1 /* 2 ******************************************************************************* 3 * 4 * Copyright (C) 1999-2013, 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 * 236 * The length can be negative for a NUL-terminated string. 237 * 238 * If the offset points to an illegal UTF-8 byte sequence, then 239 * c is set to a negative value. 240 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT. 241 * 242 * @param s const uint8_t * string 243 * @param start int32_t starting string offset 244 * @param i int32_t string offset, must be start<=i<length 245 * @param length int32_t string length 246 * @param c output UChar32 variable, set to <0 in case of an error 247 * @see U8_GET_UNSAFE 248 * @stable ICU 2.4 249 */ 250 #define U8_GET(s, start, i, length, c) { \ 251 int32_t _u8_get_index=(i); \ 252 U8_SET_CP_START(s, start, _u8_get_index); \ 253 U8_NEXT(s, _u8_get_index, length, c); \ 254 } 255 256 #ifndef U_HIDE_DRAFT_API 257 /** 258 * Get a code point from a string at a random-access offset, 259 * without changing the offset. 260 * The offset may point to either the lead byte or one of the trail bytes 261 * for a code point, in which case the macro will read all of the bytes 262 * for the code point. 263 * 264 * The length can be negative for a NUL-terminated string. 265 * 266 * If the offset points to an illegal UTF-8 byte sequence, then 267 * c is set to U+FFFD. 268 * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT_OR_FFFD. 269 * 270 * This macro does not distinguish between a real U+FFFD in the text 271 * and U+FFFD returned for an ill-formed sequence. 272 * Use U8_GET() if that distinction is important. 273 * 274 * @param s const uint8_t * string 275 * @param start int32_t starting string offset 276 * @param i int32_t string offset, must be start<=i<length 277 * @param length int32_t string length 278 * @param c output UChar32 variable, set to U+FFFD in case of an error 279 * @see U8_GET 280 * @draft ICU 51 281 */ 282 #define U8_GET_OR_FFFD(s, start, i, length, c) { \ 283 int32_t _u8_get_index=(i); \ 284 U8_SET_CP_START(s, start, _u8_get_index); \ 285 U8_NEXT_OR_FFFD(s, _u8_get_index, length, c); \ 286 } 287 #endif /* U_HIDE_DRAFT_API */ 288 289 /* definitions with forward iteration --------------------------------------- */ 290 291 /** 292 * Get a code point from a string at a code point boundary offset, 293 * and advance the offset to the next code point boundary. 294 * (Post-incrementing forward iteration.) 295 * "Unsafe" macro, assumes well-formed UTF-8. 296 * 297 * The offset may point to the lead byte of a multi-byte sequence, 298 * in which case the macro will read the whole sequence. 299 * The result is undefined if the offset points to a trail byte 300 * or an illegal UTF-8 sequence. 301 * 302 * @param s const uint8_t * string 303 * @param i string offset 304 * @param c output UChar32 variable 305 * @see U8_NEXT 306 * @stable ICU 2.4 307 */ 308 #define U8_NEXT_UNSAFE(s, i, c) { \ 309 (c)=(uint8_t)(s)[(i)++]; \ 310 if((c)>=0x80) { \ 311 if((c)<0xe0) { \ 312 (c)=(((c)&0x1f)<<6)|((s)[(i)++]&0x3f); \ 313 } else if((c)<0xf0) { \ 314 /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \ 315 (c)=(UChar)(((c)<<12)|(((s)[i]&0x3f)<<6)|((s)[(i)+1]&0x3f)); \ 316 (i)+=2; \ 317 } else { \ 318 (c)=(((c)&7)<<18)|(((s)[i]&0x3f)<<12)|(((s)[(i)+1]&0x3f)<<6)|((s)[(i)+2]&0x3f); \ 319 (i)+=3; \ 320 } \ 321 } \ 322 } 323 324 /** 325 * Get a code point from a string at a code point boundary offset, 326 * and advance the offset to the next code point boundary. 327 * (Post-incrementing forward iteration.) 328 * "Safe" macro, checks for illegal sequences and for string boundaries. 329 * 330 * The length can be negative for a NUL-terminated string. 331 * 332 * The offset may point to the lead byte of a multi-byte sequence, 333 * in which case the macro will read the whole sequence. 334 * If the offset points to a trail byte or an illegal UTF-8 sequence, then 335 * c is set to a negative value. 336 * 337 * @param s const uint8_t * string 338 * @param i int32_t string offset, must be i<length 339 * @param length int32_t string length 340 * @param c output UChar32 variable, set to <0 in case of an error 341 * @see U8_NEXT_UNSAFE 342 * @stable ICU 2.4 343 */ 344 #define U8_NEXT(s, i, length, c) { \ 345 (c)=(uint8_t)(s)[(i)++]; \ 346 if((c)>=0x80) { \ 347 uint8_t __t1, __t2; \ 348 if( /* handle U+1000..U+CFFF inline */ \ 349 (0xe0<(c) && (c)<=0xec) && \ 350 (((i)+1)<(length) || (length)<0) && \ 351 (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \ 352 (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \ 353 ) { \ 354 /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \ 355 (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \ 356 (i)+=2; \ 357 } else if( /* handle U+0080..U+07FF inline */ \ 358 ((c)<0xe0 && (c)>=0xc2) && \ 359 ((i)!=(length)) && \ 360 (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \ 361 ) { \ 362 (c)=(((c)&0x1f)<<6)|__t1; \ 363 ++(i); \ 364 } else { \ 365 /* function call for "complicated" and error cases */ \ 366 (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -1); \ 367 } \ 368 } \ 369 } 370 371 #ifndef U_HIDE_DRAFT_API 372 /** 373 * Get a code point from a string at a code point boundary offset, 374 * and advance the offset to the next code point boundary. 375 * (Post-incrementing forward iteration.) 376 * "Safe" macro, checks for illegal sequences and for string boundaries. 377 * 378 * The length can be negative for a NUL-terminated string. 379 * 380 * The offset may point to the lead byte of a multi-byte sequence, 381 * in which case the macro will read the whole sequence. 382 * If the offset points to a trail byte or an illegal UTF-8 sequence, then 383 * c is set to U+FFFD. 384 * 385 * This macro does not distinguish between a real U+FFFD in the text 386 * and U+FFFD returned for an ill-formed sequence. 387 * Use U8_NEXT() if that distinction is important. 388 * 389 * @param s const uint8_t * string 390 * @param i int32_t string offset, must be i<length 391 * @param length int32_t string length 392 * @param c output UChar32 variable, set to U+FFFD in case of an error 393 * @see U8_NEXT 394 * @draft ICU 51 395 */ 396 #define U8_NEXT_OR_FFFD(s, i, length, c) { \ 397 (c)=(uint8_t)(s)[(i)++]; \ 398 if((c)>=0x80) { \ 399 uint8_t __t1, __t2; \ 400 if( /* handle U+1000..U+CFFF inline */ \ 401 (0xe0<(c) && (c)<=0xec) && \ 402 (((i)+1)<(length) || (length)<0) && \ 403 (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \ 404 (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \ 405 ) { \ 406 /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \ 407 (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \ 408 (i)+=2; \ 409 } else if( /* handle U+0080..U+07FF inline */ \ 410 ((c)<0xe0 && (c)>=0xc2) && \ 411 ((i)!=(length)) && \ 412 (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \ 413 ) { \ 414 (c)=(((c)&0x1f)<<6)|__t1; \ 415 ++(i); \ 416 } else { \ 417 /* function call for "complicated" and error cases */ \ 418 (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -3); \ 419 } \ 420 } \ 421 } 422 #endif /* U_HIDE_DRAFT_API */ 423 424 /** 425 * Append a code point to a string, overwriting 1 to 4 bytes. 426 * The offset points to the current end of the string contents 427 * and is advanced (post-increment). 428 * "Unsafe" macro, assumes a valid code point and sufficient space in the string. 429 * Otherwise, the result is undefined. 430 * 431 * @param s const uint8_t * string buffer 432 * @param i string offset 433 * @param c code point to append 434 * @see U8_APPEND 435 * @stable ICU 2.4 436 */ 437 #define U8_APPEND_UNSAFE(s, i, c) { \ 438 if((uint32_t)(c)<=0x7f) { \ 439 (s)[(i)++]=(uint8_t)(c); \ 440 } else { \ 441 if((uint32_t)(c)<=0x7ff) { \ 442 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \ 443 } else { \ 444 if((uint32_t)(c)<=0xffff) { \ 445 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \ 446 } else { \ 447 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \ 448 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \ 449 } \ 450 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \ 451 } \ 452 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \ 453 } \ 454 } 455 456 /** 457 * Append a code point to a string, overwriting 1 to 4 bytes. 458 * The offset points to the current end of the string contents 459 * and is advanced (post-increment). 460 * "Safe" macro, checks for a valid code point. 461 * If a non-ASCII code point is written, checks for sufficient space in the string. 462 * If the code point is not valid or trail bytes do not fit, 463 * then isError is set to TRUE. 464 * 465 * @param s const uint8_t * string buffer 466 * @param i int32_t string offset, must be i<capacity 467 * @param capacity int32_t size of the string buffer 468 * @param c UChar32 code point to append 469 * @param isError output UBool set to TRUE if an error occurs, otherwise not modified 470 * @see U8_APPEND_UNSAFE 471 * @stable ICU 2.4 472 */ 473 #define U8_APPEND(s, i, capacity, c, isError) { \ 474 if((uint32_t)(c)<=0x7f) { \ 475 (s)[(i)++]=(uint8_t)(c); \ 476 } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \ 477 (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \ 478 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \ 479 } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \ 480 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \ 481 (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \ 482 (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \ 483 } else { \ 484 (i)=utf8_appendCharSafeBody(s, (i), (capacity), c, &(isError)); \ 485 } \ 486 } 487 488 /** 489 * Advance the string offset from one code point boundary to the next. 490 * (Post-incrementing iteration.) 491 * "Unsafe" macro, assumes well-formed UTF-8. 492 * 493 * @param s const uint8_t * string 494 * @param i string offset 495 * @see U8_FWD_1 496 * @stable ICU 2.4 497 */ 498 #define U8_FWD_1_UNSAFE(s, i) { \ 499 (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((uint8_t)(s)[i]); \ 500 } 501 502 /** 503 * Advance the string offset from one code point boundary to the next. 504 * (Post-incrementing iteration.) 505 * "Safe" macro, checks for illegal sequences and for string boundaries. 506 * 507 * The length can be negative for a NUL-terminated string. 508 * 509 * @param s const uint8_t * string 510 * @param i int32_t string offset, must be i<length 511 * @param length int32_t string length 512 * @see U8_FWD_1_UNSAFE 513 * @stable ICU 2.4 514 */ 515 #define U8_FWD_1(s, i, length) { \ 516 uint8_t __b=(uint8_t)(s)[(i)++]; \ 517 if(U8_IS_LEAD(__b)) { \ 518 uint8_t __count=U8_COUNT_TRAIL_BYTES(__b); \ 519 if((i)+__count>(length) && (length)>=0) { \ 520 __count=(uint8_t)((length)-(i)); \ 521 } \ 522 while(__count>0 && U8_IS_TRAIL((s)[i])) { \ 523 ++(i); \ 524 --__count; \ 525 } \ 526 } \ 527 } 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) { \ 542 int32_t __N=(n); \ 543 while(__N>0) { \ 544 U8_FWD_1_UNSAFE(s, i); \ 545 --__N; \ 546 } \ 547 } 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) { \ 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 } 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) { \ 586 while(U8_IS_TRAIL((s)[i])) { --(i); } \ 587 } 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 * "Safe" macro, checks for illegal sequences and for string boundaries. 596 * 597 * @param s const uint8_t * string 598 * @param start int32_t starting string offset (usually 0) 599 * @param i int32_t string offset, must be start<=i 600 * @see U8_SET_CP_START_UNSAFE 601 * @stable ICU 2.4 602 */ 603 #define U8_SET_CP_START(s, start, i) { \ 604 if(U8_IS_TRAIL((s)[(i)])) { \ 605 (i)=utf8_back1SafeBody(s, start, (i)); \ 606 } \ 607 } 608 609 /* definitions with backward iteration -------------------------------------- */ 610 611 /** 612 * Move the string offset from one code point boundary to the previous one 613 * and get the code point between them. 614 * (Pre-decrementing backward iteration.) 615 * "Unsafe" macro, assumes well-formed UTF-8. 616 * 617 * The input offset may be the same as the string length. 618 * If the offset is behind a multi-byte sequence, then the macro will read 619 * the whole sequence. 620 * If the offset is behind a lead byte, then that itself 621 * will be returned as the code point. 622 * The result is undefined if the offset is behind an illegal UTF-8 sequence. 623 * 624 * @param s const uint8_t * string 625 * @param i string offset 626 * @param c output UChar32 variable 627 * @see U8_PREV 628 * @stable ICU 2.4 629 */ 630 #define U8_PREV_UNSAFE(s, i, c) { \ 631 (c)=(uint8_t)(s)[--(i)]; \ 632 if(U8_IS_TRAIL(c)) { \ 633 uint8_t __b, __count=1, __shift=6; \ 634 \ 635 /* c is a trail byte */ \ 636 (c)&=0x3f; \ 637 for(;;) { \ 638 __b=(uint8_t)(s)[--(i)]; \ 639 if(__b>=0xc0) { \ 640 U8_MASK_LEAD_BYTE(__b, __count); \ 641 (c)|=(UChar32)__b<<__shift; \ 642 break; \ 643 } else { \ 644 (c)|=(UChar32)(__b&0x3f)<<__shift; \ 645 ++__count; \ 646 __shift+=6; \ 647 } \ 648 } \ 649 } \ 650 } 651 652 /** 653 * Move the string offset from one code point boundary to the previous one 654 * and get the code point between them. 655 * (Pre-decrementing backward iteration.) 656 * "Safe" macro, checks for illegal sequences and for string boundaries. 657 * 658 * The input offset may be the same as the string length. 659 * If the offset is behind a multi-byte sequence, then the macro will read 660 * the whole sequence. 661 * If the offset is behind a lead byte, then that itself 662 * will be returned as the code point. 663 * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value. 664 * 665 * @param s const uint8_t * string 666 * @param start int32_t starting string offset (usually 0) 667 * @param i int32_t string offset, must be start<i 668 * @param c output UChar32 variable, set to <0 in case of an error 669 * @see U8_PREV_UNSAFE 670 * @stable ICU 2.4 671 */ 672 #define U8_PREV(s, start, i, c) { \ 673 (c)=(uint8_t)(s)[--(i)]; \ 674 if((c)>=0x80) { \ 675 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \ 676 } \ 677 } 678 679 #ifndef U_HIDE_DRAFT_API 680 /** 681 * Move the string offset from one code point boundary to the previous one 682 * and get the code point between them. 683 * (Pre-decrementing backward iteration.) 684 * "Safe" macro, checks for illegal sequences and for string boundaries. 685 * 686 * The input offset may be the same as the string length. 687 * If the offset is behind a multi-byte sequence, then the macro will read 688 * the whole sequence. 689 * If the offset is behind a lead byte, then that itself 690 * will be returned as the code point. 691 * If the offset is behind an illegal UTF-8 sequence, then c is set to U+FFFD. 692 * 693 * This macro does not distinguish between a real U+FFFD in the text 694 * and U+FFFD returned for an ill-formed sequence. 695 * Use U8_PREV() if that distinction is important. 696 * 697 * @param s const uint8_t * string 698 * @param start int32_t starting string offset (usually 0) 699 * @param i int32_t string offset, must be start<i 700 * @param c output UChar32 variable, set to U+FFFD in case of an error 701 * @see U8_PREV 702 * @draft ICU 51 703 */ 704 #define U8_PREV_OR_FFFD(s, start, i, c) { \ 705 (c)=(uint8_t)(s)[--(i)]; \ 706 if((c)>=0x80) { \ 707 (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -3); \ 708 } \ 709 } 710 #endif /* U_HIDE_DRAFT_API */ 711 712 /** 713 * Move the string offset from one code point boundary to the previous one. 714 * (Pre-decrementing backward iteration.) 715 * The input offset may be the same as the string length. 716 * "Unsafe" macro, assumes well-formed UTF-8. 717 * 718 * @param s const uint8_t * string 719 * @param i string offset 720 * @see U8_BACK_1 721 * @stable ICU 2.4 722 */ 723 #define U8_BACK_1_UNSAFE(s, i) { \ 724 while(U8_IS_TRAIL((s)[--(i)])) {} \ 725 } 726 727 /** 728 * Move the string offset from one code point boundary to the previous one. 729 * (Pre-decrementing backward iteration.) 730 * The input offset may be the same as the string length. 731 * "Safe" macro, checks for illegal sequences and for string boundaries. 732 * 733 * @param s const uint8_t * string 734 * @param start int32_t starting string offset (usually 0) 735 * @param i int32_t string offset, must be start<i 736 * @see U8_BACK_1_UNSAFE 737 * @stable ICU 2.4 738 */ 739 #define U8_BACK_1(s, start, i) { \ 740 if(U8_IS_TRAIL((s)[--(i)])) { \ 741 (i)=utf8_back1SafeBody(s, start, (i)); \ 742 } \ 743 } 744 745 /** 746 * Move the string offset from one code point boundary to the n-th one before it, 747 * i.e., move backward by n code points. 748 * (Pre-decrementing backward iteration.) 749 * The input offset may be the same as the string length. 750 * "Unsafe" macro, assumes well-formed UTF-8. 751 * 752 * @param s const uint8_t * string 753 * @param i string offset 754 * @param n number of code points to skip 755 * @see U8_BACK_N 756 * @stable ICU 2.4 757 */ 758 #define U8_BACK_N_UNSAFE(s, i, n) { \ 759 int32_t __N=(n); \ 760 while(__N>0) { \ 761 U8_BACK_1_UNSAFE(s, i); \ 762 --__N; \ 763 } \ 764 } 765 766 /** 767 * Move the string offset from one code point boundary to the n-th one before it, 768 * i.e., move backward by n code points. 769 * (Pre-decrementing backward iteration.) 770 * The input offset may be the same as the string length. 771 * "Safe" macro, checks for illegal sequences and for string boundaries. 772 * 773 * @param s const uint8_t * string 774 * @param start int32_t index of the start of the string 775 * @param i int32_t string offset, must be start<i 776 * @param n number of code points to skip 777 * @see U8_BACK_N_UNSAFE 778 * @stable ICU 2.4 779 */ 780 #define U8_BACK_N(s, start, i, n) { \ 781 int32_t __N=(n); \ 782 while(__N>0 && (i)>(start)) { \ 783 U8_BACK_1(s, start, i); \ 784 --__N; \ 785 } \ 786 } 787 788 /** 789 * Adjust a random-access offset to a code point boundary after a code point. 790 * If the offset is behind a partial multi-byte sequence, 791 * then the offset is incremented to behind the whole sequence. 792 * Otherwise, it is not modified. 793 * The input offset may be the same as the string length. 794 * "Unsafe" macro, assumes well-formed UTF-8. 795 * 796 * @param s const uint8_t * string 797 * @param i string offset 798 * @see U8_SET_CP_LIMIT 799 * @stable ICU 2.4 800 */ 801 #define U8_SET_CP_LIMIT_UNSAFE(s, i) { \ 802 U8_BACK_1_UNSAFE(s, i); \ 803 U8_FWD_1_UNSAFE(s, i); \ 804 } 805 806 /** 807 * Adjust a random-access offset to a code point boundary after a code point. 808 * If the offset is behind a partial multi-byte sequence, 809 * then the offset is incremented to behind the whole sequence. 810 * Otherwise, it is not modified. 811 * The input offset may be the same as the string length. 812 * "Safe" macro, checks for illegal sequences and for string boundaries. 813 * 814 * The length can be negative for a NUL-terminated string. 815 * 816 * @param s const uint8_t * string 817 * @param start int32_t starting string offset (usually 0) 818 * @param i int32_t string offset, must be start<=i<=length 819 * @param length int32_t string length 820 * @see U8_SET_CP_LIMIT_UNSAFE 821 * @stable ICU 2.4 822 */ 823 #define U8_SET_CP_LIMIT(s, start, i, length) { \ 824 if((start)<(i) && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \ 825 U8_BACK_1(s, start, i); \ 826 U8_FWD_1(s, i, length); \ 827 } \ 828 } 829 830 #endif 831