINE_FINI 0
889 #define MIN_NEEDED_FROM 1
890 #define MAX_NEEDED_FROM 6
891 #define MIN_NEEDED_TO 4
892 #define FROM_DIRECTION 1
893 #define FROM_LOOP utf8_internal_loop
894 #define TO_LOOP utf8_internal_loop /* This is not used. */
895 #define FUNCTION_NAME __gconv_transform_utf8_internal
896 #define ONE_DIRECTION 1
898 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
899 #define MAX_NEEDED_INPUT MAX_NEEDED_FROM
900 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
901 #define LOOPFCT FROM_LOOP
908 /* Next input byte. */ \
913 /* One byte sequence. */ \
919 if (ch >= 0xc2 && ch < 0xe0) \
921 /* We expect two bytes. The first byte cannot be 0xc0 or 0xc1, \
922 otherwise the wide character could have been represented \
923 using a single byte. */ \
927 else if (__builtin_expect ((ch & 0xf0) == 0xe0, 1)) \
929 /* We expect three bytes. */ \
933 else if (__builtin_expect ((ch & 0xf8) == 0xf0, 1)) \
935 /* We expect four bytes. */ \
939 else if (__builtin_expect ((ch & 0xfc) == 0xf8, 1)) \
941 /* We expect five bytes. */ \
945 else if (__builtin_expect ((ch & 0xfe) == 0xfc, 1)) \
947 /* We expect six bytes. */ \
955 /* Search the end of this ill-formed UTF-8 character. This \
956 is the next byte with (x & 0xc0) != 0x80. */ \
960 while (inptr + skipped < inend \
961 && (*(inptr + skipped) & 0xc0) == 0x80 \
964 STANDARD_FROM_LOOP_ERR_HANDLER (skipped); \
967 if (__builtin_expect (inptr + cnt > inend, 0)) \
969 /* We don't have enough input. But before we report that check \
970 that all the bytes are correct. */ \
971 for (i = 1; inptr + i < inend; ++i) \
972 if ((inptr[i] & 0xc0) != 0x80) \
975 if (__builtin_expect (inptr + i == inend, 1)) \
977 result = __GCONV_INCOMPLETE_INPUT; \
981 STANDARD_FROM_LOOP_ERR_HANDLER (i); \
984 /* Read the possible remaining bytes. */ \
985 for (i = 1; i < cnt; ++i) \
987 uint32_t byte = inptr[i]; \
989 if ((byte & 0xc0) != 0x80) \
990 /* This is an illegal encoding. */ \
997 /* If i < cnt, some trail byte was not >= 0x80, < 0xc0. \
998 If cnt > 2 and ch < 2^(5*cnt-4), the wide character ch could \
999 have been represented with fewer than cnt bytes. */ \
1000 if (i < cnt || (cnt > 2 && (ch >> (5 * cnt - 4)) == 0)) \
1002 /* This is an illegal encoding. */ \
1003 STANDARD_FROM_LOOP_ERR_HANDLER (i); \
1009 /* Now adjust the pointers and store the result. */ \
1010 *((uint32_t *) outptr)++ = ch; \
1012 #define LOOP_NEED_FLAGS
1014 #define STORE_REST \
1016 /* We store the remaining bytes while converting them into the UCS4 \
1017 format. We can assume that the first byte in the buffer is \
1018 correct and that it requires a larger number of bytes than there \
1019 are in the input buffer. */ \
1020 wint_t ch = **inptrp; \
1023 state->__count = inend - *inptrp; \
1025 if (ch >= 0xc2 && ch < 0xe0) \
1027 /* We expect two bytes. The first byte cannot be 0xc0 or \
1028 0xc1, otherwise the wide character could have been \
1029 represented using a single byte. */ \
1033 else if (__builtin_expect ((ch & 0xf0) == 0xe0, 1)) \
1035 /* We expect three bytes. */ \
1039 else if (__builtin_expect ((ch & 0xf8) == 0xf0, 1)) \
1041 /* We expect four bytes. */ \
1045 else if (__builtin_expect ((ch & 0xfc) == 0xf8, 1)) \
1047 /* We expect five bytes. */ \
1053 /* We expect six bytes. */ \
1058 /* The first byte is already consumed. */ \
1060 while (++(*inptrp) < inend) \
1063 ch |= **inptrp & 0x3f; \
1067 /* Shift for the so far missing bytes. */ \
1070 /* Store the value. */ \
1071 state->__value.__wch = ch; \
1074 #define UNPACK_BYTES \
1076 wint_t wch = state->__value.__wch; \
1078 inlen = state->__count; \
1080 if (state->__value.__wch <= 0x7ff) \
1082 bytebuf[0] = 0xc0; \
1085 else if (__builtin_expect (state->__value.__wch <= 0xffff, 1)) \
1087 bytebuf[0] = 0xe0; \
1090 else if (__builtin_expect (state->__value.__wch < 0x1fffff, 1)) \
1092 bytebuf[0] = 0xf0; \
1095 else if (__builtin_expect (state->__value.__wch < 0x3ffffff, 1)) \
1097 bytebuf[0] = 0xf8; \
1102 bytebuf[0] = 0xfc; \
1108 if (--ntotal < inlen) \
1109 bytebuf[ntotal] = 0x80 | (wch & 0x3f); \
1112 while (ntotal > 1); \
1114 bytebuf[0] |= wch; \
1117 #include <iconv/loop.c>
1118 #include <iconv/skeleton.c>
1121 /* Convert from UCS2 to the internal (UCS4-like) format. */
1122 #define DEFINE_INIT 0
1123 #define DEFINE_FINI 0
1124 #define MIN_NEEDED_FROM 2
1125 #define MIN_NEEDED_TO 4
1126 #define FROM_DIRECTION 1
1127 #define FROM_LOOP ucs2_internal_loop
1128 #define TO_LOOP ucs2_internal_loop /* This is not used. */
1129 #define FUNCTION_NAME __gconv_transform_ucs2_internal
1130 #define ONE_DIRECTION 1
1132 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
1133 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
1134 #define LOOPFCT FROM_LOOP
1137 uint16_t u1 = *((const uint16_t *) inptr); \
1139 if (__builtin_expect (u1 >= 0xd800 && u1 < 0xe000, 0)) \
1141 /* Surrogate characters in UCS-2 input are not valid. Reject \
1142 them. (Catching this here is not security relevant.) */ \
1143 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
1146 *((uint32_t *) outptr)++ = u1; \
1149 #define LOOP_NEED_FLAGS
1150 #include <iconv/loop.c>
1151 #include <iconv/skeleton.c>
1154 /* Convert from the internal (UCS4-like) format to UCS2. */
1155 #define DEFINE_INIT 0
1156 #define DEFINE_FINI 0
1157 #define MIN_NEEDED_FROM 4
1158 #define MIN_NEEDED_TO 2
1159 #define FROM_DIRECTION 1
1160 #define FROM_LOOP internal_ucs2_loop
1161 #define TO_LOOP internal_ucs2_loop /* This is not used. */
1162 #define FUNCTION_NAME __gconv_transform_internal_ucs2
1163 #define ONE_DIRECTION 1
1165 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
1166 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
1167 #define LOOPFCT FROM_LOOP
1170 uint32_t val = *((const uint32_t *) inptr); \
1172 if (__builtin_expect (val >= 0x10000, 0)) \
1174 UNICODE_TAG_HANDLER (val, 4); \
1175 STANDARD_TO_LOOP_ERR_HANDLER (4); \
1177 else if (__builtin_expect (val >= 0xd800 && val < 0xe000, 0)) \
1179 /* Surrogate characters in UCS-4 input are not valid. \
1180 We must catch this, because the UCS-2 output might be \
1181 interpreted as UTF-16 by other programs. If we let \
1182 surrogates pass through, attackers could make a security \
1183 hole exploit by synthesizing any desired plane 1-16 \
1185 result = __GCONV_ILLEGAL_INPUT; \
1186 if (! ignore_errors_p ()) \
1194 *((uint16_t *) outptr)++ = val; \
1198 #define LOOP_NEED_FLAGS
1199 #include <iconv/loop.c>
1200 #include <iconv/skeleton.c>
1203 /* Convert from UCS2 in other endianness to the internal (UCS4-like) format. */
1204 #define DEFINE_INIT 0
1205 #define DEFINE_FINI 0
1206 #define MIN_NEEDED_FROM 2
1207 #define MIN_NEEDED_TO 4
1208 #define FROM_DIRECTION 1
1209 #define FROM_LOOP ucs2reverse_internal_loop
1210 #define TO_LOOP ucs2reverse_internal_loop/* This is not used.*/
1211 #define FUNCTION_NAME __gconv_transform_ucs2reverse_internal
1212 #define ONE_DIRECTION 1
1214 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
1215 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
1216 #define LOOPFCT FROM_LOOP
1219 uint16_t u1 = bswap_16 (*((const uint16_t *) inptr)); \
1221 if (__builtin_expect (u1 >= 0xd800 && u1 < 0xe000, 0)) \
1223 /* Surrogate characters in UCS-2 input are not valid. Reject \
1224 them. (Catching this here is not security relevant.) */ \
1225 if (! ignore_errors_p ()) \
1227 result = __GCONV_ILLEGAL_INPUT; \
1235 *((uint32_t *) outptr)++ = u1; \
1238 #define LOOP_NEED_FLAGS
1239 #include <iconv/loop.c>
1240 #include <iconv/skeleton.c>
1243 /* Convert from the internal (UCS4-like) format to UCS2 in other endianness. */
1244 #define DEFINE_INIT 0
1245 #define DEFINE_FINI 0
1246 #define MIN_NEEDED_FROM 4
1247 #define MIN_NEEDED_TO 2
1248 #define FROM_DIRECTION 1
1249 #define FROM_LOOP internal_ucs2reverse_loop
1250 #define TO_LOOP internal_ucs2reverse_loop/* This is not used.*/
1251 #define FUNCTION_NAME __gconv_transform_internal_ucs2reverse
1252 #define ONE_DIRECTION 1
1254 #define MIN_NEEDED_INPUT MIN_NEEDED_FROM
1255 #define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
1256 #define LOOPFCT FROM_LOOP
1259 uint32_t val = *((const uint32_t *) inptr); \
1260 if (__builtin_expect (val >= 0x10000, 0)) \
1262 UNICODE_TAG_HANDLER (val, 4); \
1263 STANDARD_TO_LOOP_ERR_HANDLER (4); \
1265 else if (__builtin_expect (val >= 0xd800 && val < 0xe000, 0)) \
1267 /* Surrogate characters in UCS-4 input are not valid. \
1268 We must catch this, because the UCS-2 output might be \
1269 interpreted as UTF-16 by other programs. If we let \
1270 surrogates pass through, attackers could make a security \
1271 hole exploit by synthesizing any desired plane 1-16 \
1273 if (! ignore_errors_p ()) \
1275 result = __GCONV_ILLEGAL_INPUT; \
1284 *((uint16_t *) outptr)++ = bswap_16 (val); \
1288 #define LOOP_NEED_FLAGS
1289 #include <iconv/loop.c>
1290 #include <iconv/skeleton.c>