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 * (https://unicode-org.github.io/icu/userguide/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 <stdbool.h>
38#include "unicode/umachine.h"
39#ifndef __UTF_H__
40# include "unicode/utf.h"
41#endif
42
43/* internal definitions ----------------------------------------------------- */
44
45/**
46 * Counts the trail bytes for a UTF-8 lead byte.
47 * Returns 0 for 0..0xc1 as well as for 0xf5..0xff.
48 * leadByte might be evaluated multiple times.
49 *
50 * This is internal since it is not meant to be called directly by external clients;
51 * however it is called by public macros in this file and thus must remain stable.
52 *
53 * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
54 * @internal
55 */
56#define U8_COUNT_TRAIL_BYTES(leadByte) \
57 (U8_IS_LEAD(leadByte) ? \
58 ((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0)+1 : 0)
59
60/**
61 * Counts the trail bytes for a UTF-8 lead byte of a valid UTF-8 sequence.
62 * Returns 0 for 0..0xc1. Undefined for 0xf5..0xff.
63 * leadByte might be evaluated multiple times.
64 *
65 * This is internal since it is not meant to be called directly by external clients;
66 * however it is called by public macros in this file and thus must remain stable.
67 *
68 * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
69 * @internal
70 */
71#define U8_COUNT_TRAIL_BYTES_UNSAFE(leadByte) \
72 (((uint8_t)(leadByte)>=0xc2)+((uint8_t)(leadByte)>=0xe0)+((uint8_t)(leadByte)>=0xf0))
73
74/**
75 * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value.
76 *
77 * This is internal since it is not meant to be called directly by external clients;
78 * however it is called by public macros in this file and thus must remain stable.
79 * @internal
80 */
81#define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
82
83/**
84 * Internal bit vector for 3-byte UTF-8 validity check, for use in U8_IS_VALID_LEAD3_AND_T1.
85 * Each bit indicates whether one lead byte + first trail byte pair starts a valid sequence.
86 * Lead byte E0..EF bits 3..0 are used as byte index,
87 * first trail byte bits 7..5 are used as bit index into that byte.
88 * @see U8_IS_VALID_LEAD3_AND_T1
89 * @internal
90 */
91#define U8_LEAD3_T1_BITS "\x20\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x10\x30\x30"
92
93/**
94 * Internal 3-byte UTF-8 validity check.
95 * Non-zero if lead byte E0..EF and first trail byte 00..FF start a valid sequence.
96 * @internal
97 */
98#define U8_IS_VALID_LEAD3_AND_T1(lead, t1) (U8_LEAD3_T1_BITS[(lead)&0xf]&(1<<((uint8_t)(t1)>>5)))
99
100/**
101 * Internal bit vector for 4-byte UTF-8 validity check, for use in U8_IS_VALID_LEAD4_AND_T1.
102 * Each bit indicates whether one lead byte + first trail byte pair starts a valid sequence.
103 * First trail byte bits 7..4 are used as byte index,
104 * lead byte F0..F4 bits 2..0 are used as bit index into that byte.
105 * @see U8_IS_VALID_LEAD4_AND_T1
106 * @internal
107 */
108#define U8_LEAD4_T1_BITS "\x00\x00\x00\x00\x00\x00\x00\x00\x1E\x0F\x0F\x0F\x00\x00\x00\x00"
109
110/**
111 * Internal 4-byte UTF-8 validity check.
112 * Non-zero if lead byte F0..F4 and first trail byte 00..FF start a valid sequence.
113 * @internal
114 */
115#define U8_IS_VALID_LEAD4_AND_T1(lead, t1) (U8_LEAD4_T1_BITS[(uint8_t)(t1)>>4]&(1<<((lead)&7)))
116
117/**
118 * Function for handling "next code point" with error-checking.
119 *
120 * This is internal since it is not meant to be called directly by external clients;
121 * however it is called by public macros in this
122 * file and thus must remain stable, and should not be hidden when other internal
123 * functions are hidden (otherwise public macros would fail to compile).
124 * @internal
125 */
126U_CAPI UChar32 U_EXPORT2
127utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
128
129/**
130 * Function for handling "append code point" with error-checking.
131 *
132 * This is internal since it is not meant to be called directly by external clients;
133 * however it is called by public macros in this
134 * file and thus must remain stable, and should not be hidden when other internal
135 * functions are hidden (otherwise public macros would fail to compile).
136 * @internal
137 */
138U_CAPI int32_t U_EXPORT2
139utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
140
141/**
142 * Function for handling "previous code point" with error-checking.
143 *
144 * This is internal since it is not meant to be called directly by external clients;
145 * however it is called by public macros in this
146 * file and thus must remain stable, and should not be hidden when other internal
147 * functions are hidden (otherwise public macros would fail to compile).
148 * @internal
149 */
150U_CAPI UChar32 U_EXPORT2
151utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
152
153/**
154 * Function for handling "skip backward one code point" with error-checking.
155 *
156 * This is internal since it is not meant to be called directly by external clients;
157 * however it is called by public macros in this
158 * file and thus must remain stable, and should not be hidden when other internal
159 * functions are hidden (otherwise public macros would fail to compile).
160 * @internal
161 */
162U_CAPI int32_t U_EXPORT2
163utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
164
165/* single-code point definitions -------------------------------------------- */
166
167/**
168 * Does this code unit (byte) encode a code point by itself (US-ASCII 0..0x7f)?
169 * @param c 8-bit code unit (byte)
170 * @return true or false
171 * @stable ICU 2.4
172 */
173#define U8_IS_SINGLE(c) (((c)&0x80)==0)
174
175/**
176 * Is this code unit (byte) a UTF-8 lead byte? (0xC2..0xF4)
177 * @param c 8-bit code unit (byte)
178 * @return true or false
179 * @stable ICU 2.4
180 */
181#define U8_IS_LEAD(c) ((uint8_t)((c)-0xc2)<=0x32)
182// 0x32=0xf4-0xc2
183
184/**
185 * Is this code unit (byte) a UTF-8 trail byte? (0x80..0xBF)
186 * @param c 8-bit code unit (byte)
187 * @return true or false
188 * @stable ICU 2.4
189 */
190#define U8_IS_TRAIL(c) ((int8_t)(c)<-0x40)
191
192/**
193 * How many code units (bytes) are used for the UTF-8 encoding
194 * of this Unicode code point?
195 * @param c 32-bit code point
196 * @return 1..4, or 0 if c is a surrogate or not a Unicode code point
197 * @stable ICU 2.4
198 */
199#define U8_LENGTH(c) \
200 ((uint32_t)(c)<=0x7f ? 1 : \
201 ((uint32_t)(c)<=0x7ff ? 2 : \
202 ((uint32_t)(c)<=0xd7ff ? 3 : \
203 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
204 ((uint32_t)(c)<=0xffff ? 3 : 4)\
205 ) \
206 ) \
207 ) \
208 )
209
210/**
211 * The maximum number of UTF-8 code units (bytes) per Unicode code point (U+0000..U+10ffff).
212 * @return 4
213 * @stable ICU 2.4
214 */
215#define U8_MAX_LENGTH 4
216
217/**
218 * Get a code point from a string at a random-access offset,
219 * without changing the offset.
220 * The offset may point to either the lead byte or one of the trail bytes
221 * for a code point, in which case the macro will read all of the bytes
222 * for the code point.
223 * The result is undefined if the offset points to an illegal UTF-8
224 * byte sequence.
225 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
226 *
227 * @param s const uint8_t * string
228 * @param i string offset
229 * @param c output UChar32 variable
230 * @see U8_GET
231 * @stable ICU 2.4
232 */
233#define U8_GET_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
234 int32_t _u8_get_unsafe_index=(int32_t)(i); \
235 U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
236 U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
237} UPRV_BLOCK_MACRO_END
238
239/**
240 * Get a code point from a string at a random-access offset,
241 * without changing the offset.
242 * The offset may point to either the lead byte or one of the trail bytes
243 * for a code point, in which case the macro will read all of the bytes
244 * for the code point.
245 *
246 * The length can be negative for a NUL-terminated string.
247 *
248 * If the offset points to an illegal UTF-8 byte sequence, then
249 * c is set to a negative value.
250 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
251 *
252 * @param s const uint8_t * string
253 * @param start int32_t starting string offset
254 * @param i int32_t string offset, must be start<=i<length
255 * @param length int32_t string length
256 * @param c output UChar32 variable, set to <0 in case of an error
257 * @see U8_GET_UNSAFE
258 * @stable ICU 2.4
259 */
260#define U8_GET(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
261 int32_t _u8_get_index=(i); \
262 U8_SET_CP_START(s, start, _u8_get_index); \
263 U8_NEXT(s, _u8_get_index, length, c); \
264} UPRV_BLOCK_MACRO_END
265
266/**
267 * Get a code point from a string at a random-access offset,
268 * without changing the offset.
269 * The offset may point to either the lead byte or one of the trail bytes
270 * for a code point, in which case the macro will read all of the bytes
271 * for the code point.
272 *
273 * The length can be negative for a NUL-terminated string.
274 *
275 * If the offset points to an illegal UTF-8 byte sequence, then
276 * c is set to U+FFFD.
277 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT_OR_FFFD.
278 *
279 * This macro does not distinguish between a real U+FFFD in the text
280 * and U+FFFD returned for an ill-formed sequence.
281 * Use U8_GET() if that distinction is important.
282 *
283 * @param s const uint8_t * string
284 * @param start int32_t starting string offset
285 * @param i int32_t string offset, must be start<=i<length
286 * @param length int32_t string length
287 * @param c output UChar32 variable, set to U+FFFD in case of an error
288 * @see U8_GET
289 * @stable ICU 51
290 */
291#define U8_GET_OR_FFFD(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \
292 int32_t _u8_get_index=(i); \
293 U8_SET_CP_START(s, start, _u8_get_index); \
294 U8_NEXT_OR_FFFD(s, _u8_get_index, length, c); \
295} UPRV_BLOCK_MACRO_END
296
297/* definitions with forward iteration --------------------------------------- */
298
299/**
300 * Get a code point from a string at a code point boundary offset,
301 * and advance the offset to the next code point boundary.
302 * (Post-incrementing forward iteration.)
303 * "Unsafe" macro, assumes well-formed UTF-8.
304 *
305 * The offset may point to the lead byte of a multi-byte sequence,
306 * in which case the macro will read the whole sequence.
307 * The result is undefined if the offset points to a trail byte
308 * or an illegal UTF-8 sequence.
309 *
310 * @param s const uint8_t * string
311 * @param i string offset
312 * @param c output UChar32 variable
313 * @see U8_NEXT
314 * @stable ICU 2.4
315 */
316#define U8_NEXT_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
317 (c)=(uint8_t)(s)[(i)++]; \
318 if(!U8_IS_SINGLE(c)) { \
319 if((c)<0xe0) { \
320 (c)=(((c)&0x1f)<<6)|((s)[(i)++]&0x3f); \
321 } else if((c)<0xf0) { \
322 /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
323 (c)=(UChar)(((c)<<12)|(((s)[i]&0x3f)<<6)|((s)[(i)+1]&0x3f)); \
324 (i)+=2; \
325 } else { \
326 (c)=(((c)&7)<<18)|(((s)[i]&0x3f)<<12)|(((s)[(i)+1]&0x3f)<<6)|((s)[(i)+2]&0x3f); \
327 (i)+=3; \
328 } \
329 } \
330} UPRV_BLOCK_MACRO_END
331
332/**
333 * Get a code point from a string at a code point boundary offset,
334 * and advance the offset to the next code point boundary.
335 * (Post-incrementing forward iteration.)
336 * "Safe" macro, checks for illegal sequences and for string boundaries.
337 *
338 * The length can be negative for a NUL-terminated string.
339 *
340 * The offset may point to the lead byte of a multi-byte sequence,
341 * in which case the macro will read the whole sequence.
342 * If the offset points to a trail byte or an illegal UTF-8 sequence, then
343 * c is set to a negative value.
344 *
345 * @param s const uint8_t * string
346 * @param i int32_t string offset, must be i<length
347 * @param length int32_t string length
348 * @param c output UChar32 variable, set to <0 in case of an error
349 * @see U8_NEXT_UNSAFE
350 * @stable ICU 2.4
351 */
352#define U8_NEXT(s, i, length, c) U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, U_SENTINEL)
353
354/**
355 * Get a code point from a string at a code point boundary offset,
356 * and advance the offset to the next code point boundary.
357 * (Post-incrementing forward iteration.)
358 * "Safe" macro, checks for illegal sequences and for string boundaries.
359 *
360 * The length can be negative for a NUL-terminated string.
361 *
362 * The offset may point to the lead byte of a multi-byte sequence,
363 * in which case the macro will read the whole sequence.
364 * If the offset points to a trail byte or an illegal UTF-8 sequence, then
365 * c is set to U+FFFD.
366 *
367 * This macro does not distinguish between a real U+FFFD in the text
368 * and U+FFFD returned for an ill-formed sequence.
369 * Use U8_NEXT() if that distinction is important.
370 *
371 * @param s const uint8_t * string
372 * @param i int32_t string offset, must be i<length
373 * @param length int32_t string length
374 * @param c output UChar32 variable, set to U+FFFD in case of an error
375 * @see U8_NEXT
376 * @stable ICU 51
377 */
378#define U8_NEXT_OR_FFFD(s, i, length, c) U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, 0xfffd)
379
380/** @internal */
381#define U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, sub) UPRV_BLOCK_MACRO_BEGIN { \
382 (c)=(uint8_t)(s)[(i)++]; \
383 if(!U8_IS_SINGLE(c)) { \
384 uint8_t __t = 0; \
385 if((i)!=(length) && \
386 /* fetch/validate/assemble all but last trail byte */ \
387 ((c)>=0xe0 ? \
388 ((c)<0xf0 ? /* U+0800..U+FFFF except surrogates */ \
389 U8_LEAD3_T1_BITS[(c)&=0xf]&(1<<((__t=(s)[i])>>5)) && \
390 (__t&=0x3f, 1) \
391 : /* U+10000..U+10FFFF */ \
392 ((c)-=0xf0)<=4 && \
393 U8_LEAD4_T1_BITS[(__t=(s)[i])>>4]&(1<<(c)) && \
394 ((c)=((c)<<6)|(__t&0x3f), ++(i)!=(length)) && \
395 (__t=(s)[i]-0x80)<=0x3f) && \
396 /* valid second-to-last trail byte */ \
397 ((c)=((c)<<6)|__t, ++(i)!=(length)) \
398 : /* U+0080..U+07FF */ \
399 (c)>=0xc2 && ((c)&=0x1f, 1)) && \
400 /* last trail byte */ \
401 (__t=(s)[i]-0x80)<=0x3f && \
402 ((c)=((c)<<6)|__t, ++(i), 1)) { \
403 } else { \
404 (c)=(sub); /* ill-formed*/ \
405 } \
406 } \
407} UPRV_BLOCK_MACRO_END
408
409/**
410 * Append a code point to a string, overwriting 1 to 4 bytes.
411 * The offset points to the current end of the string contents
412 * and is advanced (post-increment).
413 * "Unsafe" macro, assumes a valid code point and sufficient space in the string.
414 * Otherwise, the result is undefined.
415 *
416 * @param s const uint8_t * string buffer
417 * @param i string offset
418 * @param c code point to append
419 * @see U8_APPEND
420 * @stable ICU 2.4
421 */
422#define U8_APPEND_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
423 uint32_t __uc=(c); \
424 if(__uc<=0x7f) { \
425 (s)[(i)++]=(uint8_t)__uc; \
426 } else { \
427 if(__uc<=0x7ff) { \
428 (s)[(i)++]=(uint8_t)((__uc>>6)|0xc0); \
429 } else { \
430 if(__uc<=0xffff) { \
431 (s)[(i)++]=(uint8_t)((__uc>>12)|0xe0); \
432 } else { \
433 (s)[(i)++]=(uint8_t)((__uc>>18)|0xf0); \
434 (s)[(i)++]=(uint8_t)(((__uc>>12)&0x3f)|0x80); \
435 } \
436 (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
437 } \
438 (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
439 } \
440} UPRV_BLOCK_MACRO_END
441
442/**
443 * Append a code point to a string, overwriting 1 to 4 bytes.
444 * The offset points to the current end of the string contents
445 * and is advanced (post-increment).
446 * "Safe" macro, checks for a valid code point.
447 * If a non-ASCII code point is written, checks for sufficient space in the string.
448 * If the code point is not valid or trail bytes do not fit,
449 * then isError is set to true.
450 *
451 * @param s const uint8_t * string buffer
452 * @param i int32_t string offset, must be i<capacity
453 * @param capacity int32_t size of the string buffer
454 * @param c UChar32 code point to append
455 * @param isError output UBool set to true if an error occurs, otherwise not modified
456 * @see U8_APPEND_UNSAFE
457 * @stable ICU 2.4
458 */
459#define U8_APPEND(s, i, capacity, c, isError) UPRV_BLOCK_MACRO_BEGIN { \
460 uint32_t __uc=(c); \
461 if(__uc<=0x7f) { \
462 (s)[(i)++]=(uint8_t)__uc; \
463 } else if(__uc<=0x7ff && (i)+1<(capacity)) { \
464 (s)[(i)++]=(uint8_t)((__uc>>6)|0xc0); \
465 (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
466 } else if((__uc<=0xd7ff || (0xe000<=__uc && __uc<=0xffff)) && (i)+2<(capacity)) { \
467 (s)[(i)++]=(uint8_t)((__uc>>12)|0xe0); \
468 (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
469 (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
470 } else if(0xffff<__uc && __uc<=0x10ffff && (i)+3<(capacity)) { \
471 (s)[(i)++]=(uint8_t)((__uc>>18)|0xf0); \
472 (s)[(i)++]=(uint8_t)(((__uc>>12)&0x3f)|0x80); \
473 (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
474 (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
475 } else { \
476 (isError)=true; \
477 } \
478} UPRV_BLOCK_MACRO_END
479
480/**
481 * Advance the string offset from one code point boundary to the next.
482 * (Post-incrementing iteration.)
483 * "Unsafe" macro, assumes well-formed UTF-8.
484 *
485 * @param s const uint8_t * string
486 * @param i string offset
487 * @see U8_FWD_1
488 * @stable ICU 2.4
489 */
490#define U8_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
491 (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((s)[i]); \
492} UPRV_BLOCK_MACRO_END
493
494/**
495 * Advance the string offset from one code point boundary to the next.
496 * (Post-incrementing iteration.)
497 * "Safe" macro, checks for illegal sequences and for string boundaries.
498 *
499 * The length can be negative for a NUL-terminated string.
500 *
501 * @param s const uint8_t * string
502 * @param i int32_t string offset, must be i<length
503 * @param length int32_t string length
504 * @see U8_FWD_1_UNSAFE
505 * @stable ICU 2.4
506 */
507#define U8_FWD_1(s, i, length) UPRV_BLOCK_MACRO_BEGIN { \
508 uint8_t __b=(s)[(i)++]; \
509 if(U8_IS_LEAD(__b) && (i)!=(length)) { \
510 uint8_t __t1=(s)[i]; \
511 if((0xe0<=__b && __b<0xf0)) { \
512 if(U8_IS_VALID_LEAD3_AND_T1(__b, __t1) && \
513 ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
514 ++(i); \
515 } \
516 } else if(__b<0xe0) { \
517 if(U8_IS_TRAIL(__t1)) { \
518 ++(i); \
519 } \
520 } else /* c>=0xf0 */ { \
521 if(U8_IS_VALID_LEAD4_AND_T1(__b, __t1) && \
522 ++(i)!=(length) && U8_IS_TRAIL((s)[i]) && \
523 ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
524 ++(i); \
525 } \
526 } \
527 } \
528} UPRV_BLOCK_MACRO_END
529
530/**
531 * Advance the string offset from one code point boundary to the n-th next one,
532 * i.e., move forward by n code points.
533 * (Post-incrementing iteration.)
534 * "Unsafe" macro, assumes well-formed UTF-8.
535 *
536 * @param s const uint8_t * string
537 * @param i string offset
538 * @param n number of code points to skip
539 * @see U8_FWD_N
540 * @stable ICU 2.4
541 */
542#define U8_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
543 int32_t __N=(n); \
544 while(__N>0) { \
545 U8_FWD_1_UNSAFE(s, i); \
546 --__N; \
547 } \
548} UPRV_BLOCK_MACRO_END
549
550/**
551 * Advance the string offset from one code point boundary to the n-th next one,
552 * i.e., move forward by n code points.
553 * (Post-incrementing iteration.)
554 * "Safe" macro, checks for illegal sequences and for string boundaries.
555 *
556 * The length can be negative for a NUL-terminated string.
557 *
558 * @param s const uint8_t * string
559 * @param i int32_t string offset, must be i<length
560 * @param length int32_t string length
561 * @param n number of code points to skip
562 * @see U8_FWD_N_UNSAFE
563 * @stable ICU 2.4
564 */
565#define U8_FWD_N(s, i, length, n) UPRV_BLOCK_MACRO_BEGIN { \
566 int32_t __N=(n); \
567 while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
568 U8_FWD_1(s, i, length); \
569 --__N; \
570 } \
571} UPRV_BLOCK_MACRO_END
572
573/**
574 * Adjust a random-access offset to a code point boundary
575 * at the start of a code point.
576 * If the offset points to a UTF-8 trail byte,
577 * then the offset is moved backward to the corresponding lead byte.
578 * Otherwise, it is not modified.
579 * "Unsafe" macro, assumes well-formed UTF-8.
580 *
581 * @param s const uint8_t * string
582 * @param i string offset
583 * @see U8_SET_CP_START
584 * @stable ICU 2.4
585 */
586#define U8_SET_CP_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
587 while(U8_IS_TRAIL((s)[i])) { --(i); } \
588} UPRV_BLOCK_MACRO_END
589
590/**
591 * Adjust a random-access offset to a code point boundary
592 * at the start of a code point.
593 * If the offset points to a UTF-8 trail byte,
594 * then the offset is moved backward to the corresponding lead byte.
595 * Otherwise, it is not modified.
596 *
597 * "Safe" macro, checks for illegal sequences and for string boundaries.
598 * Unlike U8_TRUNCATE_IF_INCOMPLETE(), this macro always reads s[i].
599 *
600 * @param s const uint8_t * string
601 * @param start int32_t starting string offset (usually 0)
602 * @param i int32_t string offset, must be start<=i
603 * @see U8_SET_CP_START_UNSAFE
604 * @see U8_TRUNCATE_IF_INCOMPLETE
605 * @stable ICU 2.4
606 */
607#define U8_SET_CP_START(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
608 if(U8_IS_TRAIL((s)[(i)])) { \
609 (i)=utf8_back1SafeBody(s, start, (i)); \
610 } \
611} UPRV_BLOCK_MACRO_END
612
613/**
614 * If the string ends with a UTF-8 byte sequence that is valid so far
615 * but incomplete, then reduce the length of the string to end before
616 * the lead byte of that incomplete sequence.
617 * For example, if the string ends with E1 80, the length is reduced by 2.
618 *
619 * In all other cases (the string ends with a complete sequence, or it is not
620 * possible for any further trail byte to extend the trailing sequence)
621 * the length remains unchanged.
622 *
623 * Useful for processing text split across multiple buffers
624 * (save the incomplete sequence for later)
625 * and for optimizing iteration
626 * (check for string length only once per character).
627 *
628 * "Safe" macro, checks for illegal sequences and for string boundaries.
629 * Unlike U8_SET_CP_START(), this macro never reads s[length].
630 *
631 * (In UTF-16, simply check for U16_IS_LEAD(last code unit).)
632 *
633 * @param s const uint8_t * string
634 * @param start int32_t starting string offset (usually 0)
635 * @param length int32_t string length (usually start<=length)
636 * @see U8_SET_CP_START
637 * @stable ICU 61
638 */
639#define U8_TRUNCATE_IF_INCOMPLETE(s, start, length) UPRV_BLOCK_MACRO_BEGIN { \
640 if((length)>(start)) { \
641 uint8_t __b1=s[(length)-1]; \
642 if(U8_IS_SINGLE(__b1)) { \
643 /* common ASCII character */ \
644 } else if(U8_IS_LEAD(__b1)) { \
645 --(length); \
646 } else if(U8_IS_TRAIL(__b1) && ((length)-2)>=(start)) { \
647 uint8_t __b2=s[(length)-2]; \
648 if(0xe0<=__b2 && __b2<=0xf4) { \
649 if(__b2<0xf0 ? U8_IS_VALID_LEAD3_AND_T1(__b2, __b1) : \
650 U8_IS_VALID_LEAD4_AND_T1(__b2, __b1)) { \
651 (length)-=2; \
652 } \
653 } else if(U8_IS_TRAIL(__b2) && ((length)-3)>=(start)) { \
654 uint8_t __b3=s[(length)-3]; \
655 if(0xf0<=__b3 && __b3<=0xf4 && U8_IS_VALID_LEAD4_AND_T1(__b3, __b2)) { \
656 (length)-=3; \
657 } \
658 } \
659 } \
660 } \
661} UPRV_BLOCK_MACRO_END
662
663/* definitions with backward iteration -------------------------------------- */
664
665/**
666 * Move the string offset from one code point boundary to the previous one
667 * and get the code point between them.
668 * (Pre-decrementing backward iteration.)
669 * "Unsafe" macro, assumes well-formed UTF-8.
670 *
671 * The input offset may be the same as the string length.
672 * If the offset is behind a multi-byte sequence, then the macro will read
673 * the whole sequence.
674 * If the offset is behind a lead byte, then that itself
675 * will be returned as the code point.
676 * The result is undefined if the offset is behind an illegal UTF-8 sequence.
677 *
678 * @param s const uint8_t * string
679 * @param i string offset
680 * @param c output UChar32 variable
681 * @see U8_PREV
682 * @stable ICU 2.4
683 */
684#define U8_PREV_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
685 (c)=(uint8_t)(s)[--(i)]; \
686 if(U8_IS_TRAIL(c)) { \
687 uint8_t __b, __count=1, __shift=6; \
688\
689 /* c is a trail byte */ \
690 (c)&=0x3f; \
691 for(;;) { \
692 __b=(s)[--(i)]; \
693 if(__b>=0xc0) { \
694 U8_MASK_LEAD_BYTE(__b, __count); \
695 (c)|=(UChar32)__b<<__shift; \
696 break; \
697 } else { \
698 (c)|=(UChar32)(__b&0x3f)<<__shift; \
699 ++__count; \
700 __shift+=6; \
701 } \
702 } \
703 } \
704} UPRV_BLOCK_MACRO_END
705
706/**
707 * Move the string offset from one code point boundary to the previous one
708 * and get the code point between them.
709 * (Pre-decrementing backward iteration.)
710 * "Safe" macro, checks for illegal sequences and for string boundaries.
711 *
712 * The input offset may be the same as the string length.
713 * If the offset is behind a multi-byte sequence, then the macro will read
714 * the whole sequence.
715 * If the offset is behind a lead byte, then that itself
716 * will be returned as the code point.
717 * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value.
718 *
719 * @param s const uint8_t * string
720 * @param start int32_t starting string offset (usually 0)
721 * @param i int32_t string offset, must be start<i
722 * @param c output UChar32 variable, set to <0 in case of an error
723 * @see U8_PREV_UNSAFE
724 * @stable ICU 2.4
725 */
726#define U8_PREV(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
727 (c)=(uint8_t)(s)[--(i)]; \
728 if(!U8_IS_SINGLE(c)) { \
729 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
730 } \
731} UPRV_BLOCK_MACRO_END
732
733/**
734 * Move the string offset from one code point boundary to the previous one
735 * and get the code point between them.
736 * (Pre-decrementing backward iteration.)
737 * "Safe" macro, checks for illegal sequences and for string boundaries.
738 *
739 * The input offset may be the same as the string length.
740 * If the offset is behind a multi-byte sequence, then the macro will read
741 * the whole sequence.
742 * If the offset is behind a lead byte, then that itself
743 * will be returned as the code point.
744 * If the offset is behind an illegal UTF-8 sequence, then c is set to U+FFFD.
745 *
746 * This macro does not distinguish between a real U+FFFD in the text
747 * and U+FFFD returned for an ill-formed sequence.
748 * Use U8_PREV() if that distinction is important.
749 *
750 * @param s const uint8_t * string
751 * @param start int32_t starting string offset (usually 0)
752 * @param i int32_t string offset, must be start<i
753 * @param c output UChar32 variable, set to U+FFFD in case of an error
754 * @see U8_PREV
755 * @stable ICU 51
756 */
757#define U8_PREV_OR_FFFD(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
758 (c)=(uint8_t)(s)[--(i)]; \
759 if(!U8_IS_SINGLE(c)) { \
760 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -3); \
761 } \
762} UPRV_BLOCK_MACRO_END
763
764/**
765 * Move the string offset from one code point boundary to the previous one.
766 * (Pre-decrementing backward iteration.)
767 * The input offset may be the same as the string length.
768 * "Unsafe" macro, assumes well-formed UTF-8.
769 *
770 * @param s const uint8_t * string
771 * @param i string offset
772 * @see U8_BACK_1
773 * @stable ICU 2.4
774 */
775#define U8_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
776 while(U8_IS_TRAIL((s)[--(i)])) {} \
777} UPRV_BLOCK_MACRO_END
778
779/**
780 * Move the string offset from one code point boundary to the previous one.
781 * (Pre-decrementing backward iteration.)
782 * The input offset may be the same as the string length.
783 * "Safe" macro, checks for illegal sequences and for string boundaries.
784 *
785 * @param s const uint8_t * string
786 * @param start int32_t starting string offset (usually 0)
787 * @param i int32_t string offset, must be start<i
788 * @see U8_BACK_1_UNSAFE
789 * @stable ICU 2.4
790 */
791#define U8_BACK_1(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
792 if(U8_IS_TRAIL((s)[--(i)])) { \
793 (i)=utf8_back1SafeBody(s, start, (i)); \
794 } \
795} UPRV_BLOCK_MACRO_END
796
797/**
798 * Move the string offset from one code point boundary to the n-th one before it,
799 * i.e., move backward by n code points.
800 * (Pre-decrementing backward iteration.)
801 * The input offset may be the same as the string length.
802 * "Unsafe" macro, assumes well-formed UTF-8.
803 *
804 * @param s const uint8_t * string
805 * @param i string offset
806 * @param n number of code points to skip
807 * @see U8_BACK_N
808 * @stable ICU 2.4
809 */
810#define U8_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
811 int32_t __N=(n); \
812 while(__N>0) { \
813 U8_BACK_1_UNSAFE(s, i); \
814 --__N; \
815 } \
816} UPRV_BLOCK_MACRO_END
817
818/**
819 * Move the string offset from one code point boundary to the n-th one before it,
820 * i.e., move backward by n code points.
821 * (Pre-decrementing backward iteration.)
822 * The input offset may be the same as the string length.
823 * "Safe" macro, checks for illegal sequences and for string boundaries.
824 *
825 * @param s const uint8_t * string
826 * @param start int32_t index of the start of the string
827 * @param i int32_t string offset, must be start<i
828 * @param n number of code points to skip
829 * @see U8_BACK_N_UNSAFE
830 * @stable ICU 2.4
831 */
832#define U8_BACK_N(s, start, i, n) UPRV_BLOCK_MACRO_BEGIN { \
833 int32_t __N=(n); \
834 while(__N>0 && (i)>(start)) { \
835 U8_BACK_1(s, start, i); \
836 --__N; \
837 } \
838} UPRV_BLOCK_MACRO_END
839
840/**
841 * Adjust a random-access offset to a code point boundary after a code point.
842 * If the offset is behind a partial multi-byte sequence,
843 * then the offset is incremented to behind the whole sequence.
844 * Otherwise, it is not modified.
845 * The input offset may be the same as the string length.
846 * "Unsafe" macro, assumes well-formed UTF-8.
847 *
848 * @param s const uint8_t * string
849 * @param i string offset
850 * @see U8_SET_CP_LIMIT
851 * @stable ICU 2.4
852 */
853#define U8_SET_CP_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
854 U8_BACK_1_UNSAFE(s, i); \
855 U8_FWD_1_UNSAFE(s, i); \
856} UPRV_BLOCK_MACRO_END
857
858/**
859 * Adjust a random-access offset to a code point boundary after a code point.
860 * If the offset is behind a partial multi-byte sequence,
861 * then the offset is incremented to behind the whole sequence.
862 * Otherwise, it is not modified.
863 * The input offset may be the same as the string length.
864 * "Safe" macro, checks for illegal sequences and for string boundaries.
865 *
866 * The length can be negative for a NUL-terminated string.
867 *
868 * @param s const uint8_t * string
869 * @param start int32_t starting string offset (usually 0)
870 * @param i int32_t string offset, must be start<=i<=length
871 * @param length int32_t string length
872 * @see U8_SET_CP_LIMIT_UNSAFE
873 * @stable ICU 2.4
874 */
875#define U8_SET_CP_LIMIT(s, start, i, length) UPRV_BLOCK_MACRO_BEGIN { \
876 if((start)<(i) && ((i)<(length) || (length)<0)) { \
877 U8_BACK_1(s, start, i); \
878 U8_FWD_1(s, i, length); \
879 } \
880} UPRV_BLOCK_MACRO_END
881
882#endif
883

source code of include/unicode/utf8.h