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) 1999-2015, International Business Machines
7* Corporation and others. All Rights Reserved.
8*
9*******************************************************************************
10* file name: utf8.h
11* encoding: UTF-8
12* tab size: 8 (not used)
13* indentation:4
14*
15* created on: 1999sep13
16* created by: Markus W. Scherer
17*/
18
19/**
20 * \file
21 * \brief C API: 8-bit Unicode handling macros
22 *
23 * This file defines macros to deal with 8-bit Unicode (UTF-8) code units (bytes) and strings.
24 *
25 * For more information see utf.h and the ICU User Guide Strings chapter
26 * (http://userguide.icu-project.org/strings).
27 *
28 * <em>Usage:</em>
29 * ICU coding guidelines for if() statements should be followed when using these macros.
30 * Compound statements (curly braces {}) must be used for if-else-while...
31 * bodies and all macro statements should be terminated with semicolon.
32 */
33
34#ifndef __UTF8_H__
35#define __UTF8_H__
36
37#include "unicode/umachine.h"
38#ifndef __UTF_H__
39# include "unicode/utf.h"
40#endif
41
42/* internal definitions ----------------------------------------------------- */
43
44/**
45 * Counts the trail bytes for a UTF-8 lead byte.
46 * Returns 0 for 0..0xc1 as well as for 0xf5..0xff.
47 * leadByte might be evaluated multiple times.
48 *
49 * This is internal since it is not meant to be called directly by external clients;
50 * however it is called by public macros in this file and thus must remain stable.
51 *
52 * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
53 * @internal
54 */
55#define U8_COUNT_TRAIL_BYTES(leadByte) \
56 (U8_IS_LEAD(leadByte) ? \
57 ((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0)+1 : 0)
58
59/**
60 * Counts the trail bytes for a UTF-8 lead byte of a valid UTF-8 sequence.
61 * Returns 0 for 0..0xc1. Undefined for 0xf5..0xff.
62 * leadByte might be evaluated multiple times.
63 *
64 * This is internal since it is not meant to be called directly by external clients;
65 * however it is called by public macros in this file and thus must remain stable.
66 *
67 * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
68 * @internal
69 */
70#define U8_COUNT_TRAIL_BYTES_UNSAFE(leadByte) \
71 (((uint8_t)(leadByte)>=0xc2)+((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0))
72
73/**
74 * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value.
75 *
76 * This is internal since it is not meant to be called directly by external clients;
77 * however it is called by public macros in this file and thus must remain stable.
78 * @internal
79 */
80#define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
81
82/**
83 * Internal bit vector for 3-byte UTF-8 validity check, for use in U8_IS_VALID_LEAD3_AND_T1.
84 * Each bit indicates whether one lead byte + first trail byte pair starts a valid sequence.
85 * Lead byte E0..EF bits 3..0 are used as byte index,
86 * first trail byte bits 7..5 are used as bit index into that byte.
87 * @see U8_IS_VALID_LEAD3_AND_T1
88 * @internal
89 */
90#define U8_LEAD3_T1_BITS "\x20\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x10\x30\x30"
91
92/**
93 * Internal 3-byte UTF-8 validity check.
94 * Non-zero if lead byte E0..EF and first trail byte 00..FF start a valid sequence.
95 * @internal
96 */
97#define U8_IS_VALID_LEAD3_AND_T1(lead, t1) (U8_LEAD3_T1_BITS[(lead)&0xf]&(1<<((uint8_t)(t1)>>5)))
98
99/**
100 * Internal bit vector for 4-byte UTF-8 validity check, for use in U8_IS_VALID_LEAD4_AND_T1.
101 * Each bit indicates whether one lead byte + first trail byte pair starts a valid sequence.
102 * First trail byte bits 7..4 are used as byte index,
103 * lead byte F0..F4 bits 2..0 are used as bit index into that byte.
104 * @see U8_IS_VALID_LEAD4_AND_T1
105 * @internal
106 */
107#define U8_LEAD4_T1_BITS "\x00\x00\x00\x00\x00\x00\x00\x00\x1E\x0F\x0F\x0F\x00\x00\x00\x00"
108
109/**
110 * Internal 4-byte UTF-8 validity check.
111 * Non-zero if lead byte F0..F4 and first trail byte 00..FF start a valid sequence.
112 * @internal
113 */
114#define U8_IS_VALID_LEAD4_AND_T1(lead, t1) (U8_LEAD4_T1_BITS[(uint8_t)(t1)>>4]&(1<<((lead)&7)))
115
116/**
117 * Function for handling "next code point" with error-checking.
118 *
119 * This is internal since it is not meant to be called directly by external clients;
120 * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
121 * file and thus must remain stable, and should not be hidden when other internal
122 * functions are hidden (otherwise public macros would fail to compile).
123 * @internal
124 */
125U_STABLE UChar32 U_EXPORT2
126utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
127
128/**
129 * Function for handling "append code point" with error-checking.
130 *
131 * This is internal since it is not meant to be called directly by external clients;
132 * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
133 * file and thus must remain stable, and should not be hidden when other internal
134 * functions are hidden (otherwise public macros would fail to compile).
135 * @internal
136 */
137U_STABLE int32_t U_EXPORT2
138utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
139
140/**
141 * Function for handling "previous code point" with error-checking.
142 *
143 * This is internal since it is not meant to be called directly by external clients;
144 * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
145 * file and thus must remain stable, and should not be hidden when other internal
146 * functions are hidden (otherwise public macros would fail to compile).
147 * @internal
148 */
149U_STABLE UChar32 U_EXPORT2
150utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
151
152/**
153 * Function for handling "skip backward one code point" with error-checking.
154 *
155 * This is internal since it is not meant to be called directly by external clients;
156 * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
157 * file and thus must remain stable, and should not be hidden when other internal
158 * functions are hidden (otherwise public macros would fail to compile).
159 * @internal
160 */
161U_STABLE int32_t U_EXPORT2
162utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
163
164/* single-code point definitions -------------------------------------------- */
165
166/**
167 * Does this code unit (byte) encode a code point by itself (US-ASCII 0..0x7f)?
168 * @param c 8-bit code unit (byte)
169 * @return TRUE or FALSE
170 * @stable ICU 2.4
171 */
172#define U8_IS_SINGLE(c) (((c)&0x80)==0)
173
174/**
175 * Is this code unit (byte) a UTF-8 lead byte? (0xC2..0xF4)
176 * @param c 8-bit code unit (byte)
177 * @return TRUE or FALSE
178 * @stable ICU 2.4
179 */
180#define U8_IS_LEAD(c) ((uint8_t)((c)-0xc2)<=0x32)
181// 0x32=0xf4-0xc2
182
183/**
184 * Is this code unit (byte) a UTF-8 trail byte? (0x80..0xBF)
185 * @param c 8-bit code unit (byte)
186 * @return TRUE or FALSE
187 * @stable ICU 2.4
188 */
189#define U8_IS_TRAIL(c) ((int8_t)(c)<-0x40)
190
191/**
192 * How many code units (bytes) are used for the UTF-8 encoding
193 * of this Unicode code point?
194 * @param c 32-bit code point
195 * @return 1..4, or 0 if c is a surrogate or not a Unicode code point
196 * @stable ICU 2.4
197 */
198#define U8_LENGTH(c) \
199 ((uint32_t)(c)<=0x7f ? 1 : \
200 ((uint32_t)(c)<=0x7ff ? 2 : \
201 ((uint32_t)(c)<=0xd7ff ? 3 : \
202 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
203 ((uint32_t)(c)<=0xffff ? 3 : 4)\
204 ) \
205 ) \
206 ) \
207 )
208
209/**
210 * The maximum number of UTF-8 code units (bytes) per Unicode code point (U+0000..U+10ffff).
211 * @return 4
212 * @stable ICU 2.4
213 */
214#define U8_MAX_LENGTH 4
215
216/**
217 * Get a code point from a string at a random-access offset,
218 * without changing the offset.
219 * The offset may point to either the lead byte or one of the trail bytes
220 * for a code point, in which case the macro will read all of the bytes
221 * for the code point.
222 * The result is undefined if the offset points to an illegal UTF-8
223 * byte sequence.
224 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
225 *
226 * @param s const uint8_t * string
227 * @param i string offset
228 * @param c output UChar32 variable
229 * @see U8_GET
230 * @stable ICU 2.4
231 */
232#define U8_GET_UNSAFE(s, i, c) { \
233 int32_t _u8_get_unsafe_index=(int32_t)(i); \
234 U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
235 U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
236}
237
238/**
239 * Get a code point from a string at a random-access offset,
240 * without changing the offset.
241 * The offset may point to either the lead byte or one of the trail bytes
242 * for a code point, in which case the macro will read all of the bytes
243 * for the code point.
244 *
245 * The length can be negative for a NUL-terminated string.
246 *
247 * If the offset points to an illegal UTF-8 byte sequence, then
248 * c is set to a negative value.
249 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
250 *
251 * @param s const uint8_t * string
252 * @param start int32_t starting string offset
253 * @param i int32_t string offset, must be start<=i<length
254 * @param length int32_t string length
255 * @param c output UChar32 variable, set to <0 in case of an error
256 * @see U8_GET_UNSAFE
257 * @stable ICU 2.4
258 */
259#define U8_GET(s, start, i, length, c) { \
260 int32_t _u8_get_index=(i); \
261 U8_SET_CP_START(s, start, _u8_get_index); \
262 U8_NEXT(s, _u8_get_index, length, c); \
263}
264
265/**
266 * Get a code point from a string at a random-access offset,
267 * without changing the offset.
268 * The offset may point to either the lead byte or one of the trail bytes
269 * for a code point, in which case the macro will read all of the bytes
270 * for the code point.
271 *
272 * The length can be negative for a NUL-terminated string.
273 *
274 * If the offset points to an illegal UTF-8 byte sequence, then
275 * c is set to U+FFFD.
276 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT_OR_FFFD.
277 *
278 * This macro does not distinguish between a real U+FFFD in the text
279 * and U+FFFD returned for an ill-formed sequence.
280 * Use U8_GET() if that distinction is important.
281 *
282 * @param s const uint8_t * string
283 * @param start int32_t starting string offset
284 * @param i int32_t string offset, must be start<=i<length
285 * @param length int32_t string length
286 * @param c output UChar32 variable, set to U+FFFD in case of an error
287 * @see U8_GET
288 * @stable ICU 51
289 */
290#define U8_GET_OR_FFFD(s, start, i, length, c) { \
291 int32_t _u8_get_index=(i); \
292 U8_SET_CP_START(s, start, _u8_get_index); \
293 U8_NEXT_OR_FFFD(s, _u8_get_index, length, c); \
294}
295
296/* definitions with forward iteration --------------------------------------- */
297
298/**
299 * Get a code point from a string at a code point boundary offset,
300 * and advance the offset to the next code point boundary.
301 * (Post-incrementing forward iteration.)
302 * "Unsafe" macro, assumes well-formed UTF-8.
303 *
304 * The offset may point to the lead byte of a multi-byte sequence,
305 * in which case the macro will read the whole sequence.
306 * The result is undefined if the offset points to a trail byte
307 * or an illegal UTF-8 sequence.
308 *
309 * @param s const uint8_t * string
310 * @param i string offset
311 * @param c output UChar32 variable
312 * @see U8_NEXT
313 * @stable ICU 2.4
314 */
315#define U8_NEXT_UNSAFE(s, i, c) { \
316 (c)=(uint8_t)(s)[(i)++]; \
317 if(!U8_IS_SINGLE(c)) { \
318 if((c)<0xe0) { \
319 (c)=(((c)&0x1f)<<6)|((s)[(i)++]&0x3f); \
320 } else if((c)<0xf0) { \
321 /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
322 (c)=(UChar)(((c)<<12)|(((s)[i]&0x3f)<<6)|((s)[(i)+1]&0x3f)); \
323 (i)+=2; \
324 } else { \
325 (c)=(((c)&7)<<18)|(((s)[i]&0x3f)<<12)|(((s)[(i)+1]&0x3f)<<6)|((s)[(i)+2]&0x3f); \
326 (i)+=3; \
327 } \
328 } \
329}
330
331/**
332 * Get a code point from a string at a code point boundary offset,
333 * and advance the offset to the next code point boundary.
334 * (Post-incrementing forward iteration.)
335 * "Safe" macro, checks for illegal sequences and for string boundaries.
336 *
337 * The length can be negative for a NUL-terminated string.
338 *
339 * The offset may point to the lead byte of a multi-byte sequence,
340 * in which case the macro will read the whole sequence.
341 * If the offset points to a trail byte or an illegal UTF-8 sequence, then
342 * c is set to a negative value.
343 *
344 * @param s const uint8_t * string
345 * @param i int32_t string offset, must be i<length
346 * @param length int32_t string length
347 * @param c output UChar32 variable, set to <0 in case of an error
348 * @see U8_NEXT_UNSAFE
349 * @stable ICU 2.4
350 */
351#define U8_NEXT(s, i, length, c) { \
352 (c)=(uint8_t)(s)[(i)++]; \
353 if(!U8_IS_SINGLE(c)) { \
354 uint8_t __t1, __t2; \
355 if( /* handle U+0800..U+FFFF inline */ \
356 (0xe0<=(c) && (c)<0xf0) && \
357 (((i)+1)<(length) || (length)<0) && \
358 U8_IS_VALID_LEAD3_AND_T1((c), __t1=(s)[i]) && \
359 (__t2=(s)[(i)+1]-0x80)<=0x3f) { \
360 (c)=(((c)&0xf)<<12)|((__t1&0x3f)<<6)|__t2; \
361 (i)+=2; \
362 } else if( /* handle U+0080..U+07FF inline */ \
363 ((c)<0xe0 && (c)>=0xc2) && \
364 ((i)!=(length)) && \
365 (__t1=(s)[i]-0x80)<=0x3f) { \
366 (c)=(((c)&0x1f)<<6)|__t1; \
367 ++(i); \
368 } else { \
369 /* function call for "complicated" and error cases */ \
370 (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -1); \
371 } \
372 } \
373}
374
375/**
376 * Get a code point from a string at a code point boundary offset,
377 * and advance the offset to the next code point boundary.
378 * (Post-incrementing forward iteration.)
379 * "Safe" macro, checks for illegal sequences and for string boundaries.
380 *
381 * The length can be negative for a NUL-terminated string.
382 *
383 * The offset may point to the lead byte of a multi-byte sequence,
384 * in which case the macro will read the whole sequence.
385 * If the offset points to a trail byte or an illegal UTF-8 sequence, then
386 * c is set to U+FFFD.
387 *
388 * This macro does not distinguish between a real U+FFFD in the text
389 * and U+FFFD returned for an ill-formed sequence.
390 * Use U8_NEXT() if that distinction is important.
391 *
392 * @param s const uint8_t * string
393 * @param i int32_t string offset, must be i<length
394 * @param length int32_t string length
395 * @param c output UChar32 variable, set to U+FFFD in case of an error
396 * @see U8_NEXT
397 * @stable ICU 51
398 */
399#define U8_NEXT_OR_FFFD(s, i, length, c) { \
400 (c)=(uint8_t)(s)[(i)++]; \
401 if(!U8_IS_SINGLE(c)) { \
402 uint8_t __t1, __t2; \
403 if( /* handle U+0800..U+FFFF inline */ \
404 (0xe0<=(c) && (c)<0xf0) && \
405 (((i)+1)<(length) || (length)<0) && \
406 U8_IS_VALID_LEAD3_AND_T1((c), __t1=(s)[i]) && \
407 (__t2=(s)[(i)+1]-0x80)<=0x3f) { \
408 (c)=(((c)&0xf)<<12)|((__t1&0x3f)<<6)|__t2; \
409 (i)+=2; \
410 } else if( /* handle U+0080..U+07FF inline */ \
411 ((c)<0xe0 && (c)>=0xc2) && \
412 ((i)!=(length)) && \
413 (__t1=(s)[i]-0x80)<=0x3f) { \
414 (c)=(((c)&0x1f)<<6)|__t1; \
415 ++(i); \
416 } else { \
417 /* function call for "complicated" and error cases */ \
418 (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -3); \
419 } \
420 } \
421}
422
423/**
424 * Append a code point to a string, overwriting 1 to 4 bytes.
425 * The offset points to the current end of the string contents
426 * and is advanced (post-increment).
427 * "Unsafe" macro, assumes a valid code point and sufficient space in the string.
428 * Otherwise, the result is undefined.
429 *
430 * @param s const uint8_t * string buffer
431 * @param i string offset
432 * @param c code point to append
433 * @see U8_APPEND
434 * @stable ICU 2.4
435 */
436#define U8_APPEND_UNSAFE(s, i, c) { \
437 if((uint32_t)(c)<=0x7f) { \
438 (s)[(i)++]=(uint8_t)(c); \
439 } else { \
440 if((uint32_t)(c)<=0x7ff) { \
441 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
442 } else { \
443 if((uint32_t)(c)<=0xffff) { \
444 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
445 } else { \
446 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
447 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
448 } \
449 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
450 } \
451 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
452 } \
453}
454
455/**
456 * Append a code point to a string, overwriting 1 to 4 bytes.
457 * The offset points to the current end of the string contents
458 * and is advanced (post-increment).
459 * "Safe" macro, checks for a valid code point.
460 * If a non-ASCII code point is written, checks for sufficient space in the string.
461 * If the code point is not valid or trail bytes do not fit,
462 * then isError is set to TRUE.
463 *
464 * @param s const uint8_t * string buffer
465 * @param i int32_t string offset, must be i<capacity
466 * @param capacity int32_t size of the string buffer
467 * @param c UChar32 code point to append
468 * @param isError output UBool set to TRUE if an error occurs, otherwise not modified
469 * @see U8_APPEND_UNSAFE
470 * @stable ICU 2.4
471 */
472#define U8_APPEND(s, i, capacity, c, isError) { \
473 if((uint32_t)(c)<=0x7f) { \
474 (s)[(i)++]=(uint8_t)(c); \
475 } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \
476 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
477 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
478 } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \
479 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
480 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
481 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
482 } else { \
483 (i)=utf8_appendCharSafeBody(s, (i), (capacity), c, &(isError)); \
484 } \
485}
486
487/**
488 * Advance the string offset from one code point boundary to the next.
489 * (Post-incrementing iteration.)
490 * "Unsafe" macro, assumes well-formed UTF-8.
491 *
492 * @param s const uint8_t * string
493 * @param i string offset
494 * @see U8_FWD_1
495 * @stable ICU 2.4
496 */
497#define U8_FWD_1_UNSAFE(s, i) { \
498 (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((s)[i]); \
499}
500
501/**
502 * Advance the string offset from one code point boundary to the next.
503 * (Post-incrementing iteration.)
504 * "Safe" macro, checks for illegal sequences and for string boundaries.
505 *
506 * The length can be negative for a NUL-terminated string.
507 *
508 * @param s const uint8_t * string
509 * @param i int32_t string offset, must be i<length
510 * @param length int32_t string length
511 * @see U8_FWD_1_UNSAFE
512 * @stable ICU 2.4
513 */
514#define U8_FWD_1(s, i, length) { \
515 uint8_t __b=(s)[(i)++]; \
516 if(U8_IS_LEAD(__b) && (i)!=(length)) { \
517 uint8_t __t1=(s)[i]; \
518 if((0xe0<=__b && __b<0xf0)) { \
519 if(U8_IS_VALID_LEAD3_AND_T1(__b, __t1) && \
520 ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
521 ++(i); \
522 } \
523 } else if(__b<0xe0) { \
524 if(U8_IS_TRAIL(__t1)) { \
525 ++(i); \
526 } \
527 } else /* c>=0xf0 */ { \
528 if(U8_IS_VALID_LEAD4_AND_T1(__b, __t1) && \
529 ++(i)!=(length) && U8_IS_TRAIL((s)[i]) && \
530 ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
531 ++(i); \
532 } \
533 } \
534 } \
535}
536
537/**
538 * Advance the string offset from one code point boundary to the n-th next one,
539 * i.e., move forward by n code points.
540 * (Post-incrementing iteration.)
541 * "Unsafe" macro, assumes well-formed UTF-8.
542 *
543 * @param s const uint8_t * string
544 * @param i string offset
545 * @param n number of code points to skip
546 * @see U8_FWD_N
547 * @stable ICU 2.4
548 */
549#define U8_FWD_N_UNSAFE(s, i, n) { \
550 int32_t __N=(n); \
551 while(__N>0) { \
552 U8_FWD_1_UNSAFE(s, i); \
553 --__N; \
554 } \
555}
556
557/**
558 * Advance the string offset from one code point boundary to the n-th next one,
559 * i.e., move forward by n code points.
560 * (Post-incrementing iteration.)
561 * "Safe" macro, checks for illegal sequences and for string boundaries.
562 *
563 * The length can be negative for a NUL-terminated string.
564 *
565 * @param s const uint8_t * string
566 * @param i int32_t string offset, must be i<length
567 * @param length int32_t string length
568 * @param n number of code points to skip
569 * @see U8_FWD_N_UNSAFE
570 * @stable ICU 2.4
571 */
572#define U8_FWD_N(s, i, length, n) { \
573 int32_t __N=(n); \
574 while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
575 U8_FWD_1(s, i, length); \
576 --__N; \
577 } \
578}
579
580/**
581 * Adjust a random-access offset to a code point boundary
582 * at the start of a code point.
583 * If the offset points to a UTF-8 trail byte,
584 * then the offset is moved backward to the corresponding lead byte.
585 * Otherwise, it is not modified.
586 * "Unsafe" macro, assumes well-formed UTF-8.
587 *
588 * @param s const uint8_t * string
589 * @param i string offset
590 * @see U8_SET_CP_START
591 * @stable ICU 2.4
592 */
593#define U8_SET_CP_START_UNSAFE(s, i) { \
594 while(U8_IS_TRAIL((s)[i])) { --(i); } \
595}
596
597/**
598 * Adjust a random-access offset to a code point boundary
599 * at the start of a code point.
600 * If the offset points to a UTF-8 trail byte,
601 * then the offset is moved backward to the corresponding lead byte.
602 * Otherwise, it is not modified.
603 * "Safe" macro, checks for illegal sequences and for string boundaries.
604 *
605 * @param s const uint8_t * string
606 * @param start int32_t starting string offset (usually 0)
607 * @param i int32_t string offset, must be start<=i
608 * @see U8_SET_CP_START_UNSAFE
609 * @stable ICU 2.4
610 */
611#define U8_SET_CP_START(s, start, i) { \
612 if(U8_IS_TRAIL((s)[(i)])) { \
613 (i)=utf8_back1SafeBody(s, start, (i)); \
614 } \
615}
616
617/* definitions with backward iteration -------------------------------------- */
618
619/**
620 * Move the string offset from one code point boundary to the previous one
621 * and get the code point between them.
622 * (Pre-decrementing backward iteration.)
623 * "Unsafe" macro, assumes well-formed UTF-8.
624 *
625 * The input offset may be the same as the string length.
626 * If the offset is behind a multi-byte sequence, then the macro will read
627 * the whole sequence.
628 * If the offset is behind a lead byte, then that itself
629 * will be returned as the code point.
630 * The result is undefined if the offset is behind an illegal UTF-8 sequence.
631 *
632 * @param s const uint8_t * string
633 * @param i string offset
634 * @param c output UChar32 variable
635 * @see U8_PREV
636 * @stable ICU 2.4
637 */
638#define U8_PREV_UNSAFE(s, i, c) { \
639 (c)=(uint8_t)(s)[--(i)]; \
640 if(U8_IS_TRAIL(c)) { \
641 uint8_t __b, __count=1, __shift=6; \
642\
643 /* c is a trail byte */ \
644 (c)&=0x3f; \
645 for(;;) { \
646 __b=(s)[--(i)]; \
647 if(__b>=0xc0) { \
648 U8_MASK_LEAD_BYTE(__b, __count); \
649 (c)|=(UChar32)__b<<__shift; \
650 break; \
651 } else { \
652 (c)|=(UChar32)(__b&0x3f)<<__shift; \
653 ++__count; \
654 __shift+=6; \
655 } \
656 } \
657 } \
658}
659
660/**
661 * Move the string offset from one code point boundary to the previous one
662 * and get the code point between them.
663 * (Pre-decrementing backward iteration.)
664 * "Safe" macro, checks for illegal sequences and for string boundaries.
665 *
666 * The input offset may be the same as the string length.
667 * If the offset is behind a multi-byte sequence, then the macro will read
668 * the whole sequence.
669 * If the offset is behind a lead byte, then that itself
670 * will be returned as the code point.
671 * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value.
672 *
673 * @param s const uint8_t * string
674 * @param start int32_t starting string offset (usually 0)
675 * @param i int32_t string offset, must be start<i
676 * @param c output UChar32 variable, set to <0 in case of an error
677 * @see U8_PREV_UNSAFE
678 * @stable ICU 2.4
679 */
680#define U8_PREV(s, start, i, c) { \
681 (c)=(uint8_t)(s)[--(i)]; \
682 if(!U8_IS_SINGLE(c)) { \
683 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
684 } \
685}
686
687/**
688 * Move the string offset from one code point boundary to the previous one
689 * and get the code point between them.
690 * (Pre-decrementing backward iteration.)
691 * "Safe" macro, checks for illegal sequences and for string boundaries.
692 *
693 * The input offset may be the same as the string length.
694 * If the offset is behind a multi-byte sequence, then the macro will read
695 * the whole sequence.
696 * If the offset is behind a lead byte, then that itself
697 * will be returned as the code point.
698 * If the offset is behind an illegal UTF-8 sequence, then c is set to U+FFFD.
699 *
700 * This macro does not distinguish between a real U+FFFD in the text
701 * and U+FFFD returned for an ill-formed sequence.
702 * Use U8_PREV() if that distinction is important.
703 *
704 * @param s const uint8_t * string
705 * @param start int32_t starting string offset (usually 0)
706 * @param i int32_t string offset, must be start<i
707 * @param c output UChar32 variable, set to U+FFFD in case of an error
708 * @see U8_PREV
709 * @stable ICU 51
710 */
711#define U8_PREV_OR_FFFD(s, start, i, c) { \
712 (c)=(uint8_t)(s)[--(i)]; \
713 if(!U8_IS_SINGLE(c)) { \
714 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -3); \
715 } \
716}
717
718/**
719 * Move the string offset from one code point boundary to the previous one.
720 * (Pre-decrementing backward iteration.)
721 * The input offset may be the same as the string length.
722 * "Unsafe" macro, assumes well-formed UTF-8.
723 *
724 * @param s const uint8_t * string
725 * @param i string offset
726 * @see U8_BACK_1
727 * @stable ICU 2.4
728 */
729#define U8_BACK_1_UNSAFE(s, i) { \
730 while(U8_IS_TRAIL((s)[--(i)])) {} \
731}
732
733/**
734 * Move the string offset from one code point boundary to the previous one.
735 * (Pre-decrementing backward iteration.)
736 * The input offset may be the same as the string length.
737 * "Safe" macro, checks for illegal sequences and for string boundaries.
738 *
739 * @param s const uint8_t * string
740 * @param start int32_t starting string offset (usually 0)
741 * @param i int32_t string offset, must be start<i
742 * @see U8_BACK_1_UNSAFE
743 * @stable ICU 2.4
744 */
745#define U8_BACK_1(s, start, i) { \
746 if(U8_IS_TRAIL((s)[--(i)])) { \
747 (i)=utf8_back1SafeBody(s, start, (i)); \
748 } \
749}
750
751/**
752 * Move the string offset from one code point boundary to the n-th one before it,
753 * i.e., move backward by n code points.
754 * (Pre-decrementing backward iteration.)
755 * The input offset may be the same as the string length.
756 * "Unsafe" macro, assumes well-formed UTF-8.
757 *
758 * @param s const uint8_t * string
759 * @param i string offset
760 * @param n number of code points to skip
761 * @see U8_BACK_N
762 * @stable ICU 2.4
763 */
764#define U8_BACK_N_UNSAFE(s, i, n) { \
765 int32_t __N=(n); \
766 while(__N>0) { \
767 U8_BACK_1_UNSAFE(s, i); \
768 --__N; \
769 } \
770}
771
772/**
773 * Move the string offset from one code point boundary to the n-th one before it,
774 * i.e., move backward by n code points.
775 * (Pre-decrementing backward iteration.)
776 * The input offset may be the same as the string length.
777 * "Safe" macro, checks for illegal sequences and for string boundaries.
778 *
779 * @param s const uint8_t * string
780 * @param start int32_t index of the start of the string
781 * @param i int32_t string offset, must be start<i
782 * @param n number of code points to skip
783 * @see U8_BACK_N_UNSAFE
784 * @stable ICU 2.4
785 */
786#define U8_BACK_N(s, start, i, n) { \
787 int32_t __N=(n); \
788 while(__N>0 && (i)>(start)) { \
789 U8_BACK_1(s, start, i); \
790 --__N; \
791 } \
792}
793
794/**
795 * Adjust a random-access offset to a code point boundary after a code point.
796 * If the offset is behind a partial multi-byte sequence,
797 * then the offset is incremented to behind the whole sequence.
798 * Otherwise, it is not modified.
799 * The input offset may be the same as the string length.
800 * "Unsafe" macro, assumes well-formed UTF-8.
801 *
802 * @param s const uint8_t * string
803 * @param i string offset
804 * @see U8_SET_CP_LIMIT
805 * @stable ICU 2.4
806 */
807#define U8_SET_CP_LIMIT_UNSAFE(s, i) { \
808 U8_BACK_1_UNSAFE(s, i); \
809 U8_FWD_1_UNSAFE(s, i); \
810}
811
812/**
813 * Adjust a random-access offset to a code point boundary after a code point.
814 * If the offset is behind a partial multi-byte sequence,
815 * then the offset is incremented to behind the whole sequence.
816 * Otherwise, it is not modified.
817 * The input offset may be the same as the string length.
818 * "Safe" macro, checks for illegal sequences and for string boundaries.
819 *
820 * The length can be negative for a NUL-terminated string.
821 *
822 * @param s const uint8_t * string
823 * @param start int32_t starting string offset (usually 0)
824 * @param i int32_t string offset, must be start<=i<=length
825 * @param length int32_t string length
826 * @see U8_SET_CP_LIMIT_UNSAFE
827 * @stable ICU 2.4
828 */
829#define U8_SET_CP_LIMIT(s, start, i, length) { \
830 if((start)<(i) && ((i)<(length) || (length)<0)) { \
831 U8_BACK_1(s, start, i); \
832 U8_FWD_1(s, i, length); \
833 } \
834}
835
836#endif
837