1 /* 2 ******************************************************************************* 3 * 4 * Copyright (C) 1999-2012, 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 * Counts the trail bytes for a UTF-8 lead byte. 64 * Returns 0 for 0..0xbf as well as for 0xfe and 0xff. 65 * 66 * This is internal since it is not meant to be called directly by external clients; 67 * however it is called by public macros in this file and thus must remain stable. 68 * 69 * Note: Beginning with ICU 50, the implementation uses a multi-condition expression 70 * which was shown in 2012 (on x86-64) to compile to fast, branch-free code. 71 * leadByte is evaluated multiple times. 72 * 73 * The pre-ICU 50 implementation used the exported array utf8_countTrailBytes: 74 * #define U8_COUNT_TRAIL_BYTES(leadByte) (utf8_countTrailBytes[leadByte]) 75 * leadByte was evaluated exactly once. 76 * 77 * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff. 78 * @internal 79 */ 80 #define U8_COUNT_TRAIL_BYTES(leadByte) \ 81 ((leadByte)<0xf0 ? \ 82 ((leadByte)>=0xc0)+((leadByte)>=0xe0) : \ 83 (leadByte)<0xfe ? 3+((leadByte)>=0xf8)+((leadByte)>=0xfc) : 0) 84 85 /** 86 * Counts the trail bytes for a UTF-8 lead byte of a valid UTF-8 sequence. 87 * The maximum supported lead byte is 0xf4 corresponding to U+10FFFF. 88 * leadByte might be evaluated multiple times. 89 * 90 * This is internal since it is not meant to be called directly by external clients; 91 * however it is called by public macros in this file and thus must remain stable. 92 * 93 * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff. 94 * @internal 95 */ 96 #define U8_COUNT_TRAIL_BYTES_UNSAFE(leadByte) \ 97 (((leadByte)>=0xc0)+((leadByte)>=0xe0)+((leadByte)>=0xf0)) 98 99 /** 100 * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value. 101 * 102 * This is internal since it is not meant to be called directly by external clients; 103 * however it is called by public macros in this file and thus must remain stable. 104 * @internal 105 */ 106 #define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1) 107 108 /** 109 * Function for handling "next code point" with error-checking. 110 * 111 * This is internal since it is not meant to be called directly by external clients; 112 * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this 113 * file and thus must remain stable, and should not be hidden when other internal 114 * functions are hidden (otherwise public macros would fail to compile). 115 * @internal 116 */ 117 U_STABLE UChar32 U_EXPORT2 118 utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict); 119 120 /** 121 * Function for handling "append code point" with error-checking. 122 * 123 * This is internal since it is not meant to be called directly by external clients; 124 * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this 125 * file and thus must remain stable, and should not be hidden when other internal 126 * functions are hidden (otherwise public macros would fail to compile). 127 * @internal 128 */ 129 U_STABLE int32_t U_EXPORT2 130 utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError); 131 132 /** 133 * Function for handling "previous code point" with error-checking. 134 * 135 * This is internal since it is not meant to be called directly by external clients; 136 * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this 137 * file and thus must remain stable, and should not be hidden when other internal 138 * functions are hidden (otherwise public macros would fail to compile). 139 * @internal 140 */ 141 U_STABLE UChar32 U_EXPORT2 142 utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict); 143 144 /** 145 * Function for handling "skip backward one code point" with error-checking. 146 * 147 * This is internal since it is not meant to be called directly by external clients; 148 * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this 149 * file and thus must remain stable, and should not be hidden when other internal 150 * functions are hidden (otherwise public macros would fail to compile). 151 * @internal 152 */ 153 U_STABLE int32_t U_EXPORT2 154 utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i); 155 156 /* single-code point definitions -------------------------------------------- */ 157 158 /** 159 * Does this code unit (byte) encode a code point by itself (US-ASCII 0..0x7f)? 160 * @param c 8-bit code unit (byte) 161 * @return TRUE or FALSE 162 * @stable ICU 2.4 163 */ 164 #define U8_IS_SINGLE(c) (((c)&0x80)==0) 165 166 /** 167 * Is this code unit (byte) a UTF-8 lead byte? 168 * @param c 8-bit code unit (byte) 169 * @return TRUE or FALSE 170 * @stable ICU 2.4 171 */ 172 #define U8_IS_LEAD(c) ((uint8_t)((c)-0xc0)<0x3e) 173 174 /** 175 * Is this code unit (byte) a UTF-8 trail byte? 176 * @param c 8-bit code unit (byte) 177 * @return TRUE or FALSE 178 * @stable ICU 2.4 179 */ 180 #define U8_IS_TRAIL(c) (((c)&0xc0)==0x80) 181 182 /** 183 * How many code units (bytes) are used for the UTF-8 encoding 184 * of this Unicode code point? 185 * @param c 32-bit code point 186 * @return 1..4, or 0 if c is a surrogate or not a Unicode code point 187 * @stable ICU 2.4 188 */ 189 #define U8_LENGTH(c) \ 190 ((uint32_t)(c)<=0x7f ? 1 : \ 191 ((uint32_t)(c)<=0x7ff ? 2 : \ 192 ((uint32_t)(c)<=0xd7ff ? 3 : \ 193 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \ 194 ((uint32_t)(c)<=0xffff ? 3 : 4)\ 195 ) \ 196 ) \ 197 ) \ 198 ) 199 200 /** 201 * The maximum number of UTF-8 code units (bytes) per Unicode code point (U+0000..U+10ffff). 202 * @return 4 203 * @stable ICU 2.4 204 */ 205 #define U8_MAX_LENGTH 4 206 207 /** 208 * Get a code point from a string at a random-access offset, 209 * without changing the offset. 210 * The offset may point to either the lead byte or one of the trail bytes 211 * for a code point, in which case the macro will read all of the bytes 212 * for the code point. 213 * The result is undefined if the offset points to an illegal UTF-8 214 * byte sequence. 215 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT. 216 * 217 * @param s const uint8_t * string 218 * @param i string offset 219 * @param c output UChar32 variable 220 * @see U8_GET 221 * @stable ICU 2.4 222 */ 223 #define U8_GET_UNSAFE(s, i, c) { \ 224 int32_t _u8_get_unsafe_index=(int32_t)(i); \ 225 U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \ 226 U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \ 227 } 228 229 /** 230 * Get a code point from a string at a random-access offset, 231 * without changing the offset. 232 * The offset may point to either the lead byte or one of the trail bytes 233 * for a code point, in which case the macro will read all of the bytes 234 * for the code point. 235 * If the offset points to an illegal UTF-8 byte sequence, then 236 * c is set to a negative value. 237 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT. 238 * 239 * @param s const uint8_t * string 240 * @param start starting string offset 241 * @param i string offset, must be start<=i<length 242 * @param length string length 243 * @param c output UChar32 variable, set to <0 in case of an error 244 * @see U8_GET_UNSAFE 245 * @stable ICU 2.4 246 */ 247 #define U8_GET(s, start, i, length, c) { \ 248 int32_t _u8_get_index=(int32_t)(i); \ 249 U8_SET_CP_START(s, start, _u8_get_index); \ 250 U8_NEXT(s, _u8_get_index, length, c); \ 251 } 252 253 /* definitions with forward iteration --------------------------------------- */ 254 255 /** 256 * Get a code point from a string at a code point boundary offset, 257 * and advance the offset to the next code point boundary. 258 * (Post-incrementing forward iteration.) 259 * "Unsafe" macro, assumes well-formed UTF-8. 260 * 261 * The offset may point to the lead byte of a multi-byte sequence, 262 * in which case the macro will read the whole sequence. 263 * The result is undefined if the offset points to a trail byte 264 * or an illegal UTF-8 sequence. 265 * 266 * @param s const uint8_t * string 267 * @param i string offset 268 * @param c output UChar32 variable 269 * @see U8_NEXT 270 * @stable ICU 2.4 271 */ 272 #define U8_NEXT_UNSAFE(s, i, c) { \ 273 (c)=(uint8_t)(s)[(i)++]; \ 274 if((c)>=0x80) { \ 275 if((c)<0xe0) { \ 276 (c)=(((c)&0x1f)<<6)|((s)[(i)++]&0x3f); \ 277 } else if((c)<0xf0) { \ 278 /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \ 279 (c)=(UChar)(((c)<<12)|(((s)[i]&0x3f)<<6)|((s)[(i)+1]&0x3f)); \ 280 (i)+=2; \ 281 } else { \ 282 (c)=(((c)&7)<<18)|(((s)[i]&0x3f)<<12)|(((s)[(i)+1]&0x3f)<<6)|((s)[(i)+2]&0x3f); \ 283 (i)+=3; \ 284 } \ 285 } \ 286 } 287 288 /** 289 * Get a code point from a string at a code point boundary offset, 290 * and advance the offset to the next code point boundary. 291 * (Post-incrementing forward iteration.) 292 * "Safe" macro, checks for illegal sequences and for string boundaries. 293 * 294 * The offset may point to the lead byte of a multi-byte sequence, 295 * in which case the macro will read the whole sequence. 296 * If the offset points to a trail byte or an illegal UTF-8 sequence, then 297 * c is set to a negative value. 298 * 299 * @param s const uint8_t * string 300 * @param i string offset, must be i<length 301 * @param length string length 302 * @param c output UChar32 variable, set to <0 in case of an error 303 * @see U8_NEXT_UNSAFE 304 * @stable ICU 2.4 305 */ 306 #define U8_NEXT(s, i, length, c) { \ 307 (c)=(uint8_t)(s)[(i)++]; \ 308 if((c)>=0x80) { \ 309 uint8_t __t1, __t2; \ 310 if( /* handle U+1000..U+CFFF inline */ \ 311 (0xe0<(c) && (c)<=0xec) && \ 312 (((i)+1)<(length)) && \ 313 (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \ 314 (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \ 315 ) { \ 316 /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \ 317 (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \ 318 (i)+=2; \ 319 } else if( /* handle U+0080..U+07FF inline */ \ 320 ((c)<0xe0 && (c)>=0xc2) && \ 321 ((i)<(length)) && \ 322 (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \ 323 ) { \ 324 (c)=(UChar)((((c)&0x1f)<<6)|__t1); \ 325 ++(i); \ 326 } else if(U8_IS_LEAD(c)) { \ 327 /* function call for "complicated" and error cases */ \ 328 (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (int32_t)(length), c, -1); \ 329 } else { \ 330 (c)=U_SENTINEL; \ 331 } \ 332 } \ 333 } 334 335 /** 336 * Append a code point to a string, overwriting 1 to 4 bytes. 337 * The offset points to the current end of the string contents 338 * and is advanced (post-increment). 339 * "Unsafe" macro, assumes a valid code point and sufficient space in the string. 340 * Otherwise, the result is undefined. 341 * 342 * @param s const uint8_t * string buffer 343 * @param i string offset 344 * @param c code point to append 345 * @see U8_APPEND 346 * @stable ICU 2.4 347 */ 348 #define U8_APPEND_UNSAFE(s, i, c) { \ 349 if((uint32_t)(c)<=0x7f) { \ 350 (s)[(i)++]=(uint8_t)(c); \ 351 } else { \ 352 if((uint32_t)(c)<=0x7ff) { \ 353 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \ 354 } else { \ 355 if((uint32_t)(c)<=0xffff) { \ 356 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \ 357 } else { \ 358 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \ 359 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \ 360 } \ 361 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \ 362 } \ 363 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \ 364 } \ 365 } 366 367 /** 368 * Append a code point to a string, overwriting 1 to 4 bytes. 369 * The offset points to the current end of the string contents 370 * and is advanced (post-increment). 371 * "Safe" macro, checks for a valid code point. 372 * If a non-ASCII code point is written, checks for sufficient space in the string. 373 * If the code point is not valid or trail bytes do not fit, 374 * then isError is set to TRUE. 375 * 376 * @param s const uint8_t * string buffer 377 * @param i string offset, must be i<capacity 378 * @param capacity size of the string buffer 379 * @param c code point to append 380 * @param isError output UBool set to TRUE if an error occurs, otherwise not modified 381 * @see U8_APPEND_UNSAFE 382 * @stable ICU 2.4 383 */ 384 #define U8_APPEND(s, i, capacity, c, isError) { \ 385 if((uint32_t)(c)<=0x7f) { \ 386 (s)[(i)++]=(uint8_t)(c); \ 387 } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \ 388 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \ 389 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \ 390 } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \ 391 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \ 392 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \ 393 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \ 394 } else { \ 395 (i)=utf8_appendCharSafeBody(s, (int32_t)(i), (int32_t)(capacity), c, &(isError)); \ 396 } \ 397 } 398 399 /** 400 * Advance the string offset from one code point boundary to the next. 401 * (Post-incrementing iteration.) 402 * "Unsafe" macro, assumes well-formed UTF-8. 403 * 404 * @param s const uint8_t * string 405 * @param i string offset 406 * @see U8_FWD_1 407 * @stable ICU 2.4 408 */ 409 #define U8_FWD_1_UNSAFE(s, i) { \ 410 (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((uint8_t)(s)[i]); \ 411 } 412 413 /** 414 * Advance the string offset from one code point boundary to the next. 415 * (Post-incrementing iteration.) 416 * "Safe" macro, checks for illegal sequences and for string boundaries. 417 * 418 * @param s const uint8_t * string 419 * @param i string offset, must be i<length 420 * @param length string length 421 * @see U8_FWD_1_UNSAFE 422 * @stable ICU 2.4 423 */ 424 #define U8_FWD_1(s, i, length) { \ 425 uint8_t __b=(uint8_t)(s)[(i)++]; \ 426 if(U8_IS_LEAD(__b)) { \ 427 uint8_t __count=U8_COUNT_TRAIL_BYTES(__b); \ 428 if((i)+__count>(length)) { \ 429 __count=(uint8_t)((length)-(i)); \ 430 } \ 431 while(__count>0 && U8_IS_TRAIL((s)[i])) { \ 432 ++(i); \ 433 --__count; \ 434 } \ 435 } \ 436 } 437 438 /** 439 * Advance the string offset from one code point boundary to the n-th next one, 440 * i.e., move forward by n code points. 441 * (Post-incrementing iteration.) 442 * "Unsafe" macro, assumes well-formed UTF-8. 443 * 444 * @param s const uint8_t * string 445 * @param i string offset 446 * @param n number of code points to skip 447 * @see U8_FWD_N 448 * @stable ICU 2.4 449 */ 450 #define U8_FWD_N_UNSAFE(s, i, n) { \ 451 int32_t __N=(n); \ 452 while(__N>0) { \ 453 U8_FWD_1_UNSAFE(s, i); \ 454 --__N; \ 455 } \ 456 } 457 458 /** 459 * Advance the string offset from one code point boundary to the n-th next one, 460 * i.e., move forward by n code points. 461 * (Post-incrementing iteration.) 462 * "Safe" macro, checks for illegal sequences and for string boundaries. 463 * 464 * @param s const uint8_t * string 465 * @param i string offset, must be i<length 466 * @param length string length 467 * @param n number of code points to skip 468 * @see U8_FWD_N_UNSAFE 469 * @stable ICU 2.4 470 */ 471 #define U8_FWD_N(s, i, length, n) { \ 472 int32_t __N=(n); \ 473 while(__N>0 && (i)<(length)) { \ 474 U8_FWD_1(s, i, length); \ 475 --__N; \ 476 } \ 477 } 478 479 /** 480 * Adjust a random-access offset to a code point boundary 481 * at the start of a code point. 482 * If the offset points to a UTF-8 trail byte, 483 * then the offset is moved backward to the corresponding lead byte. 484 * Otherwise, it is not modified. 485 * "Unsafe" macro, assumes well-formed UTF-8. 486 * 487 * @param s const uint8_t * string 488 * @param i string offset 489 * @see U8_SET_CP_START 490 * @stable ICU 2.4 491 */ 492 #define U8_SET_CP_START_UNSAFE(s, i) { \ 493 while(U8_IS_TRAIL((s)[i])) { --(i); } \ 494 } 495 496 /** 497 * Adjust a random-access offset to a code point boundary 498 * at the start of a code point. 499 * If the offset points to a UTF-8 trail byte, 500 * then the offset is moved backward to the corresponding lead byte. 501 * Otherwise, it is not modified. 502 * "Safe" macro, checks for illegal sequences and for string boundaries. 503 * 504 * @param s const uint8_t * string 505 * @param start starting string offset (usually 0) 506 * @param i string offset, must be start<=i 507 * @see U8_SET_CP_START_UNSAFE 508 * @stable ICU 2.4 509 */ 510 #define U8_SET_CP_START(s, start, i) { \ 511 if(U8_IS_TRAIL((s)[(i)])) { \ 512 (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \ 513 } \ 514 } 515 516 /* definitions with backward iteration -------------------------------------- */ 517 518 /** 519 * Move the string offset from one code point boundary to the previous one 520 * and get the code point between them. 521 * (Pre-decrementing backward iteration.) 522 * "Unsafe" macro, assumes well-formed UTF-8. 523 * 524 * The input offset may be the same as the string length. 525 * If the offset is behind a multi-byte sequence, then the macro will read 526 * the whole sequence. 527 * If the offset is behind a lead byte, then that itself 528 * will be returned as the code point. 529 * The result is undefined if the offset is behind an illegal UTF-8 sequence. 530 * 531 * @param s const uint8_t * string 532 * @param i string offset 533 * @param c output UChar32 variable 534 * @see U8_PREV 535 * @stable ICU 2.4 536 */ 537 #define U8_PREV_UNSAFE(s, i, c) { \ 538 (c)=(uint8_t)(s)[--(i)]; \ 539 if(U8_IS_TRAIL(c)) { \ 540 uint8_t __b, __count=1, __shift=6; \ 541 \ 542 /* c is a trail byte */ \ 543 (c)&=0x3f; \ 544 for(;;) { \ 545 __b=(uint8_t)(s)[--(i)]; \ 546 if(__b>=0xc0) { \ 547 U8_MASK_LEAD_BYTE(__b, __count); \ 548 (c)|=(UChar32)__b<<__shift; \ 549 break; \ 550 } else { \ 551 (c)|=(UChar32)(__b&0x3f)<<__shift; \ 552 ++__count; \ 553 __shift+=6; \ 554 } \ 555 } \ 556 } \ 557 } 558 559 /** 560 * Move the string offset from one code point boundary to the previous one 561 * and get the code point between them. 562 * (Pre-decrementing backward iteration.) 563 * "Safe" macro, checks for illegal sequences and for string boundaries. 564 * 565 * The input offset may be the same as the string length. 566 * If the offset is behind a multi-byte sequence, then the macro will read 567 * the whole sequence. 568 * If the offset is behind a lead byte, then that itself 569 * will be returned as the code point. 570 * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value. 571 * 572 * @param s const uint8_t * string 573 * @param start starting string offset (usually 0) 574 * @param i string offset, must be start<i 575 * @param c output UChar32 variable, set to <0 in case of an error 576 * @see U8_PREV_UNSAFE 577 * @stable ICU 2.4 578 */ 579 #define U8_PREV(s, start, i, c) { \ 580 (c)=(uint8_t)(s)[--(i)]; \ 581 if((c)>=0x80) { \ 582 if((c)<=0xbf) { \ 583 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \ 584 } else { \ 585 (c)=U_SENTINEL; \ 586 } \ 587 } \ 588 } 589 590 /** 591 * Move the string offset from one code point boundary to the previous one. 592 * (Pre-decrementing backward iteration.) 593 * The input offset may be the same as the string length. 594 * "Unsafe" macro, assumes well-formed UTF-8. 595 * 596 * @param s const uint8_t * string 597 * @param i string offset 598 * @see U8_BACK_1 599 * @stable ICU 2.4 600 */ 601 #define U8_BACK_1_UNSAFE(s, i) { \ 602 while(U8_IS_TRAIL((s)[--(i)])) {} \ 603 } 604 605 /** 606 * Move the string offset from one code point boundary to the previous one. 607 * (Pre-decrementing backward iteration.) 608 * The input offset may be the same as the string length. 609 * "Safe" macro, checks for illegal sequences and for string boundaries. 610 * 611 * @param s const uint8_t * string 612 * @param start starting string offset (usually 0) 613 * @param i string offset, must be start<i 614 * @see U8_BACK_1_UNSAFE 615 * @stable ICU 2.4 616 */ 617 #define U8_BACK_1(s, start, i) { \ 618 if(U8_IS_TRAIL((s)[--(i)])) { \ 619 (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \ 620 } \ 621 } 622 623 /** 624 * Move the string offset from one code point boundary to the n-th one before it, 625 * i.e., move backward by n code points. 626 * (Pre-decrementing backward iteration.) 627 * The input offset may be the same as the string length. 628 * "Unsafe" macro, assumes well-formed UTF-8. 629 * 630 * @param s const uint8_t * string 631 * @param i string offset 632 * @param n number of code points to skip 633 * @see U8_BACK_N 634 * @stable ICU 2.4 635 */ 636 #define U8_BACK_N_UNSAFE(s, i, n) { \ 637 int32_t __N=(n); \ 638 while(__N>0) { \ 639 U8_BACK_1_UNSAFE(s, i); \ 640 --__N; \ 641 } \ 642 } 643 644 /** 645 * Move the string offset from one code point boundary to the n-th one before it, 646 * i.e., move backward by n code points. 647 * (Pre-decrementing backward iteration.) 648 * The input offset may be the same as the string length. 649 * "Safe" macro, checks for illegal sequences and for string boundaries. 650 * 651 * @param s const uint8_t * string 652 * @param start index of the start of the string 653 * @param i string offset, must be start<i 654 * @param n number of code points to skip 655 * @see U8_BACK_N_UNSAFE 656 * @stable ICU 2.4 657 */ 658 #define U8_BACK_N(s, start, i, n) { \ 659 int32_t __N=(n); \ 660 while(__N>0 && (i)>(start)) { \ 661 U8_BACK_1(s, start, i); \ 662 --__N; \ 663 } \ 664 } 665 666 /** 667 * Adjust a random-access offset to a code point boundary after a code point. 668 * If the offset is behind a partial multi-byte sequence, 669 * then the offset is incremented to behind the whole sequence. 670 * Otherwise, it is not modified. 671 * The input offset may be the same as the string length. 672 * "Unsafe" macro, assumes well-formed UTF-8. 673 * 674 * @param s const uint8_t * string 675 * @param i string offset 676 * @see U8_SET_CP_LIMIT 677 * @stable ICU 2.4 678 */ 679 #define U8_SET_CP_LIMIT_UNSAFE(s, i) { \ 680 U8_BACK_1_UNSAFE(s, i); \ 681 U8_FWD_1_UNSAFE(s, i); \ 682 } 683 684 /** 685 * Adjust a random-access offset to a code point boundary after a code point. 686 * If the offset is behind a partial multi-byte sequence, 687 * then the offset is incremented to behind the whole sequence. 688 * Otherwise, it is not modified. 689 * The input offset may be the same as the string length. 690 * "Safe" macro, checks for illegal sequences and for string boundaries. 691 * 692 * @param s const uint8_t * string 693 * @param start starting string offset (usually 0) 694 * @param i string offset, must be start<=i<=length 695 * @param length string length 696 * @see U8_SET_CP_LIMIT_UNSAFE 697 * @stable ICU 2.4 698 */ 699 #define U8_SET_CP_LIMIT(s, start, i, length) { \ 700 if((start)<(i) && (i)<(length)) { \ 701 U8_BACK_1(s, start, i); \ 702 U8_FWD_1(s, i, length); \ 703 } \ 704 } 705 706 #endif 707