ICU 70.1  70.1
utf16.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 *
6 * Copyright (C) 1999-2012, International Business Machines
7 * Corporation and others. All Rights Reserved.
8 *
9 *******************************************************************************
10 * file name: utf16.h
11 * encoding: UTF-8
12 * tab size: 8 (not used)
13 * indentation:4
14 *
15 * created on: 1999sep09
16 * created by: Markus W. Scherer
17 */
18 
34 #ifndef __UTF16_H__
35 #define __UTF16_H__
36 
37 #include <stdbool.h>
38 #include "unicode/umachine.h"
39 #ifndef __UTF_H__
40 # include "unicode/utf.h"
41 #endif
42 
43 /* single-code point definitions -------------------------------------------- */
44 
51 #define U16_IS_SINGLE(c) !U_IS_SURROGATE(c)
52 
59 #define U16_IS_LEAD(c) (((c)&0xfffffc00)==0xd800)
60 
67 #define U16_IS_TRAIL(c) (((c)&0xfffffc00)==0xdc00)
68 
75 #define U16_IS_SURROGATE(c) U_IS_SURROGATE(c)
76 
84 #define U16_IS_SURROGATE_LEAD(c) (((c)&0x400)==0)
85 
93 #define U16_IS_SURROGATE_TRAIL(c) (((c)&0x400)!=0)
94 
99 #define U16_SURROGATE_OFFSET ((0xd800<<10UL)+0xdc00-0x10000)
100 
112 #define U16_GET_SUPPLEMENTARY(lead, trail) \
113  (((UChar32)(lead)<<10UL)+(UChar32)(trail)-U16_SURROGATE_OFFSET)
114 
115 
123 #define U16_LEAD(supplementary) (UChar)(((supplementary)>>10)+0xd7c0)
124 
132 #define U16_TRAIL(supplementary) (UChar)(((supplementary)&0x3ff)|0xdc00)
133 
141 #define U16_LENGTH(c) ((uint32_t)(c)<=0xffff ? 1 : 2)
142 
148 #define U16_MAX_LENGTH 2
149 
167 #define U16_GET_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
168  (c)=(s)[i]; \
169  if(U16_IS_SURROGATE(c)) { \
170  if(U16_IS_SURROGATE_LEAD(c)) { \
171  (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)+1]); \
172  } else { \
173  (c)=U16_GET_SUPPLEMENTARY((s)[(i)-1], (c)); \
174  } \
175  } \
176 } UPRV_BLOCK_MACRO_END
177 
201 #define U16_GET(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
202  (c)=(s)[i]; \
203  if(U16_IS_SURROGATE(c)) { \
204  uint16_t __c2; \
205  if(U16_IS_SURROGATE_LEAD(c)) { \
206  if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \
207  (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
208  } \
209  } else { \
210  if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
211  (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
212  } \
213  } \
214  } \
215 } UPRV_BLOCK_MACRO_END
216 
240 #define U16_GET_OR_FFFD(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
241  (c)=(s)[i]; \
242  if(U16_IS_SURROGATE(c)) { \
243  uint16_t __c2; \
244  if(U16_IS_SURROGATE_LEAD(c)) { \
245  if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \
246  (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
247  } else { \
248  (c)=0xfffd; \
249  } \
250  } else { \
251  if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
252  (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
253  } else { \
254  (c)=0xfffd; \
255  } \
256  } \
257  } \
258 } UPRV_BLOCK_MACRO_END
259 
260 /* definitions with forward iteration --------------------------------------- */
261 
281 #define U16_NEXT_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
282  (c)=(s)[(i)++]; \
283  if(U16_IS_LEAD(c)) { \
284  (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)++]); \
285  } \
286 } UPRV_BLOCK_MACRO_END
287 
309 #define U16_NEXT(s, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
310  (c)=(s)[(i)++]; \
311  if(U16_IS_LEAD(c)) { \
312  uint16_t __c2; \
313  if((i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \
314  ++(i); \
315  (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
316  } \
317  } \
318 } UPRV_BLOCK_MACRO_END
319 
341 #define U16_NEXT_OR_FFFD(s, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
342  (c)=(s)[(i)++]; \
343  if(U16_IS_SURROGATE(c)) { \
344  uint16_t __c2; \
345  if(U16_IS_SURROGATE_LEAD(c) && (i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \
346  ++(i); \
347  (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
348  } else { \
349  (c)=0xfffd; \
350  } \
351  } \
352 } UPRV_BLOCK_MACRO_END
353 
367 #define U16_APPEND_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
368  if((uint32_t)(c)<=0xffff) { \
369  (s)[(i)++]=(uint16_t)(c); \
370  } else { \
371  (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
372  (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
373  } \
374 } UPRV_BLOCK_MACRO_END
375 
393 #define U16_APPEND(s, i, capacity, c, isError) UPRV_BLOCK_MACRO_BEGIN { \
394  if((uint32_t)(c)<=0xffff) { \
395  (s)[(i)++]=(uint16_t)(c); \
396  } else if((uint32_t)(c)<=0x10ffff && (i)+1<(capacity)) { \
397  (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \
398  (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \
399  } else /* c>0x10ffff or not enough space */ { \
400  (isError)=true; \
401  } \
402 } UPRV_BLOCK_MACRO_END
403 
414 #define U16_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
415  if(U16_IS_LEAD((s)[(i)++])) { \
416  ++(i); \
417  } \
418 } UPRV_BLOCK_MACRO_END
419 
433 #define U16_FWD_1(s, i, length) UPRV_BLOCK_MACRO_BEGIN { \
434  if(U16_IS_LEAD((s)[(i)++]) && (i)!=(length) && U16_IS_TRAIL((s)[i])) { \
435  ++(i); \
436  } \
437 } UPRV_BLOCK_MACRO_END
438 
451 #define U16_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
452  int32_t __N=(n); \
453  while(__N>0) { \
454  U16_FWD_1_UNSAFE(s, i); \
455  --__N; \
456  } \
457 } UPRV_BLOCK_MACRO_END
458 
474 #define U16_FWD_N(s, i, length, n) UPRV_BLOCK_MACRO_BEGIN { \
475  int32_t __N=(n); \
476  while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
477  U16_FWD_1(s, i, length); \
478  --__N; \
479  } \
480 } UPRV_BLOCK_MACRO_END
481 
495 #define U16_SET_CP_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
496  if(U16_IS_TRAIL((s)[i])) { \
497  --(i); \
498  } \
499 } UPRV_BLOCK_MACRO_END
500 
515 #define U16_SET_CP_START(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
516  if(U16_IS_TRAIL((s)[i]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
517  --(i); \
518  } \
519 } UPRV_BLOCK_MACRO_END
520 
521 /* definitions with backward iteration -------------------------------------- */
522 
543 #define U16_PREV_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
544  (c)=(s)[--(i)]; \
545  if(U16_IS_TRAIL(c)) { \
546  (c)=U16_GET_SUPPLEMENTARY((s)[--(i)], (c)); \
547  } \
548 } UPRV_BLOCK_MACRO_END
549 
570 #define U16_PREV(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
571  (c)=(s)[--(i)]; \
572  if(U16_IS_TRAIL(c)) { \
573  uint16_t __c2; \
574  if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
575  --(i); \
576  (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
577  } \
578  } \
579 } UPRV_BLOCK_MACRO_END
580 
601 #define U16_PREV_OR_FFFD(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
602  (c)=(s)[--(i)]; \
603  if(U16_IS_SURROGATE(c)) { \
604  uint16_t __c2; \
605  if(U16_IS_SURROGATE_TRAIL(c) && (i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \
606  --(i); \
607  (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
608  } else { \
609  (c)=0xfffd; \
610  } \
611  } \
612 } UPRV_BLOCK_MACRO_END
613 
625 #define U16_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
626  if(U16_IS_TRAIL((s)[--(i)])) { \
627  --(i); \
628  } \
629 } UPRV_BLOCK_MACRO_END
630 
643 #define U16_BACK_1(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
644  if(U16_IS_TRAIL((s)[--(i)]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \
645  --(i); \
646  } \
647 } UPRV_BLOCK_MACRO_END
648 
662 #define U16_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
663  int32_t __N=(n); \
664  while(__N>0) { \
665  U16_BACK_1_UNSAFE(s, i); \
666  --__N; \
667  } \
668 } UPRV_BLOCK_MACRO_END
669 
684 #define U16_BACK_N(s, start, i, n) UPRV_BLOCK_MACRO_BEGIN { \
685  int32_t __N=(n); \
686  while(__N>0 && (i)>(start)) { \
687  U16_BACK_1(s, start, i); \
688  --__N; \
689  } \
690 } UPRV_BLOCK_MACRO_END
691 
705 #define U16_SET_CP_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
706  if(U16_IS_LEAD((s)[(i)-1])) { \
707  ++(i); \
708  } \
709 } UPRV_BLOCK_MACRO_END
710 
728 #define U16_SET_CP_LIMIT(s, start, i, length) UPRV_BLOCK_MACRO_BEGIN { \
729  if((start)<(i) && ((i)<(length) || (length)<0) && U16_IS_LEAD((s)[(i)-1]) && U16_IS_TRAIL((s)[i])) { \
730  ++(i); \
731  } \
732 } UPRV_BLOCK_MACRO_END
733 
734 #endif
Basic types and constants for UTF.
C API: Code point macros.