ICU 71.1: utf8.h Source File

00001 // © 2016 and later: Unicode, Inc. and others.
00002 // License & terms of use: http://www.unicode.org/copyright.html
00003 /*
00004 *******************************************************************************
00005 *
00006 *   Copyright (C) 1999-2015, International Business Machines
00007 *   Corporation and others.  All Rights Reserved.
00008 *
00009 *******************************************************************************
00010 *   file name:  utf8.h
00011 *   encoding:   UTF-8
00012 *   tab size:   8 (not used)
00013 *   indentation:4
00014 *
00015 *   created on: 1999sep13
00016 *   created by: Markus W. Scherer
00017 */
00018 
00034 #ifndef __UTF8_H__
00035 #define __UTF8_H__
00036 
00037 #include <stdbool.h>
00038 #include "unicode/umachine.h"
00039 #ifndef __UTF_H__
00040 #   include "unicode/utf.h"
00041 #endif
00042 
00043 /* internal definitions ----------------------------------------------------- */
00044 
00056 #define U8_COUNT_TRAIL_BYTES(leadByte) \
00057     (U8_IS_LEAD(leadByte) ? \
00058         ((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0)+1 : 0)
00059 
00071 #define U8_COUNT_TRAIL_BYTES_UNSAFE(leadByte) \
00072     (((uint8_t)(leadByte)>=0xc2)+((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0))
00073 
00081 #define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
00082 
00091 #define U8_LEAD3_T1_BITS "\x20\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x10\x30\x30"
00092 
00098 #define U8_IS_VALID_LEAD3_AND_T1(lead, t1) (U8_LEAD3_T1_BITS[(lead)&0xf]&(1<<((uint8_t)(t1)>>5)))
00099 
00108 #define U8_LEAD4_T1_BITS "\x00\x00\x00\x00\x00\x00\x00\x00\x1E\x0F\x0F\x0F\x00\x00\x00\x00"
00109 
00115 #define U8_IS_VALID_LEAD4_AND_T1(lead, t1) (U8_LEAD4_T1_BITS[(uint8_t)(t1)>>4]&(1<<((lead)&7)))
00116 
00126 U_CAPI UChar32 U_EXPORT2
00127 utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
00128 
00138 U_CAPI int32_t U_EXPORT2
00139 utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
00140 
00150 U_CAPI UChar32 U_EXPORT2
00151 utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
00152 
00162 U_CAPI int32_t U_EXPORT2
00163 utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
00164 
00165 /* single-code point definitions -------------------------------------------- */
00166 
00173 #define U8_IS_SINGLE(c) (((c)&0x80)==0)
00174 
00181 #define U8_IS_LEAD(c) ((uint8_t)((c)-0xc2)<=0x32)
00182 // 0x32=0xf4-0xc2
00183 
00190 #define U8_IS_TRAIL(c) ((int8_t)(c)<-0x40)
00191 
00199 #define U8_LENGTH(c) \
00200     ((uint32_t)(c)<=0x7f ? 1 : \
00201         ((uint32_t)(c)<=0x7ff ? 2 : \
00202             ((uint32_t)(c)<=0xd7ff ? 3 : \
00203                 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
00204                     ((uint32_t)(c)<=0xffff ? 3 : 4)\
00205                 ) \
00206             ) \
00207         ) \
00208     )
00209 
00215 #define U8_MAX_LENGTH 4
00216 
00233 #define U8_GET_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00234     int32_t _u8_get_unsafe_index=(int32_t)(i); \
00235     U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
00236     U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
00237 } UPRV_BLOCK_MACRO_END
00238 
00260 #define U8_GET(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
00261     int32_t _u8_get_index=(i); \
00262     U8_SET_CP_START(s, start, _u8_get_index); \
00263     U8_NEXT(s, _u8_get_index, length, c); \
00264 } UPRV_BLOCK_MACRO_END
00265 
00291 #define U8_GET_OR_FFFD(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
00292     int32_t _u8_get_index=(i); \
00293     U8_SET_CP_START(s, start, _u8_get_index); \
00294     U8_NEXT_OR_FFFD(s, _u8_get_index, length, c); \
00295 } UPRV_BLOCK_MACRO_END
00296 
00297 /* definitions with forward iteration --------------------------------------- */
00298 
00316 #define U8_NEXT_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00317     (c)=(uint8_t)(s)[(i)++]; \
00318     if(!U8_IS_SINGLE(c)) { \
00319         if((c)<0xe0) { \
00320             (c)=(((c)&0x1f)<<6)|((s)[(i)++]&0x3f); \
00321         } else if((c)<0xf0) { \
00322             /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
00323             (c)=(UChar)(((c)<<12)|(((s)[i]&0x3f)<<6)|((s)[(i)+1]&0x3f)); \
00324             (i)+=2; \
00325         } else { \
00326             (c)=(((c)&7)<<18)|(((s)[i]&0x3f)<<12)|(((s)[(i)+1]&0x3f)<<6)|((s)[(i)+2]&0x3f); \
00327             (i)+=3; \
00328         } \
00329     } \
00330 } UPRV_BLOCK_MACRO_END
00331 
00352 #define U8_NEXT(s, i, length, c) U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, U_SENTINEL)
00353 
00378 #define U8_NEXT_OR_FFFD(s, i, length, c) U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, 0xfffd)
00379 
00381 #define U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, sub) UPRV_BLOCK_MACRO_BEGIN { \
00382     (c)=(uint8_t)(s)[(i)++]; \
00383     if(!U8_IS_SINGLE(c)) { \
00384         uint8_t __t = 0; \
00385         if((i)!=(length) && \
00386             /* fetch/validate/assemble all but last trail byte */ \
00387             ((c)>=0xe0 ? \
00388                 ((c)<0xf0 ?  /* U+0800..U+FFFF except surrogates */ \
00389                     U8_LEAD3_T1_BITS[(c)&=0xf]&(1<<((__t=(s)[i])>>5)) && \
00390                     (__t&=0x3f, 1) \
00391                 :  /* U+10000..U+10FFFF */ \
00392                     ((c)-=0xf0)<=4 && \
00393                     U8_LEAD4_T1_BITS[(__t=(s)[i])>>4]&(1<<(c)) && \
00394                     ((c)=((c)<<6)|(__t&0x3f), ++(i)!=(length)) && \
00395                     (__t=(s)[i]-0x80)<=0x3f) && \
00396                 /* valid second-to-last trail byte */ \
00397                 ((c)=((c)<<6)|__t, ++(i)!=(length)) \
00398             :  /* U+0080..U+07FF */ \
00399                 (c)>=0xc2 && ((c)&=0x1f, 1)) && \
00400             /* last trail byte */ \
00401             (__t=(s)[i]-0x80)<=0x3f && \
00402             ((c)=((c)<<6)|__t, ++(i), 1)) { \
00403         } else { \
00404             (c)=(sub);  /* ill-formed*/ \
00405         } \
00406     } \
00407 } UPRV_BLOCK_MACRO_END
00408 
00422 #define U8_APPEND_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00423     uint32_t __uc=(c); \
00424     if(__uc<=0x7f) { \
00425         (s)[(i)++]=(uint8_t)__uc; \
00426     } else { \
00427         if(__uc<=0x7ff) { \
00428             (s)[(i)++]=(uint8_t)((__uc>>6)|0xc0); \
00429         } else { \
00430             if(__uc<=0xffff) { \
00431                 (s)[(i)++]=(uint8_t)((__uc>>12)|0xe0); \
00432             } else { \
00433                 (s)[(i)++]=(uint8_t)((__uc>>18)|0xf0); \
00434                 (s)[(i)++]=(uint8_t)(((__uc>>12)&0x3f)|0x80); \
00435             } \
00436             (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
00437         } \
00438         (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
00439     } \
00440 } UPRV_BLOCK_MACRO_END
00441 
00459 #define U8_APPEND(s, i, capacity, c, isError) UPRV_BLOCK_MACRO_BEGIN { \
00460     uint32_t __uc=(c); \
00461     if(__uc<=0x7f) { \
00462         (s)[(i)++]=(uint8_t)__uc; \
00463     } else if(__uc<=0x7ff && (i)+1<(capacity)) { \
00464         (s)[(i)++]=(uint8_t)((__uc>>6)|0xc0); \
00465         (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
00466     } else if((__uc<=0xd7ff || (0xe000<=__uc && __uc<=0xffff)) && (i)+2<(capacity)) { \
00467         (s)[(i)++]=(uint8_t)((__uc>>12)|0xe0); \
00468         (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
00469         (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
00470     } else if(0xffff<__uc && __uc<=0x10ffff && (i)+3<(capacity)) { \
00471         (s)[(i)++]=(uint8_t)((__uc>>18)|0xf0); \
00472         (s)[(i)++]=(uint8_t)(((__uc>>12)&0x3f)|0x80); \
00473         (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
00474         (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
00475     } else { \
00476         (isError)=true; \
00477     } \
00478 } UPRV_BLOCK_MACRO_END
00479 
00490 #define U8_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
00491     (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((s)[i]); \
00492 } UPRV_BLOCK_MACRO_END
00493 
00507 #define U8_FWD_1(s, i, length) UPRV_BLOCK_MACRO_BEGIN { \
00508     uint8_t __b=(s)[(i)++]; \
00509     if(U8_IS_LEAD(__b) && (i)!=(length)) { \
00510         uint8_t __t1=(s)[i]; \
00511         if((0xe0<=__b && __b<0xf0)) { \
00512             if(U8_IS_VALID_LEAD3_AND_T1(__b, __t1) && \
00513                     ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
00514                 ++(i); \
00515             } \
00516         } else if(__b<0xe0) { \
00517             if(U8_IS_TRAIL(__t1)) { \
00518                 ++(i); \
00519             } \
00520         } else /* c>=0xf0 */ { \
00521             if(U8_IS_VALID_LEAD4_AND_T1(__b, __t1) && \
00522                     ++(i)!=(length) && U8_IS_TRAIL((s)[i]) && \
00523                     ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
00524                 ++(i); \
00525             } \
00526         } \
00527     } \
00528 } UPRV_BLOCK_MACRO_END
00529 
00542 #define U8_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
00543     int32_t __N=(n); \
00544     while(__N>0) { \
00545         U8_FWD_1_UNSAFE(s, i); \
00546         --__N; \
00547     } \
00548 } UPRV_BLOCK_MACRO_END
00549 
00565 #define U8_FWD_N(s, i, length, n) UPRV_BLOCK_MACRO_BEGIN { \
00566     int32_t __N=(n); \
00567     while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
00568         U8_FWD_1(s, i, length); \
00569         --__N; \
00570     } \
00571 } UPRV_BLOCK_MACRO_END
00572 
00586 #define U8_SET_CP_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
00587     while(U8_IS_TRAIL((s)[i])) { --(i); } \
00588 } UPRV_BLOCK_MACRO_END
00589 
00607 #define U8_SET_CP_START(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
00608     if(U8_IS_TRAIL((s)[(i)])) { \
00609         (i)=utf8_back1SafeBody(s, start, (i)); \
00610     } \
00611 } UPRV_BLOCK_MACRO_END
00612 
00639 #define U8_TRUNCATE_IF_INCOMPLETE(s, start, length) UPRV_BLOCK_MACRO_BEGIN { \
00640     if((length)>(start)) { \
00641         uint8_t __b1=s[(length)-1]; \
00642         if(U8_IS_SINGLE(__b1)) { \
00643             /* common ASCII character */ \
00644         } else if(U8_IS_LEAD(__b1)) { \
00645             --(length); \
00646         } else if(U8_IS_TRAIL(__b1) && ((length)-2)>=(start)) { \
00647             uint8_t __b2=s[(length)-2]; \
00648             if(0xe0<=__b2 && __b2<=0xf4) { \
00649                 if(__b2<0xf0 ? U8_IS_VALID_LEAD3_AND_T1(__b2, __b1) : \
00650                         U8_IS_VALID_LEAD4_AND_T1(__b2, __b1)) { \
00651                     (length)-=2; \
00652                 } \
00653             } else if(U8_IS_TRAIL(__b2) && ((length)-3)>=(start)) { \
00654                 uint8_t __b3=s[(length)-3]; \
00655                 if(0xf0<=__b3 && __b3<=0xf4 && U8_IS_VALID_LEAD4_AND_T1(__b3, __b2)) { \
00656                     (length)-=3; \
00657                 } \
00658             } \
00659         } \
00660     } \
00661 } UPRV_BLOCK_MACRO_END
00662 
00663 /* definitions with backward iteration -------------------------------------- */
00664 
00684 #define U8_PREV_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00685     (c)=(uint8_t)(s)[--(i)]; \
00686     if(U8_IS_TRAIL(c)) { \
00687         uint8_t __b, __count=1, __shift=6; \
00688 \
00689         /* c is a trail byte */ \
00690         (c)&=0x3f; \
00691         for(;;) { \
00692             __b=(s)[--(i)]; \
00693             if(__b>=0xc0) { \
00694                 U8_MASK_LEAD_BYTE(__b, __count); \
00695                 (c)|=(UChar32)__b<<__shift; \
00696                 break; \
00697             } else { \
00698                 (c)|=(UChar32)(__b&0x3f)<<__shift; \
00699                 ++__count; \
00700                 __shift+=6; \
00701             } \
00702         } \
00703     } \
00704 } UPRV_BLOCK_MACRO_END
00705 
00726 #define U8_PREV(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00727     (c)=(uint8_t)(s)[--(i)]; \
00728     if(!U8_IS_SINGLE(c)) { \
00729         (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
00730     } \
00731 } UPRV_BLOCK_MACRO_END
00732 
00757 #define U8_PREV_OR_FFFD(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00758     (c)=(uint8_t)(s)[--(i)]; \
00759     if(!U8_IS_SINGLE(c)) { \
00760         (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -3); \
00761     } \
00762 } UPRV_BLOCK_MACRO_END
00763 
00775 #define U8_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
00776     while(U8_IS_TRAIL((s)[--(i)])) {} \
00777 } UPRV_BLOCK_MACRO_END
00778 
00791 #define U8_BACK_1(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
00792     if(U8_IS_TRAIL((s)[--(i)])) { \
00793         (i)=utf8_back1SafeBody(s, start, (i)); \
00794     } \
00795 } UPRV_BLOCK_MACRO_END
00796 
00810 #define U8_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
00811     int32_t __N=(n); \
00812     while(__N>0) { \
00813         U8_BACK_1_UNSAFE(s, i); \
00814         --__N; \
00815     } \
00816 } UPRV_BLOCK_MACRO_END
00817 
00832 #define U8_BACK_N(s, start, i, n) UPRV_BLOCK_MACRO_BEGIN { \
00833     int32_t __N=(n); \
00834     while(__N>0 && (i)>(start)) { \
00835         U8_BACK_1(s, start, i); \
00836         --__N; \
00837     } \
00838 } UPRV_BLOCK_MACRO_END
00839 
00853 #define U8_SET_CP_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
00854     U8_BACK_1_UNSAFE(s, i); \
00855     U8_FWD_1_UNSAFE(s, i); \
00856 } UPRV_BLOCK_MACRO_END
00857 
00875 #define U8_SET_CP_LIMIT(s, start, i, length) UPRV_BLOCK_MACRO_BEGIN { \
00876     if((start)<(i) && ((i)<(length) || (length)<0)) { \
00877         U8_BACK_1(s, start, i); \
00878         U8_FWD_1(s, i, length); \
00879     } \
00880 } UPRV_BLOCK_MACRO_END
00881 
00882 #endif