ICU 60.1: 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 "unicode/umachine.h"
00038 #ifndef __UTF_H__
00039 #   include "unicode/utf.h"
00040 #endif
00041 
00042 /* internal definitions ----------------------------------------------------- */
00043 
00055 #define U8_COUNT_TRAIL_BYTES(leadByte) \
00056     (U8_IS_LEAD(leadByte) ? \
00057         ((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0)+1 : 0)
00058 
00070 #define U8_COUNT_TRAIL_BYTES_UNSAFE(leadByte) \
00071     (((uint8_t)(leadByte)>=0xc2)+((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0))
00072 
00080 #define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
00081 
00090 #define U8_LEAD3_T1_BITS "\x20\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x10\x30\x30"
00091 
00097 #define U8_IS_VALID_LEAD3_AND_T1(lead, t1) (U8_LEAD3_T1_BITS[(lead)&0xf]&(1<<((uint8_t)(t1)>>5)))
00098 
00107 #define U8_LEAD4_T1_BITS "\x00\x00\x00\x00\x00\x00\x00\x00\x1E\x0F\x0F\x0F\x00\x00\x00\x00"
00108 
00114 #define U8_IS_VALID_LEAD4_AND_T1(lead, t1) (U8_LEAD4_T1_BITS[(uint8_t)(t1)>>4]&(1<<((lead)&7)))
00115 
00125 U_STABLE UChar32 U_EXPORT2
00126 utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
00127 
00137 U_STABLE int32_t U_EXPORT2
00138 utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
00139 
00149 U_STABLE UChar32 U_EXPORT2
00150 utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
00151 
00161 U_STABLE int32_t U_EXPORT2
00162 utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
00163 
00164 /* single-code point definitions -------------------------------------------- */
00165 
00172 #define U8_IS_SINGLE(c) (((c)&0x80)==0)
00173 
00180 #define U8_IS_LEAD(c) ((uint8_t)((c)-0xc2)<=0x32)
00181 // 0x32=0xf4-0xc2
00182 
00189 #define U8_IS_TRAIL(c) ((int8_t)(c)<-0x40)
00190 
00198 #define U8_LENGTH(c) \
00199     ((uint32_t)(c)<=0x7f ? 1 : \
00200         ((uint32_t)(c)<=0x7ff ? 2 : \
00201             ((uint32_t)(c)<=0xd7ff ? 3 : \
00202                 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
00203                     ((uint32_t)(c)<=0xffff ? 3 : 4)\
00204                 ) \
00205             ) \
00206         ) \
00207     )
00208 
00214 #define U8_MAX_LENGTH 4
00215 
00232 #define U8_GET_UNSAFE(s, i, c) { \
00233     int32_t _u8_get_unsafe_index=(int32_t)(i); \
00234     U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
00235     U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
00236 }
00237 
00259 #define U8_GET(s, start, i, length, c) { \
00260     int32_t _u8_get_index=(i); \
00261     U8_SET_CP_START(s, start, _u8_get_index); \
00262     U8_NEXT(s, _u8_get_index, length, c); \
00263 }
00264 
00290 #define U8_GET_OR_FFFD(s, start, i, length, c) { \
00291     int32_t _u8_get_index=(i); \
00292     U8_SET_CP_START(s, start, _u8_get_index); \
00293     U8_NEXT_OR_FFFD(s, _u8_get_index, length, c); \
00294 }
00295 
00296 /* definitions with forward iteration --------------------------------------- */
00297 
00315 #define U8_NEXT_UNSAFE(s, i, c) { \
00316     (c)=(uint8_t)(s)[(i)++]; \
00317     if(!U8_IS_SINGLE(c)) { \
00318         if((c)<0xe0) { \
00319             (c)=(((c)&0x1f)<<6)|((s)[(i)++]&0x3f); \
00320         } else if((c)<0xf0) { \
00321             /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
00322             (c)=(UChar)(((c)<<12)|(((s)[i]&0x3f)<<6)|((s)[(i)+1]&0x3f)); \
00323             (i)+=2; \
00324         } else { \
00325             (c)=(((c)&7)<<18)|(((s)[i]&0x3f)<<12)|(((s)[(i)+1]&0x3f)<<6)|((s)[(i)+2]&0x3f); \
00326             (i)+=3; \
00327         } \
00328     } \
00329 }
00330 
00351 #define U8_NEXT(s, i, length, c) { \
00352     (c)=(uint8_t)(s)[(i)++]; \
00353     if(!U8_IS_SINGLE(c)) { \
00354         uint8_t __t1, __t2; \
00355         if( /* handle U+0800..U+FFFF inline */ \
00356                 (0xe0<=(c) && (c)<0xf0) && \
00357                 (((i)+1)<(length) || (length)<0) && \
00358                 U8_IS_VALID_LEAD3_AND_T1((c), __t1=(s)[i]) && \
00359                 (__t2=(s)[(i)+1]-0x80)<=0x3f) { \
00360             (c)=(((c)&0xf)<<12)|((__t1&0x3f)<<6)|__t2; \
00361             (i)+=2; \
00362         } else if( /* handle U+0080..U+07FF inline */ \
00363                 ((c)<0xe0 && (c)>=0xc2) && \
00364                 ((i)!=(length)) && \
00365                 (__t1=(s)[i]-0x80)<=0x3f) { \
00366             (c)=(((c)&0x1f)<<6)|__t1; \
00367             ++(i); \
00368         } else { \
00369             /* function call for "complicated" and error cases */ \
00370             (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -1); \
00371         } \
00372     } \
00373 }
00374 
00399 #define U8_NEXT_OR_FFFD(s, i, length, c) { \
00400     (c)=(uint8_t)(s)[(i)++]; \
00401     if(!U8_IS_SINGLE(c)) { \
00402         uint8_t __t1, __t2; \
00403         if( /* handle U+0800..U+FFFF inline */ \
00404                 (0xe0<=(c) && (c)<0xf0) && \
00405                 (((i)+1)<(length) || (length)<0) && \
00406                 U8_IS_VALID_LEAD3_AND_T1((c), __t1=(s)[i]) && \
00407                 (__t2=(s)[(i)+1]-0x80)<=0x3f) { \
00408             (c)=(((c)&0xf)<<12)|((__t1&0x3f)<<6)|__t2; \
00409             (i)+=2; \
00410         } else if( /* handle U+0080..U+07FF inline */ \
00411                 ((c)<0xe0 && (c)>=0xc2) && \
00412                 ((i)!=(length)) && \
00413                 (__t1=(s)[i]-0x80)<=0x3f) { \
00414             (c)=(((c)&0x1f)<<6)|__t1; \
00415             ++(i); \
00416         } else { \
00417             /* function call for "complicated" and error cases */ \
00418             (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -3); \
00419         } \
00420     } \
00421 }
00422 
00436 #define U8_APPEND_UNSAFE(s, i, c) { \
00437     if((uint32_t)(c)<=0x7f) { \
00438         (s)[(i)++]=(uint8_t)(c); \
00439     } else { \
00440         if((uint32_t)(c)<=0x7ff) { \
00441             (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
00442         } else { \
00443             if((uint32_t)(c)<=0xffff) { \
00444                 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
00445             } else { \
00446                 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
00447                 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
00448             } \
00449             (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
00450         } \
00451         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00452     } \
00453 }
00454 
00472 #define U8_APPEND(s, i, capacity, c, isError) { \
00473     if((uint32_t)(c)<=0x7f) { \
00474         (s)[(i)++]=(uint8_t)(c); \
00475     } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \
00476         (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
00477         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00478     } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \
00479         (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
00480         (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
00481         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00482     } else { \
00483         (i)=utf8_appendCharSafeBody(s, (i), (capacity), c, &(isError)); \
00484     } \
00485 }
00486 
00497 #define U8_FWD_1_UNSAFE(s, i) { \
00498     (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((s)[i]); \
00499 }
00500 
00514 #define U8_FWD_1(s, i, length) { \
00515     uint8_t __b=(s)[(i)++]; \
00516     if(U8_IS_LEAD(__b) && (i)!=(length)) { \
00517         uint8_t __t1=(s)[i]; \
00518         if((0xe0<=__b && __b<0xf0)) { \
00519             if(U8_IS_VALID_LEAD3_AND_T1(__b, __t1) && \
00520                     ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
00521                 ++(i); \
00522             } \
00523         } else if(__b<0xe0) { \
00524             if(U8_IS_TRAIL(__t1)) { \
00525                 ++(i); \
00526             } \
00527         } else /* c>=0xf0 */ { \
00528             if(U8_IS_VALID_LEAD4_AND_T1(__b, __t1) && \
00529                     ++(i)!=(length) && U8_IS_TRAIL((s)[i]) && \
00530                     ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
00531                 ++(i); \
00532             } \
00533         } \
00534     } \
00535 }
00536 
00549 #define U8_FWD_N_UNSAFE(s, i, n) { \
00550     int32_t __N=(n); \
00551     while(__N>0) { \
00552         U8_FWD_1_UNSAFE(s, i); \
00553         --__N; \
00554     } \
00555 }
00556 
00572 #define U8_FWD_N(s, i, length, n) { \
00573     int32_t __N=(n); \
00574     while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
00575         U8_FWD_1(s, i, length); \
00576         --__N; \
00577     } \
00578 }
00579 
00593 #define U8_SET_CP_START_UNSAFE(s, i) { \
00594     while(U8_IS_TRAIL((s)[i])) { --(i); } \
00595 }
00596 
00611 #define U8_SET_CP_START(s, start, i) { \
00612     if(U8_IS_TRAIL((s)[(i)])) { \
00613         (i)=utf8_back1SafeBody(s, start, (i)); \
00614     } \
00615 }
00616 
00617 /* definitions with backward iteration -------------------------------------- */
00618 
00638 #define U8_PREV_UNSAFE(s, i, c) { \
00639     (c)=(uint8_t)(s)[--(i)]; \
00640     if(U8_IS_TRAIL(c)) { \
00641         uint8_t __b, __count=1, __shift=6; \
00642 \
00643         /* c is a trail byte */ \
00644         (c)&=0x3f; \
00645         for(;;) { \
00646             __b=(s)[--(i)]; \
00647             if(__b>=0xc0) { \
00648                 U8_MASK_LEAD_BYTE(__b, __count); \
00649                 (c)|=(UChar32)__b<<__shift; \
00650                 break; \
00651             } else { \
00652                 (c)|=(UChar32)(__b&0x3f)<<__shift; \
00653                 ++__count; \
00654                 __shift+=6; \
00655             } \
00656         } \
00657     } \
00658 }
00659 
00680 #define U8_PREV(s, start, i, c) { \
00681     (c)=(uint8_t)(s)[--(i)]; \
00682     if(!U8_IS_SINGLE(c)) { \
00683         (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
00684     } \
00685 }
00686 
00711 #define U8_PREV_OR_FFFD(s, start, i, c) { \
00712     (c)=(uint8_t)(s)[--(i)]; \
00713     if(!U8_IS_SINGLE(c)) { \
00714         (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -3); \
00715     } \
00716 }
00717 
00729 #define U8_BACK_1_UNSAFE(s, i) { \
00730     while(U8_IS_TRAIL((s)[--(i)])) {} \
00731 }
00732 
00745 #define U8_BACK_1(s, start, i) { \
00746     if(U8_IS_TRAIL((s)[--(i)])) { \
00747         (i)=utf8_back1SafeBody(s, start, (i)); \
00748     } \
00749 }
00750 
00764 #define U8_BACK_N_UNSAFE(s, i, n) { \
00765     int32_t __N=(n); \
00766     while(__N>0) { \
00767         U8_BACK_1_UNSAFE(s, i); \
00768         --__N; \
00769     } \
00770 }
00771 
00786 #define U8_BACK_N(s, start, i, n) { \
00787     int32_t __N=(n); \
00788     while(__N>0 && (i)>(start)) { \
00789         U8_BACK_1(s, start, i); \
00790         --__N; \
00791     } \
00792 }
00793 
00807 #define U8_SET_CP_LIMIT_UNSAFE(s, i) { \
00808     U8_BACK_1_UNSAFE(s, i); \
00809     U8_FWD_1_UNSAFE(s, i); \
00810 }
00811 
00829 #define U8_SET_CP_LIMIT(s, start, i, length) { \
00830     if((start)<(i) && ((i)<(length) || (length)<0)) { \
00831         U8_BACK_1(s, start, i); \
00832         U8_FWD_1(s, i, length); \
00833     } \
00834 }
00835 
00836 #endif