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
00216 #ifndef U_HIDE_DRAFT_API
00217
00241 #define U16_GET_OR_FFFD(s, start, i, length, c) { \
00242 (c)=(s)[i]; \
00243 if(U16_IS_SURROGATE(c)) { \
00244 uint16_t __c2; \
00245 if(U16_IS_SURROGATE_LEAD(c)) { \
00246 if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \
00247 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00248 } else { \
00249 (c)=0xfffd; \
00250 } \
00251 } else { \
00252 if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00253 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00254 } else { \
00255 (c)=0xfffd; \
00256 } \
00257 } \
00258 } \
00259 }
00260
00261 #endif // U_HIDE_DRAFT_API
00262
00263
00264
00284 #define U16_NEXT_UNSAFE(s, i, c) { \
00285 (c)=(s)[(i)++]; \
00286 if(U16_IS_LEAD(c)) { \
00287 (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)++]); \
00288 } \
00289 }
00290
00312 #define U16_NEXT(s, i, length, c) { \
00313 (c)=(s)[(i)++]; \
00314 if(U16_IS_LEAD(c)) { \
00315 uint16_t __c2; \
00316 if((i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \
00317 ++(i); \
00318 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00319 } \
00320 } \
00321 }
00322
00323 #ifndef U_HIDE_DRAFT_API
00324
00346 #define U16_NEXT_OR_FFFD(s, i, length, c) { \
00347 (c)=(s)[(i)++]; \
00348 if(U16_IS_SURROGATE(c)) { \
00349 uint16_t __c2; \
00350 if(U16_IS_SURROGATE_LEAD(c) && (i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \
00351 ++(i); \
00352 (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
00353 } else { \
00354 (c)=0xfffd; \
00355 } \
00356 } \
00357 }
00358
00359 #endif // U_HIDE_DRAFT_API
00360
00374 #define U16_APPEND_UNSAFE(s, i, c) { \
00375 if((uint32_t)(c)<=0xffff) { \
00376 (s)[(i)++]=(uint16_t)(c); \
00377 } else { \
00378 (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
00379 (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
00380 } \
00381 }
00382
00400 #define U16_APPEND(s, i, capacity, c, isError) { \
00401 if((uint32_t)(c)<=0xffff) { \
00402 (s)[(i)++]=(uint16_t)(c); \
00403 } else if((uint32_t)(c)<=0x10ffff && (i)+1<(capacity)) { \
00404 (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
00405 (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
00406 } else { \
00407 (isError)=TRUE; \
00408 } \
00409 }
00410
00421 #define U16_FWD_1_UNSAFE(s, i) { \
00422 if(U16_IS_LEAD((s)[(i)++])) { \
00423 ++(i); \
00424 } \
00425 }
00426
00440 #define U16_FWD_1(s, i, length) { \
00441 if(U16_IS_LEAD((s)[(i)++]) && (i)!=(length) && U16_IS_TRAIL((s)[i])) { \
00442 ++(i); \
00443 } \
00444 }
00445
00458 #define U16_FWD_N_UNSAFE(s, i, n) { \
00459 int32_t __N=(n); \
00460 while(__N>0) { \
00461 U16_FWD_1_UNSAFE(s, i); \
00462 --__N; \
00463 } \
00464 }
00465
00481 #define U16_FWD_N(s, i, length, n) { \
00482 int32_t __N=(n); \
00483 while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
00484 U16_FWD_1(s, i, length); \
00485 --__N; \
00486 } \
00487 }
00488
00502 #define U16_SET_CP_START_UNSAFE(s, i) { \
00503 if(U16_IS_TRAIL((s)[i])) { \
00504 --(i); \
00505 } \
00506 }
00507
00522 #define U16_SET_CP_START(s, start, i) { \
00523 if(U16_IS_TRAIL((s)[i]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
00524 --(i); \
00525 } \
00526 }
00527
00528
00529
00550 #define U16_PREV_UNSAFE(s, i, c) { \
00551 (c)=(s)[--(i)]; \
00552 if(U16_IS_TRAIL(c)) { \
00553 (c)=U16_GET_SUPPLEMENTARY((s)[--(i)], (c)); \
00554 } \
00555 }
00556
00577 #define U16_PREV(s, start, i, c) { \
00578 (c)=(s)[--(i)]; \
00579 if(U16_IS_TRAIL(c)) { \
00580 uint16_t __c2; \
00581 if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00582 --(i); \
00583 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00584 } \
00585 } \
00586 }
00587
00588 #ifndef U_HIDE_DRAFT_API
00589
00610 #define U16_PREV_OR_FFFD(s, start, i, c) { \
00611 (c)=(s)[--(i)]; \
00612 if(U16_IS_SURROGATE(c)) { \
00613 uint16_t __c2; \
00614 if(U16_IS_SURROGATE_TRAIL(c) && (i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
00615 --(i); \
00616 (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
00617 } else { \
00618 (c)=0xfffd; \
00619 } \
00620 } \
00621 }
00622
00623 #endif // U_HIDE_DRAFT_API
00624
00636 #define U16_BACK_1_UNSAFE(s, i) { \
00637 if(U16_IS_TRAIL((s)[--(i)])) { \
00638 --(i); \
00639 } \
00640 }
00641
00654 #define U16_BACK_1(s, start, i) { \
00655 if(U16_IS_TRAIL((s)[--(i)]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
00656 --(i); \
00657 } \
00658 }
00659
00673 #define U16_BACK_N_UNSAFE(s, i, n) { \
00674 int32_t __N=(n); \
00675 while(__N>0) { \
00676 U16_BACK_1_UNSAFE(s, i); \
00677 --__N; \
00678 } \
00679 }
00680
00695 #define U16_BACK_N(s, start, i, n) { \
00696 int32_t __N=(n); \
00697 while(__N>0 && (i)>(start)) { \
00698 U16_BACK_1(s, start, i); \
00699 --__N; \
00700 } \
00701 }
00702
00716 #define U16_SET_CP_LIMIT_UNSAFE(s, i) { \
00717 if(U16_IS_LEAD((s)[(i)-1])) { \
00718 ++(i); \
00719 } \
00720 }
00721
00739 #define U16_SET_CP_LIMIT(s, start, i, length) { \
00740 if((start)<(i) && ((i)<(length) || (length)<0) && U16_IS_LEAD((s)[(i)-1]) && U16_IS_TRAIL((s)[i])) { \
00741 ++(i); \
00742 } \
00743 }
00744
00745 #endif