00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00034 #ifndef __UTF16_H__
00035 #define __UTF16_H__
00036
00037 #include "unicode/umachine.h"
00038 #ifndef __UTF_H__
00039 # include "unicode/utf.h"
00040 #endif
00041
00042
00043
00050 #define U16_IS_SINGLE(c) !U_IS_SURROGATE(c)
00051
00058 #define U16_IS_LEAD(c) (((c)&0xfffffc00)==0xd800)
00059
00066 #define U16_IS_TRAIL(c) (((c)&0xfffffc00)==0xdc00)
00067
00074 #define U16_IS_SURROGATE(c) U_IS_SURROGATE(c)
00075
00083 #define U16_IS_SURROGATE_LEAD(c) (((c)&0x400)==0)
00084
00092 #define U16_IS_SURROGATE_TRAIL(c) (((c)&0x400)!=0)
00093
00098 #define U16_SURROGATE_OFFSET ((0xd800<<10UL)+0xdc00-0x10000)
00099
00111 #define U16_GET_SUPPLEMENTARY(lead, trail) \
00112 (((UChar32)(lead)<<10UL)+(UChar32)(trail)-U16_SURROGATE_OFFSET)
00113
00114
00122 #define U16_LEAD(supplementary) (UChar)(((supplementary)>>10)+0xd7c0)
00123
00131 #define U16_TRAIL(supplementary) (UChar)(((supplementary)&0x3ff)|0xdc00)
00132
00140 #define U16_LENGTH(c) ((uint32_t)(c)<=0xffff ? 1 : 2)
00141
00147 #define U16_MAX_LENGTH 2
00148
00166 #define U16_GET_UNSAFE(s, i, c) { \
00167 (c)=(s)[i]; \
00168 if(U16_IS_SURROGATE(c)) { \
00169 if(U16_IS_SURROGATE_LEAD(c)) { \
00170 (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)+1]); \
00171 } else { \
00172 (c)=U16_GET_SUPPLEMENTARY((s)[(i)-1], (c)); \
00173 } \
00174 } \
00175 }
00176
00200 #define U16_GET(s, start, i, length, c) { \
00201 (c)=(s)[i]; \
00202 if(U16_IS_SURROGATE(c)) { \
00203 uint16_t __c2; \
00204 if(U16_IS_SURROGATE_LEAD(c)) { \
00205 if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \
00206 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00207 } \
00208 } else { \
00209 if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00210 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00211 } \
00212 } \
00213 } \
00214 }
00215
00239 #define U16_GET_OR_FFFD(s, start, i, length, c) { \
00240 (c)=(s)[i]; \
00241 if(U16_IS_SURROGATE(c)) { \
00242 uint16_t __c2; \
00243 if(U16_IS_SURROGATE_LEAD(c)) { \
00244 if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \
00245 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00246 } else { \
00247 (c)=0xfffd; \
00248 } \
00249 } else { \
00250 if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00251 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00252 } else { \
00253 (c)=0xfffd; \
00254 } \
00255 } \
00256 } \
00257 }
00258
00259
00260
00280 #define U16_NEXT_UNSAFE(s, i, c) { \
00281 (c)=(s)[(i)++]; \
00282 if(U16_IS_LEAD(c)) { \
00283 (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)++]); \
00284 } \
00285 }
00286
00308 #define U16_NEXT(s, i, length, c) { \
00309 (c)=(s)[(i)++]; \
00310 if(U16_IS_LEAD(c)) { \
00311 uint16_t __c2; \
00312 if((i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \
00313 ++(i); \
00314 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00315 } \
00316 } \
00317 }
00318
00340 #define U16_NEXT_OR_FFFD(s, i, length, c) { \
00341 (c)=(s)[(i)++]; \
00342 if(U16_IS_SURROGATE(c)) { \
00343 uint16_t __c2; \
00344 if(U16_IS_SURROGATE_LEAD(c) && (i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \
00345 ++(i); \
00346 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00347 } else { \
00348 (c)=0xfffd; \
00349 } \
00350 } \
00351 }
00352
00366 #define U16_APPEND_UNSAFE(s, i, c) { \
00367 if((uint32_t)(c)<=0xffff) { \
00368 (s)[(i)++]=(uint16_t)(c); \
00369 } else { \
00370 (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
00371 (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
00372 } \
00373 }
00374
00392 #define U16_APPEND(s, i, capacity, c, isError) { \
00393 if((uint32_t)(c)<=0xffff) { \
00394 (s)[(i)++]=(uint16_t)(c); \
00395 } else if((uint32_t)(c)<=0x10ffff && (i)+1<(capacity)) { \
00396 (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
00397 (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
00398 } else { \
00399 (isError)=TRUE; \
00400 } \
00401 }
00402
00413 #define U16_FWD_1_UNSAFE(s, i) { \
00414 if(U16_IS_LEAD((s)[(i)++])) { \
00415 ++(i); \
00416 } \
00417 }
00418
00432 #define U16_FWD_1(s, i, length) { \
00433 if(U16_IS_LEAD((s)[(i)++]) && (i)!=(length) && U16_IS_TRAIL((s)[i])) { \
00434 ++(i); \
00435 } \
00436 }
00437
00450 #define U16_FWD_N_UNSAFE(s, i, n) { \
00451 int32_t __N=(n); \
00452 while(__N>0) { \
00453 U16_FWD_1_UNSAFE(s, i); \
00454 --__N; \
00455 } \
00456 }
00457
00473 #define U16_FWD_N(s, i, length, n) { \
00474 int32_t __N=(n); \
00475 while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
00476 U16_FWD_1(s, i, length); \
00477 --__N; \
00478 } \
00479 }
00480
00494 #define U16_SET_CP_START_UNSAFE(s, i) { \
00495 if(U16_IS_TRAIL((s)[i])) { \
00496 --(i); \
00497 } \
00498 }
00499
00514 #define U16_SET_CP_START(s, start, i) { \
00515 if(U16_IS_TRAIL((s)[i]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
00516 --(i); \
00517 } \
00518 }
00519
00520
00521
00542 #define U16_PREV_UNSAFE(s, i, c) { \
00543 (c)=(s)[--(i)]; \
00544 if(U16_IS_TRAIL(c)) { \
00545 (c)=U16_GET_SUPPLEMENTARY((s)[--(i)], (c)); \
00546 } \
00547 }
00548
00569 #define U16_PREV(s, start, i, c) { \
00570 (c)=(s)[--(i)]; \
00571 if(U16_IS_TRAIL(c)) { \
00572 uint16_t __c2; \
00573 if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00574 --(i); \
00575 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00576 } \
00577 } \
00578 }
00579
00600 #define U16_PREV_OR_FFFD(s, start, i, c) { \
00601 (c)=(s)[--(i)]; \
00602 if(U16_IS_SURROGATE(c)) { \
00603 uint16_t __c2; \
00604 if(U16_IS_SURROGATE_TRAIL(c) && (i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00605 --(i); \
00606 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00607 } else { \
00608 (c)=0xfffd; \
00609 } \
00610 } \
00611 }
00612
00624 #define U16_BACK_1_UNSAFE(s, i) { \
00625 if(U16_IS_TRAIL((s)[--(i)])) { \
00626 --(i); \
00627 } \
00628 }
00629
00642 #define U16_BACK_1(s, start, i) { \
00643 if(U16_IS_TRAIL((s)[--(i)]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
00644 --(i); \
00645 } \
00646 }
00647
00661 #define U16_BACK_N_UNSAFE(s, i, n) { \
00662 int32_t __N=(n); \
00663 while(__N>0) { \
00664 U16_BACK_1_UNSAFE(s, i); \
00665 --__N; \
00666 } \
00667 }
00668
00683 #define U16_BACK_N(s, start, i, n) { \
00684 int32_t __N=(n); \
00685 while(__N>0 && (i)>(start)) { \
00686 U16_BACK_1(s, start, i); \
00687 --__N; \
00688 } \
00689 }
00690
00704 #define U16_SET_CP_LIMIT_UNSAFE(s, i) { \
00705 if(U16_IS_LEAD((s)[(i)-1])) { \
00706 ++(i); \
00707 } \
00708 }
00709
00727 #define U16_SET_CP_LIMIT(s, start, i, length) { \
00728 if((start)<(i) && ((i)<(length) || (length)<0) && U16_IS_LEAD((s)[(i)-1]) && U16_IS_TRAIL((s)[i])) { \
00729 ++(i); \
00730 } \
00731 }
00732
00733 #endif