icuSources/common/unicode/utf8.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 1999-2007, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  utf8.h
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 1999sep13
  14 *   created by: Markus W. Scherer
  15 */
  16
  17 /**
  18  * \file
  19  * \brief C API: 8-bit Unicode handling macros
  20  *
  21  * This file defines macros to deal with 8-bit Unicode (UTF-8) code units (bytes) and strings.
  22  * utf8.h is included by utf.h after unicode/umachine.h
  23  * and some common definitions.
  24  *
  25  * For more information see utf.h and the ICU User Guide Strings chapter
  26  * (http://icu-project.org/userguide/strings.html).
  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 /* utf.h must be included first. */
  38 #ifndef __UTF_H__
  39 #   include "unicode/utf.h"
  40 #endif
  41
  42 /* internal definitions ----------------------------------------------------- */
  43
  44 /**
  45  * \var utf8_countTrailBytes
  46  * Internal array with numbers of trail bytes for any given byte used in
  47  * lead byte position.
  48  * @internal
  49  */
  50 #ifdef U_UTF8_IMPL
  51 U_EXPORT const uint8_t
  52 #elif defined(U_STATIC_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION)
  53 U_CFUNC const uint8_t
  54 #else
  55 U_CFUNC U_IMPORT const uint8_t /* U_IMPORT2? */ /*U_IMPORT*/
  56 #endif
  57 utf8_countTrailBytes[256];
  58
  59 /**
  60  * Count the trail bytes for a UTF-8 lead byte.
  61  * @internal
  62  */
  63 #define U8_COUNT_TRAIL_BYTES(leadByte) (utf8_countTrailBytes[(uint8_t)leadByte])
  64
  65 /**
  66  * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value.
  67  * @internal
  68  */
  69 #define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
  70
  71 /**
  72  * Function for handling "next code point" with error-checking.
  73  * @internal
  74  */
  75 U_INTERNAL UChar32 U_EXPORT2
  76 utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
  77
  78 /**
  79  * Function for handling "append code point" with error-checking.
  80  * @internal
  81  */
  82 U_INTERNAL int32_t U_EXPORT2
  83 utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
  84
  85 /**
  86  * Function for handling "previous code point" with error-checking.
  87  * @internal
  88  */
  89 U_INTERNAL UChar32 U_EXPORT2
  90 utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
  91
  92 /**
  93  * Function for handling "skip backward one code point" with error-checking.
  94  * @internal
  95  */
  96 U_INTERNAL int32_t U_EXPORT2
  97 utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
  98
  99 /* single-code point definitions -------------------------------------------- */
 100
 101 /**
 102  * Does this code unit (byte) encode a code point by itself (US-ASCII 0..0x7f)?
 103  * @param c 8-bit code unit (byte)
 104  * @return TRUE or FALSE
 105  * @stable ICU 2.4
 106  */
 107 #define U8_IS_SINGLE(c) (((c)&0x80)==0)
 108
 109 /**
 110  * Is this code unit (byte) a UTF-8 lead byte?
 111  * @param c 8-bit code unit (byte)
 112  * @return TRUE or FALSE
 113  * @stable ICU 2.4
 114  */
 115 #define U8_IS_LEAD(c) ((uint8_t)((c)-0xc0)<0x3e)
 116
 117 /**
 118  * Is this code unit (byte) a UTF-8 trail byte?
 119  * @param c 8-bit code unit (byte)
 120  * @return TRUE or FALSE
 121  * @stable ICU 2.4
 122  */
 123 #define U8_IS_TRAIL(c) (((c)&0xc0)==0x80)
 124
 125 /**
 126  * How many code units (bytes) are used for the UTF-8 encoding
 127  * of this Unicode code point?
 128  * @param c 32-bit code point
 129  * @return 1..4, or 0 if c is a surrogate or not a Unicode code point
 130  * @stable ICU 2.4
 131  */
 132 #define U8_LENGTH(c) \
 133     ((uint32_t)(c)<=0x7f ? 1 : \
 134         ((uint32_t)(c)<=0x7ff ? 2 : \
 135             ((uint32_t)(c)<=0xd7ff ? 3 : \
 136                 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
 137                     ((uint32_t)(c)<=0xffff ? 3 : 4)\
 138                 ) \
 139             ) \
 140         ) \
 141     )
 142
 143 /**
 144  * The maximum number of UTF-8 code units (bytes) per Unicode code point (U+0000..U+10ffff).
 145  * @return 4
 146  * @stable ICU 2.4
 147  */
 148 #define U8_MAX_LENGTH 4
 149
 150 /**
 151  * Get a code point from a string at a random-access offset,
 152  * without changing the offset.
 153  * The offset may point to either the lead byte or one of the trail bytes
 154  * for a code point, in which case the macro will read all of the bytes
 155  * for the code point.
 156  * The result is undefined if the offset points to an illegal UTF-8
 157  * byte sequence.
 158  * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
 159  *
 160  * @param s const uint8_t * string
 161  * @param i string offset
 162  * @param c output UChar32 variable
 163  * @see U8_GET
 164  * @stable ICU 2.4
 165  */
 166 #define U8_GET_UNSAFE(s, i, c) { \
 167     int32_t _u8_get_unsafe_index=(int32_t)(i); \
 168     U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
 169     U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
 170 }
 171
 172 /**
 173  * Get a code point from a string at a random-access offset,
 174  * without changing the offset.
 175  * The offset may point to either the lead byte or one of the trail bytes
 176  * for a code point, in which case the macro will read all of the bytes
 177  * for the code point.
 178  * If the offset points to an illegal UTF-8 byte sequence, then
 179  * c is set to a negative value.
 180  * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
 181  *
 182  * @param s const uint8_t * string
 183  * @param start starting string offset
 184  * @param i string offset, must be start<=i<length
 185  * @param length string length
 186  * @param c output UChar32 variable, set to <0 in case of an error
 187  * @see U8_GET_UNSAFE
 188  * @stable ICU 2.4
 189  */
 190 #define U8_GET(s, start, i, length, c) { \
 191     int32_t _u8_get_index=(int32_t)(i); \
 192     U8_SET_CP_START(s, start, _u8_get_index); \
 193     U8_NEXT(s, _u8_get_index, length, c); \
 194 }
 195
 196 /* definitions with forward iteration --------------------------------------- */
 197
 198 /**
 199  * Get a code point from a string at a code point boundary offset,
 200  * and advance the offset to the next code point boundary.
 201  * (Post-incrementing forward iteration.)
 202  * "Unsafe" macro, assumes well-formed UTF-8.
 203  *
 204  * The offset may point to the lead byte of a multi-byte sequence,
 205  * in which case the macro will read the whole sequence.
 206  * The result is undefined if the offset points to a trail byte
 207  * or an illegal UTF-8 sequence.
 208  *
 209  * @param s const uint8_t * string
 210  * @param i string offset
 211  * @param c output UChar32 variable
 212  * @see U8_NEXT
 213  * @stable ICU 2.4
 214  */
 215 #define U8_NEXT_UNSAFE(s, i, c) { \
 216     (c)=(uint8_t)(s)[(i)++]; \
 217     if((uint8_t)((c)-0xc0)<0x35) { \
 218         uint8_t __count=U8_COUNT_TRAIL_BYTES(c); \
 219         U8_MASK_LEAD_BYTE(c, __count); \
 220         switch(__count) { \
 221         /* each following branch falls through to the next one */ \
 222         case 3: \
 223             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
 224         case 2: \
 225             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
 226         case 1: \
 227             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
 228         /* no other branches to optimize switch() */ \
 229             break; \
 230         } \
 231     } \
 232 }
 233
 234 /**
 235  * Get a code point from a string at a code point boundary offset,
 236  * and advance the offset to the next code point boundary.
 237  * (Post-incrementing forward iteration.)
 238  * "Safe" macro, checks for illegal sequences and for string boundaries.
 239  *
 240  * The offset may point to the lead byte of a multi-byte sequence,
 241  * in which case the macro will read the whole sequence.
 242  * If the offset points to a trail byte or an illegal UTF-8 sequence, then
 243  * c is set to a negative value.
 244  *
 245  * @param s const uint8_t * string
 246  * @param i string offset, must be i<length
 247  * @param length string length
 248  * @param c output UChar32 variable, set to <0 in case of an error
 249  * @see U8_NEXT_UNSAFE
 250  * @stable ICU 2.4
 251  */
 252 #define U8_NEXT(s, i, length, c) { \
 253     (c)=(uint8_t)(s)[(i)++]; \
 254     if((c)>=0x80) { \
 255         uint8_t __t1, __t2; \
 256         if( /* handle U+1000..U+CFFF inline */ \
 257             (0xe0<(c) && (c)<=0xec) && \
 258             (((i)+1)<(length)) && \
 259             (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \
 260             (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \
 261         ) { \
 262             /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
 263             (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \
 264             (i)+=2; \
 265         } else if( /* handle U+0080..U+07FF inline */ \
 266             ((c)<0xe0 && (c)>=0xc2) && \
 267             ((i)<(length)) && \
 268             (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \
 269         ) { \
 270             (c)=(UChar)((((c)&0x1f)<<6)|__t1); \
 271             ++(i); \
 272         } else if(U8_IS_LEAD(c)) { \
 273             /* function call for "complicated" and error cases */ \
 274             (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (int32_t)(length), c, -1); \
 275         } else { \
 276             (c)=U_SENTINEL; \
 277         } \
 278     } \
 279 }
 280
 281 /**
 282  * Append a code point to a string, overwriting 1 to 4 bytes.
 283  * The offset points to the current end of the string contents
 284  * and is advanced (post-increment).
 285  * "Unsafe" macro, assumes a valid code point and sufficient space in the string.
 286  * Otherwise, the result is undefined.
 287  *
 288  * @param s const uint8_t * string buffer
 289  * @param i string offset
 290  * @param c code point to append
 291  * @see U8_APPEND
 292  * @stable ICU 2.4
 293  */
 294 #define U8_APPEND_UNSAFE(s, i, c) { \
 295     if((uint32_t)(c)<=0x7f) { \
 296         (s)[(i)++]=(uint8_t)(c); \
 297     } else { \
 298         if((uint32_t)(c)<=0x7ff) { \
 299             (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
 300         } else { \
 301             if((uint32_t)(c)<=0xffff) { \
 302                 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
 303             } else { \
 304                 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
 305                 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
 306             } \
 307             (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
 308         } \
 309         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
 310     } \
 311 }
 312
 313 /**
 314  * Append a code point to a string, overwriting 1 to 4 bytes.
 315  * The offset points to the current end of the string contents
 316  * and is advanced (post-increment).
 317  * "Safe" macro, checks for a valid code point.
 318  * If a non-ASCII code point is written, checks for sufficient space in the string.
 319  * If the code point is not valid or trail bytes do not fit,
 320  * then isError is set to TRUE.
 321  *
 322  * @param s const uint8_t * string buffer
 323  * @param i string offset, must be i<capacity
 324  * @param capacity size of the string buffer
 325  * @param c code point to append
 326  * @param isError output UBool set to TRUE if an error occurs, otherwise not modified
 327  * @see U8_APPEND_UNSAFE
 328  * @stable ICU 2.4
 329  */
 330 #define U8_APPEND(s, i, capacity, c, isError) { \
 331     if((uint32_t)(c)<=0x7f) { \
 332         (s)[(i)++]=(uint8_t)(c); \
 333     } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \
 334         (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
 335         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
 336     } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \
 337         (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
 338         (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
 339         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
 340     } else { \
 341         (i)=utf8_appendCharSafeBody(s, (int32_t)(i), (int32_t)(capacity), c, &(isError)); \
 342     } \
 343 }
 344
 345 /**
 346  * Advance the string offset from one code point boundary to the next.
 347  * (Post-incrementing iteration.)
 348  * "Unsafe" macro, assumes well-formed UTF-8.
 349  *
 350  * @param s const uint8_t * string
 351  * @param i string offset
 352  * @see U8_FWD_1
 353  * @stable ICU 2.4
 354  */
 355 #define U8_FWD_1_UNSAFE(s, i) { \
 356     (i)+=1+U8_COUNT_TRAIL_BYTES((s)[i]); \
 357 }
 358
 359 /**
 360  * Advance the string offset from one code point boundary to the next.
 361  * (Post-incrementing iteration.)
 362  * "Safe" macro, checks for illegal sequences and for string boundaries.
 363  *
 364  * @param s const uint8_t * string
 365  * @param i string offset, must be i<length
 366  * @param length string length
 367  * @see U8_FWD_1_UNSAFE
 368  * @stable ICU 2.4
 369  */
 370 #define U8_FWD_1(s, i, length) { \
 371     uint8_t __b=(uint8_t)(s)[(i)++]; \
 372     if(U8_IS_LEAD(__b)) { \
 373         uint8_t __count=U8_COUNT_TRAIL_BYTES(__b); \
 374         if((i)+__count>(length)) { \
 375             __count=(uint8_t)((length)-(i)); \
 376         } \
 377         while(__count>0 && U8_IS_TRAIL((s)[i])) { \
 378             ++(i); \
 379             --__count; \
 380         } \
 381     } \
 382 }
 383
 384 /**
 385  * Advance the string offset from one code point boundary to the n-th next one,
 386  * i.e., move forward by n code points.
 387  * (Post-incrementing iteration.)
 388  * "Unsafe" macro, assumes well-formed UTF-8.
 389  *
 390  * @param s const uint8_t * string
 391  * @param i string offset
 392  * @param n number of code points to skip
 393  * @see U8_FWD_N
 394  * @stable ICU 2.4
 395  */
 396 #define U8_FWD_N_UNSAFE(s, i, n) { \
 397     int32_t __N=(n); \
 398     while(__N>0) { \
 399         U8_FWD_1_UNSAFE(s, i); \
 400         --__N; \
 401     } \
 402 }
 403
 404 /**
 405  * Advance the string offset from one code point boundary to the n-th next one,
 406  * i.e., move forward by n code points.
 407  * (Post-incrementing iteration.)
 408  * "Safe" macro, checks for illegal sequences and for string boundaries.
 409  *
 410  * @param s const uint8_t * string
 411  * @param i string offset, must be i<length
 412  * @param length string length
 413  * @param n number of code points to skip
 414  * @see U8_FWD_N_UNSAFE
 415  * @stable ICU 2.4
 416  */
 417 #define U8_FWD_N(s, i, length, n) { \
 418     int32_t __N=(n); \
 419     while(__N>0 && (i)<(length)) { \
 420         U8_FWD_1(s, i, length); \
 421         --__N; \
 422     } \
 423 }
 424
 425 /**
 426  * Adjust a random-access offset to a code point boundary
 427  * at the start of a code point.
 428  * If the offset points to a UTF-8 trail byte,
 429  * then the offset is moved backward to the corresponding lead byte.
 430  * Otherwise, it is not modified.
 431  * "Unsafe" macro, assumes well-formed UTF-8.
 432  *
 433  * @param s const uint8_t * string
 434  * @param i string offset
 435  * @see U8_SET_CP_START
 436  * @stable ICU 2.4
 437  */
 438 #define U8_SET_CP_START_UNSAFE(s, i) { \
 439     while(U8_IS_TRAIL((s)[i])) { --(i); } \
 440 }
 441
 442 /**
 443  * Adjust a random-access offset to a code point boundary
 444  * at the start of a code point.
 445  * If the offset points to a UTF-8 trail byte,
 446  * then the offset is moved backward to the corresponding lead byte.
 447  * Otherwise, it is not modified.
 448  * "Safe" macro, checks for illegal sequences and for string boundaries.
 449  *
 450  * @param s const uint8_t * string
 451  * @param start starting string offset (usually 0)
 452  * @param i string offset, must be start<=i
 453  * @see U8_SET_CP_START_UNSAFE
 454  * @stable ICU 2.4
 455  */
 456 #define U8_SET_CP_START(s, start, i) { \
 457     if(U8_IS_TRAIL((s)[(i)])) { \
 458         (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \
 459     } \
 460 }
 461
 462 /* definitions with backward iteration -------------------------------------- */
 463
 464 /**
 465  * Move the string offset from one code point boundary to the previous one
 466  * and get the code point between them.
 467  * (Pre-decrementing backward iteration.)
 468  * "Unsafe" macro, assumes well-formed UTF-8.
 469  *
 470  * The input offset may be the same as the string length.
 471  * If the offset is behind a multi-byte sequence, then the macro will read
 472  * the whole sequence.
 473  * If the offset is behind a lead byte, then that itself
 474  * will be returned as the code point.
 475  * The result is undefined if the offset is behind an illegal UTF-8 sequence.
 476  *
 477  * @param s const uint8_t * string
 478  * @param i string offset
 479  * @param c output UChar32 variable
 480  * @see U8_PREV
 481  * @stable ICU 2.4
 482  */
 483 #define U8_PREV_UNSAFE(s, i, c) { \
 484     (c)=(uint8_t)(s)[--(i)]; \
 485     if(U8_IS_TRAIL(c)) { \
 486         uint8_t __b, __count=1, __shift=6; \
 487 \
 488         /* c is a trail byte */ \
 489         (c)&=0x3f; \
 490         for(;;) { \
 491             __b=(uint8_t)(s)[--(i)]; \
 492             if(__b>=0xc0) { \
 493                 U8_MASK_LEAD_BYTE(__b, __count); \
 494                 (c)|=(UChar32)__b<<__shift; \
 495                 break; \
 496             } else { \
 497                 (c)|=(UChar32)(__b&0x3f)<<__shift; \
 498                 ++__count; \
 499                 __shift+=6; \
 500             } \
 501         } \
 502     } \
 503 }
 504
 505 /**
 506  * Move the string offset from one code point boundary to the previous one
 507  * and get the code point between them.
 508  * (Pre-decrementing backward iteration.)
 509  * "Safe" macro, checks for illegal sequences and for string boundaries.
 510  *
 511  * The input offset may be the same as the string length.
 512  * If the offset is behind a multi-byte sequence, then the macro will read
 513  * the whole sequence.
 514  * If the offset is behind a lead byte, then that itself
 515  * will be returned as the code point.
 516  * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value.
 517  *
 518  * @param s const uint8_t * string
 519  * @param start starting string offset (usually 0)
 520  * @param i string offset, must be start<i
 521  * @param c output UChar32 variable, set to <0 in case of an error
 522  * @see U8_PREV_UNSAFE
 523  * @stable ICU 2.4
 524  */
 525 #define U8_PREV(s, start, i, c) { \
 526     (c)=(uint8_t)(s)[--(i)]; \
 527     if((c)>=0x80) { \
 528         if((c)<=0xbf) { \
 529             (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
 530         } else { \
 531             (c)=U_SENTINEL; \
 532         } \
 533     } \
 534 }
 535
 536 /**
 537  * Move the string offset from one code point boundary to the previous one.
 538  * (Pre-decrementing backward iteration.)
 539  * The input offset may be the same as the string length.
 540  * "Unsafe" macro, assumes well-formed UTF-8.
 541  *
 542  * @param s const uint8_t * string
 543  * @param i string offset
 544  * @see U8_BACK_1
 545  * @stable ICU 2.4
 546  */
 547 #define U8_BACK_1_UNSAFE(s, i) { \
 548     while(U8_IS_TRAIL((s)[--(i)])) {} \
 549 }
 550
 551 /**
 552  * Move the string offset from one code point boundary to the previous one.
 553  * (Pre-decrementing backward iteration.)
 554  * The input offset may be the same as the string length.
 555  * "Safe" macro, checks for illegal sequences and for string boundaries.
 556  *
 557  * @param s const uint8_t * string
 558  * @param start starting string offset (usually 0)
 559  * @param i string offset, must be start<i
 560  * @see U8_BACK_1_UNSAFE
 561  * @stable ICU 2.4
 562  */
 563 #define U8_BACK_1(s, start, i) { \
 564     if(U8_IS_TRAIL((s)[--(i)])) { \
 565         (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \
 566     } \
 567 }
 568
 569 /**
 570  * Move the string offset from one code point boundary to the n-th one before it,
 571  * i.e., move backward by n code points.
 572  * (Pre-decrementing backward iteration.)
 573  * The input offset may be the same as the string length.
 574  * "Unsafe" macro, assumes well-formed UTF-8.
 575  *
 576  * @param s const uint8_t * string
 577  * @param i string offset
 578  * @param n number of code points to skip
 579  * @see U8_BACK_N
 580  * @stable ICU 2.4
 581  */
 582 #define U8_BACK_N_UNSAFE(s, i, n) { \
 583     int32_t __N=(n); \
 584     while(__N>0) { \
 585         U8_BACK_1_UNSAFE(s, i); \
 586         --__N; \
 587     } \
 588 }
 589
 590 /**
 591  * Move the string offset from one code point boundary to the n-th one before it,
 592  * i.e., move backward by n code points.
 593  * (Pre-decrementing backward iteration.)
 594  * The input offset may be the same as the string length.
 595  * "Safe" macro, checks for illegal sequences and for string boundaries.
 596  *
 597  * @param s const uint8_t * string
 598  * @param start index of the start of the string
 599  * @param i string offset, must be start<i
 600  * @param n number of code points to skip
 601  * @see U8_BACK_N_UNSAFE
 602  * @stable ICU 2.4
 603  */
 604 #define U8_BACK_N(s, start, i, n) { \
 605     int32_t __N=(n); \
 606     while(__N>0 && (i)>(start)) { \
 607         U8_BACK_1(s, start, i); \
 608         --__N; \
 609     } \
 610 }
 611
 612 /**
 613  * Adjust a random-access offset to a code point boundary after a code point.
 614  * If the offset is behind a partial multi-byte sequence,
 615  * then the offset is incremented to behind the whole sequence.
 616  * Otherwise, it is not modified.
 617  * The input offset may be the same as the string length.
 618  * "Unsafe" macro, assumes well-formed UTF-8.
 619  *
 620  * @param s const uint8_t * string
 621  * @param i string offset
 622  * @see U8_SET_CP_LIMIT
 623  * @stable ICU 2.4
 624  */
 625 #define U8_SET_CP_LIMIT_UNSAFE(s, i) { \
 626     U8_BACK_1_UNSAFE(s, i); \
 627     U8_FWD_1_UNSAFE(s, i); \
 628 }
 629
 630 /**
 631  * Adjust a random-access offset to a code point boundary after a code point.
 632  * If the offset is behind a partial multi-byte sequence,
 633  * then the offset is incremented to behind the whole sequence.
 634  * Otherwise, it is not modified.
 635  * The input offset may be the same as the string length.
 636  * "Safe" macro, checks for illegal sequences and for string boundaries.
 637  *
 638  * @param s const uint8_t * string
 639  * @param start starting string offset (usually 0)
 640  * @param i string offset, must be start<=i<=length
 641  * @param length string length
 642  * @see U8_SET_CP_LIMIT_UNSAFE
 643  * @stable ICU 2.4
 644  */
 645 #define U8_SET_CP_LIMIT(s, start, i, length) { \
 646     if((start)<(i) && (i)<(length)) { \
 647         U8_BACK_1(s, start, i); \
 648         U8_FWD_1(s, i, length); \
 649     } \
 650 }
 651
 652 #endif