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 <stdbool.h>
00038 #include "unicode/umachine.h"
00039 #ifndef __UTF_H__
00040 # include "unicode/utf.h"
00041 #endif
00042
00043
00044
00051 #define U16_IS_SINGLE(c) !U_IS_SURROGATE(c)
00052
00059 #define U16_IS_LEAD(c) (((c)&0xfffffc00)==0xd800)
00060
00067 #define U16_IS_TRAIL(c) (((c)&0xfffffc00)==0xdc00)
00068
00075 #define U16_IS_SURROGATE(c) U_IS_SURROGATE(c)
00076
00084 #define U16_IS_SURROGATE_LEAD(c) (((c)&0x400)==0)
00085
00093 #define U16_IS_SURROGATE_TRAIL(c) (((c)&0x400)!=0)
00094
00099 #define U16_SURROGATE_OFFSET ((0xd800<<10UL)+0xdc00-0x10000)
00100
00112 #define U16_GET_SUPPLEMENTARY(lead, trail) \
00113 (((UChar32)(lead)<<10UL)+(UChar32)(trail)-U16_SURROGATE_OFFSET)
00114
00115
00123 #define U16_LEAD(supplementary) (UChar)(((supplementary)>>10)+0xd7c0)
00124
00132 #define U16_TRAIL(supplementary) (UChar)(((supplementary)&0x3ff)|0xdc00)
00133
00141 #define U16_LENGTH(c) ((uint32_t)(c)<=0xffff ? 1 : 2)
00142
00148 #define U16_MAX_LENGTH 2
00149
00167 #define U16_GET_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00168 (c)=(s)[i]; \
00169 if(U16_IS_SURROGATE(c)) { \
00170 if(U16_IS_SURROGATE_LEAD(c)) { \
00171 (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)+1]); \
00172 } else { \
00173 (c)=U16_GET_SUPPLEMENTARY((s)[(i)-1], (c)); \
00174 } \
00175 } \
00176 } UPRV_BLOCK_MACRO_END
00177
00201 #define U16_GET(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
00202 (c)=(s)[i]; \
00203 if(U16_IS_SURROGATE(c)) { \
00204 uint16_t __c2; \
00205 if(U16_IS_SURROGATE_LEAD(c)) { \
00206 if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \
00207 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00208 } \
00209 } else { \
00210 if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00211 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00212 } \
00213 } \
00214 } \
00215 } UPRV_BLOCK_MACRO_END
00216
00240 #define U16_GET_OR_FFFD(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
00241 (c)=(s)[i]; \
00242 if(U16_IS_SURROGATE(c)) { \
00243 uint16_t __c2; \
00244 if(U16_IS_SURROGATE_LEAD(c)) { \
00245 if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \
00246 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00247 } else { \
00248 (c)=0xfffd; \
00249 } \
00250 } else { \
00251 if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00252 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00253 } else { \
00254 (c)=0xfffd; \
00255 } \
00256 } \
00257 } \
00258 } UPRV_BLOCK_MACRO_END
00259
00260
00261
00281 #define U16_NEXT_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00282 (c)=(s)[(i)++]; \
00283 if(U16_IS_LEAD(c)) { \
00284 (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)++]); \
00285 } \
00286 } UPRV_BLOCK_MACRO_END
00287
00309 #define U16_NEXT(s, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
00310 (c)=(s)[(i)++]; \
00311 if(U16_IS_LEAD(c)) { \
00312 uint16_t __c2; \
00313 if((i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \
00314 ++(i); \
00315 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00316 } \
00317 } \
00318 } UPRV_BLOCK_MACRO_END
00319
00341 #define U16_NEXT_OR_FFFD(s, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
00342 (c)=(s)[(i)++]; \
00343 if(U16_IS_SURROGATE(c)) { \
00344 uint16_t __c2; \
00345 if(U16_IS_SURROGATE_LEAD(c) && (i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \
00346 ++(i); \
00347 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00348 } else { \
00349 (c)=0xfffd; \
00350 } \
00351 } \
00352 } UPRV_BLOCK_MACRO_END
00353
00367 #define U16_APPEND_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00368 if((uint32_t)(c)<=0xffff) { \
00369 (s)[(i)++]=(uint16_t)(c); \
00370 } else { \
00371 (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
00372 (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
00373 } \
00374 } UPRV_BLOCK_MACRO_END
00375
00393 #define U16_APPEND(s, i, capacity, c, isError) UPRV_BLOCK_MACRO_BEGIN { \
00394 if((uint32_t)(c)<=0xffff) { \
00395 (s)[(i)++]=(uint16_t)(c); \
00396 } else if((uint32_t)(c)<=0x10ffff && (i)+1<(capacity)) { \
00397 (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
00398 (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
00399 } else { \
00400 (isError)=true; \
00401 } \
00402 } UPRV_BLOCK_MACRO_END
00403
00414 #define U16_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
00415 if(U16_IS_LEAD((s)[(i)++])) { \
00416 ++(i); \
00417 } \
00418 } UPRV_BLOCK_MACRO_END
00419
00433 #define U16_FWD_1(s, i, length) UPRV_BLOCK_MACRO_BEGIN { \
00434 if(U16_IS_LEAD((s)[(i)++]) && (i)!=(length) && U16_IS_TRAIL((s)[i])) { \
00435 ++(i); \
00436 } \
00437 } UPRV_BLOCK_MACRO_END
00438
00451 #define U16_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
00452 int32_t __N=(n); \
00453 while(__N>0) { \
00454 U16_FWD_1_UNSAFE(s, i); \
00455 --__N; \
00456 } \
00457 } UPRV_BLOCK_MACRO_END
00458
00474 #define U16_FWD_N(s, i, length, n) UPRV_BLOCK_MACRO_BEGIN { \
00475 int32_t __N=(n); \
00476 while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
00477 U16_FWD_1(s, i, length); \
00478 --__N; \
00479 } \
00480 } UPRV_BLOCK_MACRO_END
00481
00495 #define U16_SET_CP_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
00496 if(U16_IS_TRAIL((s)[i])) { \
00497 --(i); \
00498 } \
00499 } UPRV_BLOCK_MACRO_END
00500
00515 #define U16_SET_CP_START(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
00516 if(U16_IS_TRAIL((s)[i]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
00517 --(i); \
00518 } \
00519 } UPRV_BLOCK_MACRO_END
00520
00521
00522
00543 #define U16_PREV_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00544 (c)=(s)[--(i)]; \
00545 if(U16_IS_TRAIL(c)) { \
00546 (c)=U16_GET_SUPPLEMENTARY((s)[--(i)], (c)); \
00547 } \
00548 } UPRV_BLOCK_MACRO_END
00549
00570 #define U16_PREV(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00571 (c)=(s)[--(i)]; \
00572 if(U16_IS_TRAIL(c)) { \
00573 uint16_t __c2; \
00574 if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00575 --(i); \
00576 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00577 } \
00578 } \
00579 } UPRV_BLOCK_MACRO_END
00580
00601 #define U16_PREV_OR_FFFD(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
00602 (c)=(s)[--(i)]; \
00603 if(U16_IS_SURROGATE(c)) { \
00604 uint16_t __c2; \
00605 if(U16_IS_SURROGATE_TRAIL(c) && (i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00606 --(i); \
00607 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00608 } else { \
00609 (c)=0xfffd; \
00610 } \
00611 } \
00612 } UPRV_BLOCK_MACRO_END
00613
00625 #define U16_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
00626 if(U16_IS_TRAIL((s)[--(i)])) { \
00627 --(i); \
00628 } \
00629 } UPRV_BLOCK_MACRO_END
00630
00643 #define U16_BACK_1(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
00644 if(U16_IS_TRAIL((s)[--(i)]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
00645 --(i); \
00646 } \
00647 } UPRV_BLOCK_MACRO_END
00648
00662 #define U16_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
00663 int32_t __N=(n); \
00664 while(__N>0) { \
00665 U16_BACK_1_UNSAFE(s, i); \
00666 --__N; \
00667 } \
00668 } UPRV_BLOCK_MACRO_END
00669
00684 #define U16_BACK_N(s, start, i, n) UPRV_BLOCK_MACRO_BEGIN { \
00685 int32_t __N=(n); \
00686 while(__N>0 && (i)>(start)) { \
00687 U16_BACK_1(s, start, i); \
00688 --__N; \
00689 } \
00690 } UPRV_BLOCK_MACRO_END
00691
00705 #define U16_SET_CP_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
00706 if(U16_IS_LEAD((s)[(i)-1])) { \
00707 ++(i); \
00708 } \
00709 } UPRV_BLOCK_MACRO_END
00710
00728 #define U16_SET_CP_LIMIT(s, start, i, length) UPRV_BLOCK_MACRO_BEGIN { \
00729 if((start)<(i) && ((i)<(length) || (length)<0) && U16_IS_LEAD((s)[(i)-1]) && U16_IS_TRAIL((s)[i])) { \
00730 ++(i); \
00731 } \
00732 } UPRV_BLOCK_MACRO_END
00733
00734 #endif