145 #ifndef __UTF_OLD_H__
146 #define __UTF_OLD_H__
159 #ifndef U_HIDE_OBSOLETE_UTF_OLD_H
160 # define U_HIDE_OBSOLETE_UTF_OLD_H 0
163 #if !defined(U_HIDE_DEPRECATED_API) && !U_HIDE_OBSOLETE_UTF_OLD_H
171 #ifdef U_USE_UTF_DEPRECATES
179 typedef int32_t UTextOffset;
211 #define UTF8_ERROR_VALUE_1 0x15
218 #define UTF8_ERROR_VALUE_2 0x9f
226 #define UTF_ERROR_VALUE 0xffff
234 #define UTF_IS_ERROR(c) \
235 (((c)&0xfffe)==0xfffe || (c)==UTF8_ERROR_VALUE_1 || (c)==UTF8_ERROR_VALUE_2)
242 #define UTF_IS_VALID(c) \
243 (UTF_IS_UNICODE_CHAR(c) && \
244 (c)!=UTF8_ERROR_VALUE_1 && (c)!=UTF8_ERROR_VALUE_2)
250 #define UTF_IS_SURROGATE(uchar) (((uchar)&0xfffff800)==0xd800)
257 #define UTF_IS_UNICODE_NONCHAR(c) \
259 ((uint32_t)(c)<=0xfdef || ((c)&0xfffe)==0xfffe) && \
260 (uint32_t)(c)<=0x10ffff)
277 #define UTF_IS_UNICODE_CHAR(c) \
278 ((uint32_t)(c)<0xd800 || \
279 ((uint32_t)(c)>0xdfff && \
280 (uint32_t)(c)<=0x10ffff && \
281 !UTF_IS_UNICODE_NONCHAR(c)))
298 #elif defined(U_STATIC_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION)
308 #define UTF8_COUNT_TRAIL_BYTES(leadByte) (utf8_countTrailBytes[(uint8_t)leadByte])
314 #define UTF8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
317 #define UTF8_IS_SINGLE(uchar) (((uchar)&0x80)==0)
319 #define UTF8_IS_LEAD(uchar) ((uint8_t)((uchar)-0xc0)<0x3e)
321 #define UTF8_IS_TRAIL(uchar) (((uchar)&0xc0)==0x80)
324 #define UTF8_NEED_MULTIPLE_UCHAR(c) ((uint32_t)(c)>0x7f)
340 # define UTF8_CHAR_LENGTH(c) \
341 ((uint32_t)(c)<=0x7f ? 1 : \
342 ((uint32_t)(c)<=0x7ff ? 2 : \
343 ((uint32_t)((c)-0x10000)>0xfffff ? 3 : 4) \
347 # define UTF8_CHAR_LENGTH(c) \
348 ((uint32_t)(c)<=0x7f ? 1 : \
349 ((uint32_t)(c)<=0x7ff ? 2 : \
350 ((uint32_t)(c)<=0xffff ? 3 : \
351 ((uint32_t)(c)<=0x10ffff ? 4 : \
352 ((uint32_t)(c)<=0x3ffffff ? 5 : \
353 ((uint32_t)(c)<=0x7fffffff ? 6 : 3) \
362 #define UTF8_MAX_CHAR_LENGTH 4
365 #define UTF8_ARRAY_SIZE(size) ((5*(size))/2)
368 #define UTF8_GET_CHAR_UNSAFE(s, i, c) { \
369 int32_t _utf8_get_char_unsafe_index=(int32_t)(i); \
370 UTF8_SET_CHAR_START_UNSAFE(s, _utf8_get_char_unsafe_index); \
371 UTF8_NEXT_CHAR_UNSAFE(s, _utf8_get_char_unsafe_index, c); \
375 #define UTF8_GET_CHAR_SAFE(s, start, i, length, c, strict) { \
376 int32_t _utf8_get_char_safe_index=(int32_t)(i); \
377 UTF8_SET_CHAR_START_SAFE(s, start, _utf8_get_char_safe_index); \
378 UTF8_NEXT_CHAR_SAFE(s, _utf8_get_char_safe_index, length, c, strict); \
382 #define UTF8_NEXT_CHAR_UNSAFE(s, i, c) { \
384 if((uint8_t)((c)-0xc0)<0x35) { \
385 uint8_t __count=UTF8_COUNT_TRAIL_BYTES(c); \
386 UTF8_MASK_LEAD_BYTE(c, __count); \
390 (c)=((c)<<6)|((s)[(i)++]&0x3f); \
392 (c)=((c)<<6)|((s)[(i)++]&0x3f); \
394 (c)=((c)<<6)|((s)[(i)++]&0x3f); \
402 #define UTF8_APPEND_CHAR_UNSAFE(s, i, c) { \
403 if((uint32_t)(c)<=0x7f) { \
404 (s)[(i)++]=(uint8_t)(c); \
406 if((uint32_t)(c)<=0x7ff) { \
407 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
409 if((uint32_t)(c)<=0xffff) { \
410 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
412 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
413 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
415 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
417 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
422 #define UTF8_FWD_1_UNSAFE(s, i) { \
423 (i)+=1+UTF8_COUNT_TRAIL_BYTES((s)[i]); \
427 #define UTF8_FWD_N_UNSAFE(s, i, n) { \
430 UTF8_FWD_1_UNSAFE(s, i); \
436 #define UTF8_SET_CHAR_START_UNSAFE(s, i) { \
437 while(UTF8_IS_TRAIL((s)[i])) { --(i); } \
441 #define UTF8_NEXT_CHAR_SAFE(s, i, length, c, strict) { \
444 if(UTF8_IS_LEAD(c)) { \
445 (c)=utf8_nextCharSafeBody(s, &(i), (int32_t)(length), c, strict); \
447 (c)=UTF8_ERROR_VALUE_1; \
453 #define UTF8_APPEND_CHAR_SAFE(s, i, length, c) { \
454 if((uint32_t)(c)<=0x7f) { \
455 (s)[(i)++]=(uint8_t)(c); \
457 (i)=utf8_appendCharSafeBody(s, (int32_t)(i), (int32_t)(length), c, NULL); \
462 #define UTF8_FWD_1_SAFE(s, i, length) U8_FWD_1(s, i, length)
465 #define UTF8_FWD_N_SAFE(s, i, length, n) U8_FWD_N(s, i, length, n)
468 #define UTF8_SET_CHAR_START_SAFE(s, start, i) U8_SET_CP_START(s, start, i)
471 #define UTF8_PREV_CHAR_UNSAFE(s, i, c) { \
473 if(UTF8_IS_TRAIL(c)) { \
474 uint8_t __b, __count=1, __shift=6; \
481 UTF8_MASK_LEAD_BYTE(__b, __count); \
482 (c)|=(UChar32)__b<<__shift; \
485 (c)|=(UChar32)(__b&0x3f)<<__shift; \
494 #define UTF8_BACK_1_UNSAFE(s, i) { \
495 while(UTF8_IS_TRAIL((s)[--(i)])) {} \
499 #define UTF8_BACK_N_UNSAFE(s, i, n) { \
502 UTF8_BACK_1_UNSAFE(s, i); \
508 #define UTF8_SET_CHAR_LIMIT_UNSAFE(s, i) { \
509 UTF8_BACK_1_UNSAFE(s, i); \
510 UTF8_FWD_1_UNSAFE(s, i); \
514 #define UTF8_PREV_CHAR_SAFE(s, start, i, c, strict) { \
518 (c)=utf8_prevCharSafeBody(s, start, &(i), c, strict); \
520 (c)=UTF8_ERROR_VALUE_1; \
526 #define UTF8_BACK_1_SAFE(s, start, i) U8_BACK_1(s, start, i)
529 #define UTF8_BACK_N_SAFE(s, start, i, n) U8_BACK_N(s, start, i, n)
532 #define UTF8_SET_CHAR_LIMIT_SAFE(s, start, i, length) U8_SET_CP_LIMIT(s, start, i, length)
537 #define UTF_IS_FIRST_SURROGATE(uchar) (((uchar)&0xfffffc00)==0xd800)
540 #define UTF_IS_SECOND_SURROGATE(uchar) (((uchar)&0xfffffc00)==0xdc00)
543 #define UTF_IS_SURROGATE_FIRST(c) (((c)&0x400)==0)
546 #define UTF_SURROGATE_OFFSET ((0xd800<<10UL)+0xdc00-0x10000)
549 #define UTF16_GET_PAIR_VALUE(first, second) \
550 (((first)<<10UL)+(second)-UTF_SURROGATE_OFFSET)
553 #define UTF_FIRST_SURROGATE(supplementary) (UChar)(((supplementary)>>10)+0xd7c0)
556 #define UTF_SECOND_SURROGATE(supplementary) (UChar)(((supplementary)&0x3ff)|0xdc00)
559 #define UTF16_LEAD(supplementary) UTF_FIRST_SURROGATE(supplementary)
562 #define UTF16_TRAIL(supplementary) UTF_SECOND_SURROGATE(supplementary)
565 #define UTF16_IS_SINGLE(uchar) !UTF_IS_SURROGATE(uchar)
568 #define UTF16_IS_LEAD(uchar) UTF_IS_FIRST_SURROGATE(uchar)
571 #define UTF16_IS_TRAIL(uchar) UTF_IS_SECOND_SURROGATE(uchar)
574 #define UTF16_NEED_MULTIPLE_UCHAR(c) ((uint32_t)(c)>0xffff)
577 #define UTF16_CHAR_LENGTH(c) ((uint32_t)(c)<=0xffff ? 1 : 2)
580 #define UTF16_MAX_CHAR_LENGTH 2
583 #define UTF16_ARRAY_SIZE(size) (size)
596 #define UTF16_GET_CHAR_UNSAFE(s, i, c) { \
598 if(UTF_IS_SURROGATE(c)) { \
599 if(UTF_IS_SURROGATE_FIRST(c)) { \
600 (c)=UTF16_GET_PAIR_VALUE((c), (s)[(i)+1]); \
602 (c)=UTF16_GET_PAIR_VALUE((s)[(i)-1], (c)); \
608 #define UTF16_GET_CHAR_SAFE(s, start, i, length, c, strict) { \
610 if(UTF_IS_SURROGATE(c)) { \
612 if(UTF_IS_SURROGATE_FIRST(c)) { \
613 if((i)+1<(length) && UTF_IS_SECOND_SURROGATE(__c2=(s)[(i)+1])) { \
614 (c)=UTF16_GET_PAIR_VALUE((c), __c2); \
618 (c)=UTF_ERROR_VALUE; \
621 if((i)-1>=(start) && UTF_IS_FIRST_SURROGATE(__c2=(s)[(i)-1])) { \
622 (c)=UTF16_GET_PAIR_VALUE(__c2, (c)); \
626 (c)=UTF_ERROR_VALUE; \
629 } else if((strict) && !UTF_IS_UNICODE_CHAR(c)) { \
630 (c)=UTF_ERROR_VALUE; \
635 #define UTF16_NEXT_CHAR_UNSAFE(s, i, c) { \
637 if(UTF_IS_FIRST_SURROGATE(c)) { \
638 (c)=UTF16_GET_PAIR_VALUE((c), (s)[(i)++]); \
643 #define UTF16_APPEND_CHAR_UNSAFE(s, i, c) { \
644 if((uint32_t)(c)<=0xffff) { \
645 (s)[(i)++]=(uint16_t)(c); \
647 (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
648 (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
653 #define UTF16_FWD_1_UNSAFE(s, i) { \
654 if(UTF_IS_FIRST_SURROGATE((s)[(i)++])) { \
660 #define UTF16_FWD_N_UNSAFE(s, i, n) { \
663 UTF16_FWD_1_UNSAFE(s, i); \
669 #define UTF16_SET_CHAR_START_UNSAFE(s, i) { \
670 if(UTF_IS_SECOND_SURROGATE((s)[i])) { \
676 #define UTF16_NEXT_CHAR_SAFE(s, i, length, c, strict) { \
678 if(UTF_IS_FIRST_SURROGATE(c)) { \
680 if((i)<(length) && UTF_IS_SECOND_SURROGATE(__c2=(s)[(i)])) { \
682 (c)=UTF16_GET_PAIR_VALUE((c), __c2); \
686 (c)=UTF_ERROR_VALUE; \
688 } else if((strict) && !UTF_IS_UNICODE_CHAR(c)) { \
690 (c)=UTF_ERROR_VALUE; \
695 #define UTF16_APPEND_CHAR_SAFE(s, i, length, c) { \
696 if((uint32_t)(c)<=0xffff) { \
697 (s)[(i)++]=(uint16_t)(c); \
698 } else if((uint32_t)(c)<=0x10ffff) { \
699 if((i)+1<(length)) { \
700 (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
701 (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
703 (s)[(i)++]=UTF_ERROR_VALUE; \
706 (s)[(i)++]=UTF_ERROR_VALUE; \
711 #define UTF16_FWD_1_SAFE(s, i, length) U16_FWD_1(s, i, length)
714 #define UTF16_FWD_N_SAFE(s, i, length, n) U16_FWD_N(s, i, length, n)
717 #define UTF16_SET_CHAR_START_SAFE(s, start, i) U16_SET_CP_START(s, start, i)
720 #define UTF16_PREV_CHAR_UNSAFE(s, i, c) { \
722 if(UTF_IS_SECOND_SURROGATE(c)) { \
723 (c)=UTF16_GET_PAIR_VALUE((s)[--(i)], (c)); \
728 #define UTF16_BACK_1_UNSAFE(s, i) { \
729 if(UTF_IS_SECOND_SURROGATE((s)[--(i)])) { \
735 #define UTF16_BACK_N_UNSAFE(s, i, n) { \
738 UTF16_BACK_1_UNSAFE(s, i); \
744 #define UTF16_SET_CHAR_LIMIT_UNSAFE(s, i) { \
745 if(UTF_IS_FIRST_SURROGATE((s)[(i)-1])) { \
751 #define UTF16_PREV_CHAR_SAFE(s, start, i, c, strict) { \
753 if(UTF_IS_SECOND_SURROGATE(c)) { \
755 if((i)>(start) && UTF_IS_FIRST_SURROGATE(__c2=(s)[(i)-1])) { \
757 (c)=UTF16_GET_PAIR_VALUE(__c2, (c)); \
761 (c)=UTF_ERROR_VALUE; \
763 } else if((strict) && !UTF_IS_UNICODE_CHAR(c)) { \
765 (c)=UTF_ERROR_VALUE; \
770 #define UTF16_BACK_1_SAFE(s, start, i) U16_BACK_1(s, start, i)
773 #define UTF16_BACK_N_SAFE(s, start, i, n) U16_BACK_N(s, start, i, n)
776 #define UTF16_SET_CHAR_LIMIT_SAFE(s, start, i, length) U16_SET_CP_LIMIT(s, start, i, length)
796 #define UTF32_IS_SAFE(c, strict) \
798 (uint32_t)(c)<=0x10ffff : \
799 UTF_IS_UNICODE_CHAR(c))
812 #define UTF32_IS_SINGLE(uchar) 1
814 #define UTF32_IS_LEAD(uchar) 0
816 #define UTF32_IS_TRAIL(uchar) 0
821 #define UTF32_NEED_MULTIPLE_UCHAR(c) 0
823 #define UTF32_CHAR_LENGTH(c) 1
825 #define UTF32_MAX_CHAR_LENGTH 1
830 #define UTF32_ARRAY_SIZE(size) (size)
833 #define UTF32_GET_CHAR_UNSAFE(s, i, c) { \
838 #define UTF32_GET_CHAR_SAFE(s, start, i, length, c, strict) { \
840 if(!UTF32_IS_SAFE(c, strict)) { \
841 (c)=UTF_ERROR_VALUE; \
848 #define UTF32_NEXT_CHAR_UNSAFE(s, i, c) { \
853 #define UTF32_APPEND_CHAR_UNSAFE(s, i, c) { \
858 #define UTF32_FWD_1_UNSAFE(s, i) { \
863 #define UTF32_FWD_N_UNSAFE(s, i, n) { \
868 #define UTF32_SET_CHAR_START_UNSAFE(s, i) { \
872 #define UTF32_NEXT_CHAR_SAFE(s, i, length, c, strict) { \
874 if(!UTF32_IS_SAFE(c, strict)) { \
875 (c)=UTF_ERROR_VALUE; \
880 #define UTF32_APPEND_CHAR_SAFE(s, i, length, c) { \
881 if((uint32_t)(c)<=0x10ffff) { \
889 #define UTF32_FWD_1_SAFE(s, i, length) { \
894 #define UTF32_FWD_N_SAFE(s, i, length, n) { \
895 if(((i)+=(n))>(length)) { \
901 #define UTF32_SET_CHAR_START_SAFE(s, start, i) { \
907 #define UTF32_PREV_CHAR_UNSAFE(s, i, c) { \
912 #define UTF32_BACK_1_UNSAFE(s, i) { \
917 #define UTF32_BACK_N_UNSAFE(s, i, n) { \
922 #define UTF32_SET_CHAR_LIMIT_UNSAFE(s, i) { \
926 #define UTF32_PREV_CHAR_SAFE(s, start, i, c, strict) { \
928 if(!UTF32_IS_SAFE(c, strict)) { \
929 (c)=UTF_ERROR_VALUE; \
934 #define UTF32_BACK_1_SAFE(s, start, i) { \
939 #define UTF32_BACK_N_SAFE(s, start, i, n) { \
947 #define UTF32_SET_CHAR_LIMIT_SAFE(s, i, length) { \
957 #define UTF_ARRAY_SIZE(size) UTF16_ARRAY_SIZE(size)
960 #define UTF_GET_CHAR_UNSAFE(s, i, c) UTF16_GET_CHAR_UNSAFE(s, i, c)
963 #define UTF_GET_CHAR_SAFE(s, start, i, length, c, strict) UTF16_GET_CHAR_SAFE(s, start, i, length, c, strict)
967 #define UTF_NEXT_CHAR_UNSAFE(s, i, c) UTF16_NEXT_CHAR_UNSAFE(s, i, c)
970 #define UTF_NEXT_CHAR_SAFE(s, i, length, c, strict) UTF16_NEXT_CHAR_SAFE(s, i, length, c, strict)
974 #define UTF_APPEND_CHAR_UNSAFE(s, i, c) UTF16_APPEND_CHAR_UNSAFE(s, i, c)
977 #define UTF_APPEND_CHAR_SAFE(s, i, length, c) UTF16_APPEND_CHAR_SAFE(s, i, length, c)
981 #define UTF_FWD_1_UNSAFE(s, i) UTF16_FWD_1_UNSAFE(s, i)
984 #define UTF_FWD_1_SAFE(s, i, length) UTF16_FWD_1_SAFE(s, i, length)
988 #define UTF_FWD_N_UNSAFE(s, i, n) UTF16_FWD_N_UNSAFE(s, i, n)
991 #define UTF_FWD_N_SAFE(s, i, length, n) UTF16_FWD_N_SAFE(s, i, length, n)
995 #define UTF_SET_CHAR_START_UNSAFE(s, i) UTF16_SET_CHAR_START_UNSAFE(s, i)
998 #define UTF_SET_CHAR_START_SAFE(s, start, i) UTF16_SET_CHAR_START_SAFE(s, start, i)
1002 #define UTF_PREV_CHAR_UNSAFE(s, i, c) UTF16_PREV_CHAR_UNSAFE(s, i, c)
1005 #define UTF_PREV_CHAR_SAFE(s, start, i, c, strict) UTF16_PREV_CHAR_SAFE(s, start, i, c, strict)
1009 #define UTF_BACK_1_UNSAFE(s, i) UTF16_BACK_1_UNSAFE(s, i)
1012 #define UTF_BACK_1_SAFE(s, start, i) UTF16_BACK_1_SAFE(s, start, i)
1016 #define UTF_BACK_N_UNSAFE(s, i, n) UTF16_BACK_N_UNSAFE(s, i, n)
1019 #define UTF_BACK_N_SAFE(s, start, i, n) UTF16_BACK_N_SAFE(s, start, i, n)
1023 #define UTF_SET_CHAR_LIMIT_UNSAFE(s, i) UTF16_SET_CHAR_LIMIT_UNSAFE(s, i)
1026 #define UTF_SET_CHAR_LIMIT_SAFE(s, start, i, length) UTF16_SET_CHAR_LIMIT_SAFE(s, start, i, length)
1035 #define UTF_IS_SINGLE(uchar) U16_IS_SINGLE(uchar)
1042 #define UTF_IS_LEAD(uchar) U16_IS_LEAD(uchar)
1049 #define UTF_IS_TRAIL(uchar) U16_IS_TRAIL(uchar)
1056 #define UTF_NEED_MULTIPLE_UCHAR(c) UTF16_NEED_MULTIPLE_UCHAR(c)
1063 #define UTF_CHAR_LENGTH(c) U16_LENGTH(c)
1070 #define UTF_MAX_CHAR_LENGTH U16_MAX_LENGTH
1081 #define UTF_GET_CHAR(s, start, i, length, c) U16_GET(s, start, i, length, c)
1094 #define UTF_NEXT_CHAR(s, i, length, c) U16_NEXT(s, i, length, c)
1107 #define UTF_APPEND_CHAR(s, i, length, c) UTF16_APPEND_CHAR_SAFE(s, i, length, c)
1118 #define UTF_FWD_1(s, i, length) U16_FWD_1(s, i, length)
1129 #define UTF_FWD_N(s, i, length, n) U16_FWD_N(s, i, length, n)
1145 #define UTF_SET_CHAR_START(s, start, i) U16_SET_CP_START(s, start, i)
1158 #define UTF_PREV_CHAR(s, start, i, c) U16_PREV(s, start, i, c)
1171 #define UTF_BACK_1(s, start, i) U16_BACK_1(s, start, i)
1184 #define UTF_BACK_N(s, start, i, n) U16_BACK_N(s, start, i, n)
1200 #define UTF_SET_CHAR_LIMIT(s, start, i, length) U16_SET_CP_LIMIT(s, start, i, length)
1202 #endif // !U_HIDE_DEPRECATED_API && !U_HIDE_OBSOLETE_UTF_OLD_H
C API: 8-bit Unicode handling macros.
U_CFUNC U_IMPORT const uint8_t utf8_countTrailBytes[]
Internal array with numbers of trail bytes for any given byte used in lead byte position.
C API: Code point macros.
#define U_CFUNC
This is used in a declaration of a library private ICU C function.
C API: 16-bit Unicode handling macros.