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) UPRV_BLOCK_MACRO_BEGIN { \
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} UPRV_BLOCK_MACRO_END
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) UPRV_BLOCK_MACRO_BEGIN { \
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} UPRV_BLOCK_MACRO_END
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) UPRV_BLOCK_MACRO_BEGIN { \
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} UPRV_BLOCK_MACRO_END
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) UPRV_BLOCK_MACRO_BEGIN { \
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} UPRV_BLOCK_MACRO_END
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) U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, U_SENTINEL)
352
353/**
354 * Get a code point from a string at a code point boundary offset,
355 * and advance the offset to the next code point boundary.
356 * (Post-incrementing forward iteration.)
357 * "Safe" macro, checks for illegal sequences and for string boundaries.
358 *
359 * The length can be negative for a NUL-terminated string.
360 *
361 * The offset may point to the lead byte of a multi-byte sequence,
362 * in which case the macro will read the whole sequence.
363 * If the offset points to a trail byte or an illegal UTF-8 sequence, then
364 * c is set to U+FFFD.
365 *
366 * This macro does not distinguish between a real U+FFFD in the text
367 * and U+FFFD returned for an ill-formed sequence.
368 * Use U8_NEXT() if that distinction is important.
369 *
370 * @param s const uint8_t * string
371 * @param i int32_t string offset, must be i<length
372 * @param length int32_t string length
373 * @param c output UChar32 variable, set to U+FFFD in case of an error
374 * @see U8_NEXT
375 * @stable ICU 51
376 */
377#define U8_NEXT_OR_FFFD(s, i, length, c) U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, 0xfffd)
378
379/** @internal */
380#define U8_INTERNAL_NEXT_OR_SUB(s, i, length, c, sub) UPRV_BLOCK_MACRO_BEGIN { \
381 (c)=(uint8_t)(s)[(i)++]; \
382 if(!U8_IS_SINGLE(c)) { \
383 uint8_t __t = 0; \
384 if((i)!=(length) && \
385 /* fetch/validate/assemble all but last trail byte */ \
386 ((c)>=0xe0 ? \
387 ((c)<0xf0 ? /* U+0800..U+FFFF except surrogates */ \
388 U8_LEAD3_T1_BITS[(c)&=0xf]&(1<<((__t=(s)[i])>>5)) && \
389 (__t&=0x3f, 1) \
390 : /* U+10000..U+10FFFF */ \
391 ((c)-=0xf0)<=4 && \
392 U8_LEAD4_T1_BITS[(__t=(s)[i])>>4]&(1<<(c)) && \
393 ((c)=((c)<<6)|(__t&0x3f), ++(i)!=(length)) && \
394 (__t=(s)[i]-0x80)<=0x3f) && \
395 /* valid second-to-last trail byte */ \
396 ((c)=((c)<<6)|__t, ++(i)!=(length)) \
397 : /* U+0080..U+07FF */ \
398 (c)>=0xc2 && ((c)&=0x1f, 1)) && \
399 /* last trail byte */ \
400 (__t=(s)[i]-0x80)<=0x3f && \
401 ((c)=((c)<<6)|__t, ++(i), 1)) { \
402 } else { \
403 (c)=(sub); /* ill-formed*/ \
404 } \
405 } \
406} UPRV_BLOCK_MACRO_END
407
408/**
409 * Append a code point to a string, overwriting 1 to 4 bytes.
410 * The offset points to the current end of the string contents
411 * and is advanced (post-increment).
412 * "Unsafe" macro, assumes a valid code point and sufficient space in the string.
413 * Otherwise, the result is undefined.
414 *
415 * @param s const uint8_t * string buffer
416 * @param i string offset
417 * @param c code point to append
418 * @see U8_APPEND
419 * @stable ICU 2.4
420 */
421#define U8_APPEND_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
422 uint32_t __uc=(c); \
423 if(__uc<=0x7f) { \
424 (s)[(i)++]=(uint8_t)__uc; \
425 } else { \
426 if(__uc<=0x7ff) { \
427 (s)[(i)++]=(uint8_t)((__uc>>6)|0xc0); \
428 } else { \
429 if(__uc<=0xffff) { \
430 (s)[(i)++]=(uint8_t)((__uc>>12)|0xe0); \
431 } else { \
432 (s)[(i)++]=(uint8_t)((__uc>>18)|0xf0); \
433 (s)[(i)++]=(uint8_t)(((__uc>>12)&0x3f)|0x80); \
434 } \
435 (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
436 } \
437 (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
438 } \
439} UPRV_BLOCK_MACRO_END
440
441/**
442 * Append a code point to a string, overwriting 1 to 4 bytes.
443 * The offset points to the current end of the string contents
444 * and is advanced (post-increment).
445 * "Safe" macro, checks for a valid code point.
446 * If a non-ASCII code point is written, checks for sufficient space in the string.
447 * If the code point is not valid or trail bytes do not fit,
448 * then isError is set to TRUE.
449 *
450 * @param s const uint8_t * string buffer
451 * @param i int32_t string offset, must be i<capacity
452 * @param capacity int32_t size of the string buffer
453 * @param c UChar32 code point to append
454 * @param isError output UBool set to TRUE if an error occurs, otherwise not modified
455 * @see U8_APPEND_UNSAFE
456 * @stable ICU 2.4
457 */
458#define U8_APPEND(s, i, capacity, c, isError) UPRV_BLOCK_MACRO_BEGIN { \
459 uint32_t __uc=(c); \
460 if(__uc<=0x7f) { \
461 (s)[(i)++]=(uint8_t)__uc; \
462 } else if(__uc<=0x7ff && (i)+1<(capacity)) { \
463 (s)[(i)++]=(uint8_t)((__uc>>6)|0xc0); \
464 (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
465 } else if((__uc<=0xd7ff || (0xe000<=__uc && __uc<=0xffff)) && (i)+2<(capacity)) { \
466 (s)[(i)++]=(uint8_t)((__uc>>12)|0xe0); \
467 (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
468 (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
469 } else if(0xffff<__uc && __uc<=0x10ffff && (i)+3<(capacity)) { \
470 (s)[(i)++]=(uint8_t)((__uc>>18)|0xf0); \
471 (s)[(i)++]=(uint8_t)(((__uc>>12)&0x3f)|0x80); \
472 (s)[(i)++]=(uint8_t)(((__uc>>6)&0x3f)|0x80); \
473 (s)[(i)++]=(uint8_t)((__uc&0x3f)|0x80); \
474 } else { \
475 (isError)=TRUE; \
476 } \
477} UPRV_BLOCK_MACRO_END
478
479/**
480 * Advance the string offset from one code point boundary to the next.
481 * (Post-incrementing iteration.)
482 * "Unsafe" macro, assumes well-formed UTF-8.
483 *
484 * @param s const uint8_t * string
485 * @param i string offset
486 * @see U8_FWD_1
487 * @stable ICU 2.4
488 */
489#define U8_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
490 (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((s)[i]); \
491} UPRV_BLOCK_MACRO_END
492
493/**
494 * Advance the string offset from one code point boundary to the next.
495 * (Post-incrementing iteration.)
496 * "Safe" macro, checks for illegal sequences and for string boundaries.
497 *
498 * The length can be negative for a NUL-terminated string.
499 *
500 * @param s const uint8_t * string
501 * @param i int32_t string offset, must be i<length
502 * @param length int32_t string length
503 * @see U8_FWD_1_UNSAFE
504 * @stable ICU 2.4
505 */
506#define U8_FWD_1(s, i, length) UPRV_BLOCK_MACRO_BEGIN { \
507 uint8_t __b=(s)[(i)++]; \
508 if(U8_IS_LEAD(__b) && (i)!=(length)) { \
509 uint8_t __t1=(s)[i]; \
510 if((0xe0<=__b && __b<0xf0)) { \
511 if(U8_IS_VALID_LEAD3_AND_T1(__b, __t1) && \
512 ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
513 ++(i); \
514 } \
515 } else if(__b<0xe0) { \
516 if(U8_IS_TRAIL(__t1)) { \
517 ++(i); \
518 } \
519 } else /* c>=0xf0 */ { \
520 if(U8_IS_VALID_LEAD4_AND_T1(__b, __t1) && \
521 ++(i)!=(length) && U8_IS_TRAIL((s)[i]) && \
522 ++(i)!=(length) && U8_IS_TRAIL((s)[i])) { \
523 ++(i); \
524 } \
525 } \
526 } \
527} UPRV_BLOCK_MACRO_END
528
529/**
530 * Advance the string offset from one code point boundary to the n-th next one,
531 * i.e., move forward by n code points.
532 * (Post-incrementing iteration.)
533 * "Unsafe" macro, assumes well-formed UTF-8.
534 *
535 * @param s const uint8_t * string
536 * @param i string offset
537 * @param n number of code points to skip
538 * @see U8_FWD_N
539 * @stable ICU 2.4
540 */
541#define U8_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
542 int32_t __N=(n); \
543 while(__N>0) { \
544 U8_FWD_1_UNSAFE(s, i); \
545 --__N; \
546 } \
547} UPRV_BLOCK_MACRO_END
548
549/**
550 * Advance the string offset from one code point boundary to the n-th next one,
551 * i.e., move forward by n code points.
552 * (Post-incrementing iteration.)
553 * "Safe" macro, checks for illegal sequences and for string boundaries.
554 *
555 * The length can be negative for a NUL-terminated string.
556 *
557 * @param s const uint8_t * string
558 * @param i int32_t string offset, must be i<length
559 * @param length int32_t string length
560 * @param n number of code points to skip
561 * @see U8_FWD_N_UNSAFE
562 * @stable ICU 2.4
563 */
564#define U8_FWD_N(s, i, length, n) UPRV_BLOCK_MACRO_BEGIN { \
565 int32_t __N=(n); \
566 while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
567 U8_FWD_1(s, i, length); \
568 --__N; \
569 } \
570} UPRV_BLOCK_MACRO_END
571
572/**
573 * Adjust a random-access offset to a code point boundary
574 * at the start of a code point.
575 * If the offset points to a UTF-8 trail byte,
576 * then the offset is moved backward to the corresponding lead byte.
577 * Otherwise, it is not modified.
578 * "Unsafe" macro, assumes well-formed UTF-8.
579 *
580 * @param s const uint8_t * string
581 * @param i string offset
582 * @see U8_SET_CP_START
583 * @stable ICU 2.4
584 */
585#define U8_SET_CP_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
586 while(U8_IS_TRAIL((s)[i])) { --(i); } \
587} UPRV_BLOCK_MACRO_END
588
589/**
590 * Adjust a random-access offset to a code point boundary
591 * at the start of a code point.
592 * If the offset points to a UTF-8 trail byte,
593 * then the offset is moved backward to the corresponding lead byte.
594 * Otherwise, it is not modified.
595 *
596 * "Safe" macro, checks for illegal sequences and for string boundaries.
597 * Unlike U8_TRUNCATE_IF_INCOMPLETE(), this macro always reads s[i].
598 *
599 * @param s const uint8_t * string
600 * @param start int32_t starting string offset (usually 0)
601 * @param i int32_t string offset, must be start<=i
602 * @see U8_SET_CP_START_UNSAFE
603 * @see U8_TRUNCATE_IF_INCOMPLETE
604 * @stable ICU 2.4
605 */
606#define U8_SET_CP_START(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
607 if(U8_IS_TRAIL((s)[(i)])) { \
608 (i)=utf8_back1SafeBody(s, start, (i)); \
609 } \
610} UPRV_BLOCK_MACRO_END
611
612/**
613 * If the string ends with a UTF-8 byte sequence that is valid so far
614 * but incomplete, then reduce the length of the string to end before
615 * the lead byte of that incomplete sequence.
616 * For example, if the string ends with E1 80, the length is reduced by 2.
617 *
618 * In all other cases (the string ends with a complete sequence, or it is not
619 * possible for any further trail byte to extend the trailing sequence)
620 * the length remains unchanged.
621 *
622 * Useful for processing text split across multiple buffers
623 * (save the incomplete sequence for later)
624 * and for optimizing iteration
625 * (check for string length only once per character).
626 *
627 * "Safe" macro, checks for illegal sequences and for string boundaries.
628 * Unlike U8_SET_CP_START(), this macro never reads s[length].
629 *
630 * (In UTF-16, simply check for U16_IS_LEAD(last code unit).)
631 *
632 * @param s const uint8_t * string
633 * @param start int32_t starting string offset (usually 0)
634 * @param length int32_t string length (usually start<=length)
635 * @see U8_SET_CP_START
636 * @stable ICU 61
637 */
638#define U8_TRUNCATE_IF_INCOMPLETE(s, start, length) UPRV_BLOCK_MACRO_BEGIN { \
639 if((length)>(start)) { \
640 uint8_t __b1=s[(length)-1]; \
641 if(U8_IS_SINGLE(__b1)) { \
642 /* common ASCII character */ \
643 } else if(U8_IS_LEAD(__b1)) { \
644 --(length); \
645 } else if(U8_IS_TRAIL(__b1) && ((length)-2)>=(start)) { \
646 uint8_t __b2=s[(length)-2]; \
647 if(0xe0<=__b2 && __b2<=0xf4) { \
648 if(__b2<0xf0 ? U8_IS_VALID_LEAD3_AND_T1(__b2, __b1) : \
649 U8_IS_VALID_LEAD4_AND_T1(__b2, __b1)) { \
650 (length)-=2; \
651 } \
652 } else if(U8_IS_TRAIL(__b2) && ((length)-3)>=(start)) { \
653 uint8_t __b3=s[(length)-3]; \
654 if(0xf0<=__b3 && __b3<=0xf4 && U8_IS_VALID_LEAD4_AND_T1(__b3, __b2)) { \
655 (length)-=3; \
656 } \
657 } \
658 } \
659 } \
660} UPRV_BLOCK_MACRO_END
661
662/* definitions with backward iteration -------------------------------------- */
663
664/**
665 * Move the string offset from one code point boundary to the previous one
666 * and get the code point between them.
667 * (Pre-decrementing backward iteration.)
668 * "Unsafe" macro, assumes well-formed UTF-8.
669 *
670 * The input offset may be the same as the string length.
671 * If the offset is behind a multi-byte sequence, then the macro will read
672 * the whole sequence.
673 * If the offset is behind a lead byte, then that itself
674 * will be returned as the code point.
675 * The result is undefined if the offset is behind an illegal UTF-8 sequence.
676 *
677 * @param s const uint8_t * string
678 * @param i string offset
679 * @param c output UChar32 variable
680 * @see U8_PREV
681 * @stable ICU 2.4
682 */
683#define U8_PREV_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \
684 (c)=(uint8_t)(s)[--(i)]; \
685 if(U8_IS_TRAIL(c)) { \
686 uint8_t __b, __count=1, __shift=6; \
687\
688 /* c is a trail byte */ \
689 (c)&=0x3f; \
690 for(;;) { \
691 __b=(s)[--(i)]; \
692 if(__b>=0xc0) { \
693 U8_MASK_LEAD_BYTE(__b, __count); \
694 (c)|=(UChar32)__b<<__shift; \
695 break; \
696 } else { \
697 (c)|=(UChar32)(__b&0x3f)<<__shift; \
698 ++__count; \
699 __shift+=6; \
700 } \
701 } \
702 } \
703} UPRV_BLOCK_MACRO_END
704
705/**
706 * Move the string offset from one code point boundary to the previous one
707 * and get the code point between them.
708 * (Pre-decrementing backward iteration.)
709 * "Safe" macro, checks for illegal sequences and for string boundaries.
710 *
711 * The input offset may be the same as the string length.
712 * If the offset is behind a multi-byte sequence, then the macro will read
713 * the whole sequence.
714 * If the offset is behind a lead byte, then that itself
715 * will be returned as the code point.
716 * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value.
717 *
718 * @param s const uint8_t * string
719 * @param start int32_t starting string offset (usually 0)
720 * @param i int32_t string offset, must be start<i
721 * @param c output UChar32 variable, set to <0 in case of an error
722 * @see U8_PREV_UNSAFE
723 * @stable ICU 2.4
724 */
725#define U8_PREV(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
726 (c)=(uint8_t)(s)[--(i)]; \
727 if(!U8_IS_SINGLE(c)) { \
728 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
729 } \
730} UPRV_BLOCK_MACRO_END
731
732/**
733 * Move the string offset from one code point boundary to the previous one
734 * and get the code point between them.
735 * (Pre-decrementing backward iteration.)
736 * "Safe" macro, checks for illegal sequences and for string boundaries.
737 *
738 * The input offset may be the same as the string length.
739 * If the offset is behind a multi-byte sequence, then the macro will read
740 * the whole sequence.
741 * If the offset is behind a lead byte, then that itself
742 * will be returned as the code point.
743 * If the offset is behind an illegal UTF-8 sequence, then c is set to U+FFFD.
744 *
745 * This macro does not distinguish between a real U+FFFD in the text
746 * and U+FFFD returned for an ill-formed sequence.
747 * Use U8_PREV() if that distinction is important.
748 *
749 * @param s const uint8_t * string
750 * @param start int32_t starting string offset (usually 0)
751 * @param i int32_t string offset, must be start<i
752 * @param c output UChar32 variable, set to U+FFFD in case of an error
753 * @see U8_PREV
754 * @stable ICU 51
755 */
756#define U8_PREV_OR_FFFD(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \
757 (c)=(uint8_t)(s)[--(i)]; \
758 if(!U8_IS_SINGLE(c)) { \
759 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -3); \
760 } \
761} UPRV_BLOCK_MACRO_END
762
763/**
764 * Move the string offset from one code point boundary to the previous one.
765 * (Pre-decrementing backward iteration.)
766 * The input offset may be the same as the string length.
767 * "Unsafe" macro, assumes well-formed UTF-8.
768 *
769 * @param s const uint8_t * string
770 * @param i string offset
771 * @see U8_BACK_1
772 * @stable ICU 2.4
773 */
774#define U8_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
775 while(U8_IS_TRAIL((s)[--(i)])) {} \
776} UPRV_BLOCK_MACRO_END
777
778/**
779 * Move the string offset from one code point boundary to the previous one.
780 * (Pre-decrementing backward iteration.)
781 * The input offset may be the same as the string length.
782 * "Safe" macro, checks for illegal sequences and for string boundaries.
783 *
784 * @param s const uint8_t * string
785 * @param start int32_t starting string offset (usually 0)
786 * @param i int32_t string offset, must be start<i
787 * @see U8_BACK_1_UNSAFE
788 * @stable ICU 2.4
789 */
790#define U8_BACK_1(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \
791 if(U8_IS_TRAIL((s)[--(i)])) { \
792 (i)=utf8_back1SafeBody(s, start, (i)); \
793 } \
794} UPRV_BLOCK_MACRO_END
795
796/**
797 * Move the string offset from one code point boundary to the n-th one before it,
798 * i.e., move backward by n code points.
799 * (Pre-decrementing backward iteration.)
800 * The input offset may be the same as the string length.
801 * "Unsafe" macro, assumes well-formed UTF-8.
802 *
803 * @param s const uint8_t * string
804 * @param i string offset
805 * @param n number of code points to skip
806 * @see U8_BACK_N
807 * @stable ICU 2.4
808 */
809#define U8_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \
810 int32_t __N=(n); \
811 while(__N>0) { \
812 U8_BACK_1_UNSAFE(s, i); \
813 --__N; \
814 } \
815} UPRV_BLOCK_MACRO_END
816
817/**
818 * Move the string offset from one code point boundary to the n-th one before it,
819 * i.e., move backward by n code points.
820 * (Pre-decrementing backward iteration.)
821 * The input offset may be the same as the string length.
822 * "Safe" macro, checks for illegal sequences and for string boundaries.
823 *
824 * @param s const uint8_t * string
825 * @param start int32_t index of the start of the string
826 * @param i int32_t string offset, must be start<i
827 * @param n number of code points to skip
828 * @see U8_BACK_N_UNSAFE
829 * @stable ICU 2.4
830 */
831#define U8_BACK_N(s, start, i, n) UPRV_BLOCK_MACRO_BEGIN { \
832 int32_t __N=(n); \
833 while(__N>0 && (i)>(start)) { \
834 U8_BACK_1(s, start, i); \
835 --__N; \
836 } \
837} UPRV_BLOCK_MACRO_END
838
839/**
840 * Adjust a random-access offset to a code point boundary after a code point.
841 * If the offset is behind a partial multi-byte sequence,
842 * then the offset is incremented to behind the whole sequence.
843 * Otherwise, it is not modified.
844 * The input offset may be the same as the string length.
845 * "Unsafe" macro, assumes well-formed UTF-8.
846 *
847 * @param s const uint8_t * string
848 * @param i string offset
849 * @see U8_SET_CP_LIMIT
850 * @stable ICU 2.4
851 */
852#define U8_SET_CP_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \
853 U8_BACK_1_UNSAFE(s, i); \
854 U8_FWD_1_UNSAFE(s, i); \
855} UPRV_BLOCK_MACRO_END
856
857/**
858 * Adjust a random-access offset to a code point boundary after a code point.
859 * If the offset is behind a partial multi-byte sequence,
860 * then the offset is incremented to behind the whole sequence.
861 * Otherwise, it is not modified.
862 * The input offset may be the same as the string length.
863 * "Safe" macro, checks for illegal sequences and for string boundaries.
864 *
865 * The length can be negative for a NUL-terminated string.
866 *
867 * @param s const uint8_t * string
868 * @param start int32_t starting string offset (usually 0)
869 * @param i int32_t string offset, must be start<=i<=length
870 * @param length int32_t string length
871 * @see U8_SET_CP_LIMIT_UNSAFE
872 * @stable ICU 2.4
873 */
874#define U8_SET_CP_LIMIT(s, start, i, length) UPRV_BLOCK_MACRO_BEGIN { \
875 if((start)<(i) && ((i)<(length) || (length)<0)) { \
876 U8_BACK_1(s, start, i); \
877 U8_FWD_1(s, i, length); \
878 } \
879} UPRV_BLOCK_MACRO_END
880
881#endif
882