00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00034 #ifndef __UTF8_H__
00035 #define __UTF8_H__
00036
00037
00038 #ifndef __UTF_H__
00039 # include "unicode/utf.h"
00040 #endif
00041
00042
00043
00055 #ifdef U_UTF8_IMPL
00056 U_EXPORT const uint8_t
00057 #elif defined(U_STATIC_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION)
00058 U_CFUNC const uint8_t
00059 #else
00060 U_CFUNC U_IMPORT const uint8_t
00061 #endif
00062 utf8_countTrailBytes[256];
00063
00071 #define U8_COUNT_TRAIL_BYTES(leadByte) (utf8_countTrailBytes[(uint8_t)leadByte])
00072
00080 #define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
00081
00091 U_STABLE UChar32 U_EXPORT2
00092 utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
00093
00103 U_STABLE int32_t U_EXPORT2
00104 utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
00105
00115 U_STABLE UChar32 U_EXPORT2
00116 utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
00117
00127 U_STABLE int32_t U_EXPORT2
00128 utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
00129
00130
00131
00138 #define U8_IS_SINGLE(c) (((c)&0x80)==0)
00139
00146 #define U8_IS_LEAD(c) ((uint8_t)((c)-0xc0)<0x3e)
00147
00154 #define U8_IS_TRAIL(c) (((c)&0xc0)==0x80)
00155
00163 #define U8_LENGTH(c) \
00164 ((uint32_t)(c)<=0x7f ? 1 : \
00165 ((uint32_t)(c)<=0x7ff ? 2 : \
00166 ((uint32_t)(c)<=0xd7ff ? 3 : \
00167 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
00168 ((uint32_t)(c)<=0xffff ? 3 : 4)\
00169 ) \
00170 ) \
00171 ) \
00172 )
00173
00179 #define U8_MAX_LENGTH 4
00180
00197 #define U8_GET_UNSAFE(s, i, c) { \
00198 int32_t _u8_get_unsafe_index=(int32_t)(i); \
00199 U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
00200 U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
00201 }
00202
00221 #define U8_GET(s, start, i, length, c) { \
00222 int32_t _u8_get_index=(int32_t)(i); \
00223 U8_SET_CP_START(s, start, _u8_get_index); \
00224 U8_NEXT(s, _u8_get_index, length, c); \
00225 }
00226
00227
00228
00246 #define U8_NEXT_UNSAFE(s, i, c) { \
00247 (c)=(uint8_t)(s)[(i)++]; \
00248 if((uint8_t)((c)-0xc0)<0x35) { \
00249 uint8_t __count=U8_COUNT_TRAIL_BYTES(c); \
00250 U8_MASK_LEAD_BYTE(c, __count); \
00251 switch(__count) { \
00252 \
00253 case 3: \
00254 (c)=((c)<<6)|((s)[(i)++]&0x3f); \
00255 case 2: \
00256 (c)=((c)<<6)|((s)[(i)++]&0x3f); \
00257 case 1: \
00258 (c)=((c)<<6)|((s)[(i)++]&0x3f); \
00259 \
00260 break; \
00261 } \
00262 } \
00263 }
00264
00283 #define U8_NEXT(s, i, length, c) { \
00284 (c)=(uint8_t)(s)[(i)++]; \
00285 if((c)>=0x80) { \
00286 uint8_t __t1, __t2; \
00287 if( \
00288 (0xe0<(c) && (c)<=0xec) && \
00289 (((i)+1)<(length)) && \
00290 (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \
00291 (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \
00292 ) { \
00293 \
00294 (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \
00295 (i)+=2; \
00296 } else if( \
00297 ((c)<0xe0 && (c)>=0xc2) && \
00298 ((i)<(length)) && \
00299 (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \
00300 ) { \
00301 (c)=(UChar)((((c)&0x1f)<<6)|__t1); \
00302 ++(i); \
00303 } else if(U8_IS_LEAD(c)) { \
00304 \
00305 (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (int32_t)(length), c, -1); \
00306 } else { \
00307 (c)=U_SENTINEL; \
00308 } \
00309 } \
00310 }
00311
00325 #define U8_APPEND_UNSAFE(s, i, c) { \
00326 if((uint32_t)(c)<=0x7f) { \
00327 (s)[(i)++]=(uint8_t)(c); \
00328 } else { \
00329 if((uint32_t)(c)<=0x7ff) { \
00330 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
00331 } else { \
00332 if((uint32_t)(c)<=0xffff) { \
00333 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
00334 } else { \
00335 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
00336 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
00337 } \
00338 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
00339 } \
00340 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00341 } \
00342 }
00343
00361 #define U8_APPEND(s, i, capacity, c, isError) { \
00362 if((uint32_t)(c)<=0x7f) { \
00363 (s)[(i)++]=(uint8_t)(c); \
00364 } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \
00365 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
00366 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00367 } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \
00368 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
00369 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
00370 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00371 } else { \
00372 (i)=utf8_appendCharSafeBody(s, (int32_t)(i), (int32_t)(capacity), c, &(isError)); \
00373 } \
00374 }
00375
00386 #define U8_FWD_1_UNSAFE(s, i) { \
00387 (i)+=1+U8_COUNT_TRAIL_BYTES((s)[i]); \
00388 }
00389
00401 #define U8_FWD_1(s, i, length) { \
00402 uint8_t __b=(uint8_t)(s)[(i)++]; \
00403 if(U8_IS_LEAD(__b)) { \
00404 uint8_t __count=U8_COUNT_TRAIL_BYTES(__b); \
00405 if((i)+__count>(length)) { \
00406 __count=(uint8_t)((length)-(i)); \
00407 } \
00408 while(__count>0 && U8_IS_TRAIL((s)[i])) { \
00409 ++(i); \
00410 --__count; \
00411 } \
00412 } \
00413 }
00414
00427 #define U8_FWD_N_UNSAFE(s, i, n) { \
00428 int32_t __N=(n); \
00429 while(__N>0) { \
00430 U8_FWD_1_UNSAFE(s, i); \
00431 --__N; \
00432 } \
00433 }
00434
00448 #define U8_FWD_N(s, i, length, n) { \
00449 int32_t __N=(n); \
00450 while(__N>0 && (i)<(length)) { \
00451 U8_FWD_1(s, i, length); \
00452 --__N; \
00453 } \
00454 }
00455
00469 #define U8_SET_CP_START_UNSAFE(s, i) { \
00470 while(U8_IS_TRAIL((s)[i])) { --(i); } \
00471 }
00472
00487 #define U8_SET_CP_START(s, start, i) { \
00488 if(U8_IS_TRAIL((s)[(i)])) { \
00489 (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \
00490 } \
00491 }
00492
00493
00494
00514 #define U8_PREV_UNSAFE(s, i, c) { \
00515 (c)=(uint8_t)(s)[--(i)]; \
00516 if(U8_IS_TRAIL(c)) { \
00517 uint8_t __b, __count=1, __shift=6; \
00518 \
00519 \
00520 (c)&=0x3f; \
00521 for(;;) { \
00522 __b=(uint8_t)(s)[--(i)]; \
00523 if(__b>=0xc0) { \
00524 U8_MASK_LEAD_BYTE(__b, __count); \
00525 (c)|=(UChar32)__b<<__shift; \
00526 break; \
00527 } else { \
00528 (c)|=(UChar32)(__b&0x3f)<<__shift; \
00529 ++__count; \
00530 __shift+=6; \
00531 } \
00532 } \
00533 } \
00534 }
00535
00556 #define U8_PREV(s, start, i, c) { \
00557 (c)=(uint8_t)(s)[--(i)]; \
00558 if((c)>=0x80) { \
00559 if((c)<=0xbf) { \
00560 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
00561 } else { \
00562 (c)=U_SENTINEL; \
00563 } \
00564 } \
00565 }
00566
00578 #define U8_BACK_1_UNSAFE(s, i) { \
00579 while(U8_IS_TRAIL((s)[--(i)])) {} \
00580 }
00581
00594 #define U8_BACK_1(s, start, i) { \
00595 if(U8_IS_TRAIL((s)[--(i)])) { \
00596 (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \
00597 } \
00598 }
00599
00613 #define U8_BACK_N_UNSAFE(s, i, n) { \
00614 int32_t __N=(n); \
00615 while(__N>0) { \
00616 U8_BACK_1_UNSAFE(s, i); \
00617 --__N; \
00618 } \
00619 }
00620
00635 #define U8_BACK_N(s, start, i, n) { \
00636 int32_t __N=(n); \
00637 while(__N>0 && (i)>(start)) { \
00638 U8_BACK_1(s, start, i); \
00639 --__N; \
00640 } \
00641 }
00642
00656 #define U8_SET_CP_LIMIT_UNSAFE(s, i) { \
00657 U8_BACK_1_UNSAFE(s, i); \
00658 U8_FWD_1_UNSAFE(s, i); \
00659 }
00660
00676 #define U8_SET_CP_LIMIT(s, start, i, length) { \
00677 if((start)<(i) && (i)<(length)) { \
00678 U8_BACK_1(s, start, i); \
00679 U8_FWD_1(s, i, length); \
00680 } \
00681 }
00682
00683 #endif