icuSources/common/unicode/utf8.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 1999-2011, 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  *
  23  * For more information see utf.h and the ICU User Guide Strings chapter
  24  * (http://userguide.icu-project.org/strings).
  25  *
  26  * <em>Usage:</em>
  27  * ICU coding guidelines for if() statements should be followed when using these macros.
  28  * Compound statements (curly braces {}) must be used  for if-else-while...
  29  * bodies and all macro statements should be terminated with semicolon.
  30  */
  31
  32 #ifndef __UTF8_H__
  33 #define __UTF8_H__
  34
  35 #include "unicode/umachine.h"
  36 #ifndef __UTF_H__
  37 #   include "unicode/utf.h"
  38 #endif
  39
  40 /* internal definitions ----------------------------------------------------- */
  41
  42 /**
  43  * \var utf8_countTrailBytes
  44  * Internal array with numbers of trail bytes for any given byte used in
  45  * lead byte position.
  46  *
  47  * This is internal since it is not meant to be called directly by external clients;
  48  * however it is called by public macros in this file and thus must remain stable,
  49  * and should not be hidden when other internal functions are hidden (otherwise
  50  * public macros would fail to compile).
  51  * @internal
  52  */
  53 #ifdef U_UTF8_IMPL
  54 U_EXPORT const uint8_t
  55 #elif defined(U_STATIC_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION)
  56 U_CFUNC const uint8_t
  57 #else
  58 U_CFUNC U_IMPORT const uint8_t /* U_IMPORT2? */ /*U_IMPORT*/
  59 #endif
  60 utf8_countTrailBytes[256];
  61
  62 /**
  63  * Count the trail bytes for a UTF-8 lead byte.
  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  * @internal
  68  */
  69 #define U8_COUNT_TRAIL_BYTES(leadByte) (utf8_countTrailBytes[(uint8_t)leadByte])
  70
  71 /**
  72  * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value.
  73  *
  74  * This is internal since it is not meant to be called directly by external clients;
  75  * however it is called by public macros in this file and thus must remain stable.
  76  * @internal
  77  */
  78 #define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
  79
  80 /**
  81  * Function for handling "next code point" with error-checking.
  82  *
  83  * This is internal since it is not meant to be called directly by external clients;
  84  * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
  85  * file and thus must remain stable, and should not be hidden when other internal
  86  * functions are hidden (otherwise public macros would fail to compile).
  87  * @internal
  88  */
  89 U_STABLE UChar32 U_EXPORT2
  90 utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
  91
  92 /**
  93  * Function for handling "append code point" with error-checking.
  94  *
  95  * This is internal since it is not meant to be called directly by external clients;
  96  * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
  97  * file and thus must remain stable, and should not be hidden when other internal
  98  * functions are hidden (otherwise public macros would fail to compile).
  99  * @internal
 100  */
 101 U_STABLE int32_t U_EXPORT2
 102 utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
 103
 104 /**
 105  * Function for handling "previous code point" with error-checking.
 106  *
 107  * This is internal since it is not meant to be called directly by external clients;
 108  * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
 109  * file and thus must remain stable, and should not be hidden when other internal
 110  * functions are hidden (otherwise public macros would fail to compile).
 111  * @internal
 112  */
 113 U_STABLE UChar32 U_EXPORT2
 114 utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
 115
 116 /**
 117  * Function for handling "skip backward one 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  */
 125 U_STABLE int32_t U_EXPORT2
 126 utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
 127
 128 /* single-code point definitions -------------------------------------------- */
 129
 130 /**
 131  * Does this code unit (byte) encode a code point by itself (US-ASCII 0..0x7f)?
 132  * @param c 8-bit code unit (byte)
 133  * @return TRUE or FALSE
 134  * @stable ICU 2.4
 135  */
 136 #define U8_IS_SINGLE(c) (((c)&0x80)==0)
 137
 138 /**
 139  * Is this code unit (byte) a UTF-8 lead byte?
 140  * @param c 8-bit code unit (byte)
 141  * @return TRUE or FALSE
 142  * @stable ICU 2.4
 143  */
 144 #define U8_IS_LEAD(c) ((uint8_t)((c)-0xc0)<0x3e)
 145
 146 /**
 147  * Is this code unit (byte) a UTF-8 trail byte?
 148  * @param c 8-bit code unit (byte)
 149  * @return TRUE or FALSE
 150  * @stable ICU 2.4
 151  */
 152 #define U8_IS_TRAIL(c) (((c)&0xc0)==0x80)
 153
 154 /**
 155  * How many code units (bytes) are used for the UTF-8 encoding
 156  * of this Unicode code point?
 157  * @param c 32-bit code point
 158  * @return 1..4, or 0 if c is a surrogate or not a Unicode code point
 159  * @stable ICU 2.4
 160  */
 161 #define U8_LENGTH(c) \
 162     ((uint32_t)(c)<=0x7f ? 1 : \
 163         ((uint32_t)(c)<=0x7ff ? 2 : \
 164             ((uint32_t)(c)<=0xd7ff ? 3 : \
 165                 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
 166                     ((uint32_t)(c)<=0xffff ? 3 : 4)\
 167                 ) \
 168             ) \
 169         ) \
 170     )
 171
 172 /**
 173  * The maximum number of UTF-8 code units (bytes) per Unicode code point (U+0000..U+10ffff).
 174  * @return 4
 175  * @stable ICU 2.4
 176  */
 177 #define U8_MAX_LENGTH 4
 178
 179 /**
 180  * Get a code point from a string at a random-access offset,
 181  * without changing the offset.
 182  * The offset may point to either the lead byte or one of the trail bytes
 183  * for a code point, in which case the macro will read all of the bytes
 184  * for the code point.
 185  * The result is undefined if the offset points to an illegal UTF-8
 186  * byte sequence.
 187  * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
 188  *
 189  * @param s const uint8_t * string
 190  * @param i string offset
 191  * @param c output UChar32 variable
 192  * @see U8_GET
 193  * @stable ICU 2.4
 194  */
 195 #define U8_GET_UNSAFE(s, i, c) { \
 196     int32_t _u8_get_unsafe_index=(int32_t)(i); \
 197     U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
 198     U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
 199 }
 200
 201 /**
 202  * Get a code point from a string at a random-access offset,
 203  * without changing the offset.
 204  * The offset may point to either the lead byte or one of the trail bytes
 205  * for a code point, in which case the macro will read all of the bytes
 206  * for the code point.
 207  * If the offset points to an illegal UTF-8 byte sequence, then
 208  * c is set to a negative value.
 209  * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
 210  *
 211  * @param s const uint8_t * string
 212  * @param start starting string offset
 213  * @param i string offset, must be start<=i<length
 214  * @param length string length
 215  * @param c output UChar32 variable, set to <0 in case of an error
 216  * @see U8_GET_UNSAFE
 217  * @stable ICU 2.4
 218  */
 219 #define U8_GET(s, start, i, length, c) { \
 220     int32_t _u8_get_index=(int32_t)(i); \
 221     U8_SET_CP_START(s, start, _u8_get_index); \
 222     U8_NEXT(s, _u8_get_index, length, c); \
 223 }
 224
 225 /* definitions with forward iteration --------------------------------------- */
 226
 227 /**
 228  * Get a code point from a string at a code point boundary offset,
 229  * and advance the offset to the next code point boundary.
 230  * (Post-incrementing forward iteration.)
 231  * "Unsafe" macro, assumes well-formed UTF-8.
 232  *
 233  * The offset may point to the lead byte of a multi-byte sequence,
 234  * in which case the macro will read the whole sequence.
 235  * The result is undefined if the offset points to a trail byte
 236  * or an illegal UTF-8 sequence.
 237  *
 238  * @param s const uint8_t * string
 239  * @param i string offset
 240  * @param c output UChar32 variable
 241  * @see U8_NEXT
 242  * @stable ICU 2.4
 243  */
 244 #define U8_NEXT_UNSAFE(s, i, c) { \
 245     (c)=(uint8_t)(s)[(i)++]; \
 246     if((uint8_t)((c)-0xc0)<0x35) { \
 247         uint8_t __count=U8_COUNT_TRAIL_BYTES(c); \
 248         U8_MASK_LEAD_BYTE(c, __count); \
 249         switch(__count) { \
 250         /* each following branch falls through to the next one */ \
 251         case 3: \
 252             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
 253         case 2: \
 254             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
 255         case 1: \
 256             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
 257         /* no other branches to optimize switch() */ \
 258             break; \
 259         } \
 260     } \
 261 }
 262
 263 /**
 264  * Get a code point from a string at a code point boundary offset,
 265  * and advance the offset to the next code point boundary.
 266  * (Post-incrementing forward iteration.)
 267  * "Safe" macro, checks for illegal sequences and for string boundaries.
 268  *
 269  * The offset may point to the lead byte of a multi-byte sequence,
 270  * in which case the macro will read the whole sequence.
 271  * If the offset points to a trail byte or an illegal UTF-8 sequence, then
 272  * c is set to a negative value.
 273  *
 274  * @param s const uint8_t * string
 275  * @param i string offset, must be i<length
 276  * @param length string length
 277  * @param c output UChar32 variable, set to <0 in case of an error
 278  * @see U8_NEXT_UNSAFE
 279  * @stable ICU 2.4
 280  */
 281 #define U8_NEXT(s, i, length, c) { \
 282     (c)=(uint8_t)(s)[(i)++]; \
 283     if((c)>=0x80) { \
 284         uint8_t __t1, __t2; \
 285         if( /* handle U+1000..U+CFFF inline */ \
 286             (0xe0<(c) && (c)<=0xec) && \
 287             (((i)+1)<(length)) && \
 288             (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \
 289             (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \
 290         ) { \
 291             /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
 292             (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \
 293             (i)+=2; \
 294         } else if( /* handle U+0080..U+07FF inline */ \
 295             ((c)<0xe0 && (c)>=0xc2) && \
 296             ((i)<(length)) && \
 297             (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \
 298         ) { \
 299             (c)=(UChar)((((c)&0x1f)<<6)|__t1); \
 300             ++(i); \
 301         } else if(U8_IS_LEAD(c)) { \
 302             /* function call for "complicated" and error cases */ \
 303             (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (int32_t)(length), c, -1); \
 304         } else { \
 305             (c)=U_SENTINEL; \
 306         } \
 307     } \
 308 }
 309
 310 /**
 311  * Append a code point to a string, overwriting 1 to 4 bytes.
 312  * The offset points to the current end of the string contents
 313  * and is advanced (post-increment).
 314  * "Unsafe" macro, assumes a valid code point and sufficient space in the string.
 315  * Otherwise, the result is undefined.
 316  *
 317  * @param s const uint8_t * string buffer
 318  * @param i string offset
 319  * @param c code point to append
 320  * @see U8_APPEND
 321  * @stable ICU 2.4
 322  */
 323 #define U8_APPEND_UNSAFE(s, i, c) { \
 324     if((uint32_t)(c)<=0x7f) { \
 325         (s)[(i)++]=(uint8_t)(c); \
 326     } else { \
 327         if((uint32_t)(c)<=0x7ff) { \
 328             (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
 329         } else { \
 330             if((uint32_t)(c)<=0xffff) { \
 331                 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
 332             } else { \
 333                 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
 334                 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
 335             } \
 336             (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
 337         } \
 338         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
 339     } \
 340 }
 341
 342 /**
 343  * Append a code point to a string, overwriting 1 to 4 bytes.
 344  * The offset points to the current end of the string contents
 345  * and is advanced (post-increment).
 346  * "Safe" macro, checks for a valid code point.
 347  * If a non-ASCII code point is written, checks for sufficient space in the string.
 348  * If the code point is not valid or trail bytes do not fit,
 349  * then isError is set to TRUE.
 350  *
 351  * @param s const uint8_t * string buffer
 352  * @param i string offset, must be i<capacity
 353  * @param capacity size of the string buffer
 354  * @param c code point to append
 355  * @param isError output UBool set to TRUE if an error occurs, otherwise not modified
 356  * @see U8_APPEND_UNSAFE
 357  * @stable ICU 2.4
 358  */
 359 #define U8_APPEND(s, i, capacity, c, isError) { \
 360     if((uint32_t)(c)<=0x7f) { \
 361         (s)[(i)++]=(uint8_t)(c); \
 362     } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \
 363         (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
 364         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
 365     } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \
 366         (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
 367         (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
 368         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
 369     } else { \
 370         (i)=utf8_appendCharSafeBody(s, (int32_t)(i), (int32_t)(capacity), c, &(isError)); \
 371     } \
 372 }
 373
 374 /**
 375  * Advance the string offset from one code point boundary to the next.
 376  * (Post-incrementing iteration.)
 377  * "Unsafe" macro, assumes well-formed UTF-8.
 378  *
 379  * @param s const uint8_t * string
 380  * @param i string offset
 381  * @see U8_FWD_1
 382  * @stable ICU 2.4
 383  */
 384 #define U8_FWD_1_UNSAFE(s, i) { \
 385     (i)+=1+U8_COUNT_TRAIL_BYTES((s)[i]); \
 386 }
 387
 388 /**
 389  * Advance the string offset from one code point boundary to the next.
 390  * (Post-incrementing iteration.)
 391  * "Safe" macro, checks for illegal sequences and for string boundaries.
 392  *
 393  * @param s const uint8_t * string
 394  * @param i string offset, must be i<length
 395  * @param length string length
 396  * @see U8_FWD_1_UNSAFE
 397  * @stable ICU 2.4
 398  */
 399 #define U8_FWD_1(s, i, length) { \
 400     uint8_t __b=(uint8_t)(s)[(i)++]; \
 401     if(U8_IS_LEAD(__b)) { \
 402         uint8_t __count=U8_COUNT_TRAIL_BYTES(__b); \
 403         if((i)+__count>(length)) { \
 404             __count=(uint8_t)((length)-(i)); \
 405         } \
 406         while(__count>0 && U8_IS_TRAIL((s)[i])) { \
 407             ++(i); \
 408             --__count; \
 409         } \
 410     } \
 411 }
 412
 413 /**
 414  * Advance the string offset from one code point boundary to the n-th next one,
 415  * i.e., move forward by n code points.
 416  * (Post-incrementing iteration.)
 417  * "Unsafe" macro, assumes well-formed UTF-8.
 418  *
 419  * @param s const uint8_t * string
 420  * @param i string offset
 421  * @param n number of code points to skip
 422  * @see U8_FWD_N
 423  * @stable ICU 2.4
 424  */
 425 #define U8_FWD_N_UNSAFE(s, i, n) { \
 426     int32_t __N=(n); \
 427     while(__N>0) { \
 428         U8_FWD_1_UNSAFE(s, i); \
 429         --__N; \
 430     } \
 431 }
 432
 433 /**
 434  * Advance the string offset from one code point boundary to the n-th next one,
 435  * i.e., move forward by n code points.
 436  * (Post-incrementing iteration.)
 437  * "Safe" macro, checks for illegal sequences and for string boundaries.
 438  *
 439  * @param s const uint8_t * string
 440  * @param i string offset, must be i<length
 441  * @param length string length
 442  * @param n number of code points to skip
 443  * @see U8_FWD_N_UNSAFE
 444  * @stable ICU 2.4
 445  */
 446 #define U8_FWD_N(s, i, length, n) { \
 447     int32_t __N=(n); \
 448     while(__N>0 && (i)<(length)) { \
 449         U8_FWD_1(s, i, length); \
 450         --__N; \
 451     } \
 452 }
 453
 454 /**
 455  * Adjust a random-access offset to a code point boundary
 456  * at the start of a code point.
 457  * If the offset points to a UTF-8 trail byte,
 458  * then the offset is moved backward to the corresponding lead byte.
 459  * Otherwise, it is not modified.
 460  * "Unsafe" macro, assumes well-formed UTF-8.
 461  *
 462  * @param s const uint8_t * string
 463  * @param i string offset
 464  * @see U8_SET_CP_START
 465  * @stable ICU 2.4
 466  */
 467 #define U8_SET_CP_START_UNSAFE(s, i) { \
 468     while(U8_IS_TRAIL((s)[i])) { --(i); } \
 469 }
 470
 471 /**
 472  * Adjust a random-access offset to a code point boundary
 473  * at the start of a code point.
 474  * If the offset points to a UTF-8 trail byte,
 475  * then the offset is moved backward to the corresponding lead byte.
 476  * Otherwise, it is not modified.
 477  * "Safe" macro, checks for illegal sequences and for string boundaries.
 478  *
 479  * @param s const uint8_t * string
 480  * @param start starting string offset (usually 0)
 481  * @param i string offset, must be start<=i
 482  * @see U8_SET_CP_START_UNSAFE
 483  * @stable ICU 2.4
 484  */
 485 #define U8_SET_CP_START(s, start, i) { \
 486     if(U8_IS_TRAIL((s)[(i)])) { \
 487         (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \
 488     } \
 489 }
 490
 491 /* definitions with backward iteration -------------------------------------- */
 492
 493 /**
 494  * Move the string offset from one code point boundary to the previous one
 495  * and get the code point between them.
 496  * (Pre-decrementing backward iteration.)
 497  * "Unsafe" macro, assumes well-formed UTF-8.
 498  *
 499  * The input offset may be the same as the string length.
 500  * If the offset is behind a multi-byte sequence, then the macro will read
 501  * the whole sequence.
 502  * If the offset is behind a lead byte, then that itself
 503  * will be returned as the code point.
 504  * The result is undefined if the offset is behind an illegal UTF-8 sequence.
 505  *
 506  * @param s const uint8_t * string
 507  * @param i string offset
 508  * @param c output UChar32 variable
 509  * @see U8_PREV
 510  * @stable ICU 2.4
 511  */
 512 #define U8_PREV_UNSAFE(s, i, c) { \
 513     (c)=(uint8_t)(s)[--(i)]; \
 514     if(U8_IS_TRAIL(c)) { \
 515         uint8_t __b, __count=1, __shift=6; \
 516 \
 517         /* c is a trail byte */ \
 518         (c)&=0x3f; \
 519         for(;;) { \
 520             __b=(uint8_t)(s)[--(i)]; \
 521             if(__b>=0xc0) { \
 522                 U8_MASK_LEAD_BYTE(__b, __count); \
 523                 (c)|=(UChar32)__b<<__shift; \
 524                 break; \
 525             } else { \
 526                 (c)|=(UChar32)(__b&0x3f)<<__shift; \
 527                 ++__count; \
 528                 __shift+=6; \
 529             } \
 530         } \
 531     } \
 532 }
 533
 534 /**
 535  * Move the string offset from one code point boundary to the previous one
 536  * and get the code point between them.
 537  * (Pre-decrementing backward iteration.)
 538  * "Safe" macro, checks for illegal sequences and for string boundaries.
 539  *
 540  * The input offset may be the same as the string length.
 541  * If the offset is behind a multi-byte sequence, then the macro will read
 542  * the whole sequence.
 543  * If the offset is behind a lead byte, then that itself
 544  * will be returned as the code point.
 545  * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value.
 546  *
 547  * @param s const uint8_t * string
 548  * @param start starting string offset (usually 0)
 549  * @param i string offset, must be start<i
 550  * @param c output UChar32 variable, set to <0 in case of an error
 551  * @see U8_PREV_UNSAFE
 552  * @stable ICU 2.4
 553  */
 554 #define U8_PREV(s, start, i, c) { \
 555     (c)=(uint8_t)(s)[--(i)]; \
 556     if((c)>=0x80) { \
 557         if((c)<=0xbf) { \
 558             (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
 559         } else { \
 560             (c)=U_SENTINEL; \
 561         } \
 562     } \
 563 }
 564
 565 /**
 566  * Move the string offset from one code point boundary to the previous one.
 567  * (Pre-decrementing backward iteration.)
 568  * The input offset may be the same as the string length.
 569  * "Unsafe" macro, assumes well-formed UTF-8.
 570  *
 571  * @param s const uint8_t * string
 572  * @param i string offset
 573  * @see U8_BACK_1
 574  * @stable ICU 2.4
 575  */
 576 #define U8_BACK_1_UNSAFE(s, i) { \
 577     while(U8_IS_TRAIL((s)[--(i)])) {} \
 578 }
 579
 580 /**
 581  * Move the string offset from one code point boundary to the previous one.
 582  * (Pre-decrementing backward iteration.)
 583  * The input offset may be the same as the string length.
 584  * "Safe" macro, checks for illegal sequences and for string boundaries.
 585  *
 586  * @param s const uint8_t * string
 587  * @param start starting string offset (usually 0)
 588  * @param i string offset, must be start<i
 589  * @see U8_BACK_1_UNSAFE
 590  * @stable ICU 2.4
 591  */
 592 #define U8_BACK_1(s, start, i) { \
 593     if(U8_IS_TRAIL((s)[--(i)])) { \
 594         (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \
 595     } \
 596 }
 597
 598 /**
 599  * Move the string offset from one code point boundary to the n-th one before it,
 600  * i.e., move backward by n code points.
 601  * (Pre-decrementing backward iteration.)
 602  * The input offset may be the same as the string length.
 603  * "Unsafe" macro, assumes well-formed UTF-8.
 604  *
 605  * @param s const uint8_t * string
 606  * @param i string offset
 607  * @param n number of code points to skip
 608  * @see U8_BACK_N
 609  * @stable ICU 2.4
 610  */
 611 #define U8_BACK_N_UNSAFE(s, i, n) { \
 612     int32_t __N=(n); \
 613     while(__N>0) { \
 614         U8_BACK_1_UNSAFE(s, i); \
 615         --__N; \
 616     } \
 617 }
 618
 619 /**
 620  * Move the string offset from one code point boundary to the n-th one before it,
 621  * i.e., move backward by n code points.
 622  * (Pre-decrementing backward iteration.)
 623  * The input offset may be the same as the string length.
 624  * "Safe" macro, checks for illegal sequences and for string boundaries.
 625  *
 626  * @param s const uint8_t * string
 627  * @param start index of the start of the string
 628  * @param i string offset, must be start<i
 629  * @param n number of code points to skip
 630  * @see U8_BACK_N_UNSAFE
 631  * @stable ICU 2.4
 632  */
 633 #define U8_BACK_N(s, start, i, n) { \
 634     int32_t __N=(n); \
 635     while(__N>0 && (i)>(start)) { \
 636         U8_BACK_1(s, start, i); \
 637         --__N; \
 638     } \
 639 }
 640
 641 /**
 642  * Adjust a random-access offset to a code point boundary after a code point.
 643  * If the offset is behind a partial multi-byte sequence,
 644  * then the offset is incremented to behind the whole sequence.
 645  * Otherwise, it is not modified.
 646  * The input offset may be the same as the string length.
 647  * "Unsafe" macro, assumes well-formed UTF-8.
 648  *
 649  * @param s const uint8_t * string
 650  * @param i string offset
 651  * @see U8_SET_CP_LIMIT
 652  * @stable ICU 2.4
 653  */
 654 #define U8_SET_CP_LIMIT_UNSAFE(s, i) { \
 655     U8_BACK_1_UNSAFE(s, i); \
 656     U8_FWD_1_UNSAFE(s, i); \
 657 }
 658
 659 /**
 660  * Adjust a random-access offset to a code point boundary after a code point.
 661  * If the offset is behind a partial multi-byte sequence,
 662  * then the offset is incremented to behind the whole sequence.
 663  * Otherwise, it is not modified.
 664  * The input offset may be the same as the string length.
 665  * "Safe" macro, checks for illegal sequences and for string boundaries.
 666  *
 667  * @param s const uint8_t * string
 668  * @param start starting string offset (usually 0)
 669  * @param i string offset, must be start<=i<=length
 670  * @param length string length
 671  * @see U8_SET_CP_LIMIT_UNSAFE
 672  * @stable ICU 2.4
 673  */
 674 #define U8_SET_CP_LIMIT(s, start, i, length) { \
 675     if((start)<(i) && (i)<(length)) { \
 676         U8_BACK_1(s, start, i); \
 677         U8_FWD_1(s, i, length); \
 678     } \
 679 }
 680
 681 #endif