1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/unicode/utf8.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,830 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 1999-2013, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: utf8.h
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 1999sep13
1.17 +* created by: Markus W. Scherer
1.18 +*/
1.19 +
1.20 +/**
1.21 + * \file
1.22 + * \brief C API: 8-bit Unicode handling macros
1.23 + *
1.24 + * This file defines macros to deal with 8-bit Unicode (UTF-8) code units (bytes) and strings.
1.25 + *
1.26 + * For more information see utf.h and the ICU User Guide Strings chapter
1.27 + * (http://userguide.icu-project.org/strings).
1.28 + *
1.29 + * <em>Usage:</em>
1.30 + * ICU coding guidelines for if() statements should be followed when using these macros.
1.31 + * Compound statements (curly braces {}) must be used for if-else-while...
1.32 + * bodies and all macro statements should be terminated with semicolon.
1.33 + */
1.34 +
1.35 +#ifndef __UTF8_H__
1.36 +#define __UTF8_H__
1.37 +
1.38 +#include "unicode/umachine.h"
1.39 +#ifndef __UTF_H__
1.40 +# include "unicode/utf.h"
1.41 +#endif
1.42 +
1.43 +/* internal definitions ----------------------------------------------------- */
1.44 +
1.45 +/**
1.46 + * \var utf8_countTrailBytes
1.47 + * Internal array with numbers of trail bytes for any given byte used in
1.48 + * lead byte position.
1.49 + *
1.50 + * This is internal since it is not meant to be called directly by external clients;
1.51 + * however it is called by public macros in this file and thus must remain stable,
1.52 + * and should not be hidden when other internal functions are hidden (otherwise
1.53 + * public macros would fail to compile).
1.54 + * @internal
1.55 + */
1.56 +#ifdef U_UTF8_IMPL
1.57 +U_EXPORT const uint8_t
1.58 +#elif defined(U_STATIC_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION)
1.59 +U_CFUNC const uint8_t
1.60 +#else
1.61 +U_CFUNC U_IMPORT const uint8_t /* U_IMPORT2? */ /*U_IMPORT*/
1.62 +#endif
1.63 +utf8_countTrailBytes[256];
1.64 +
1.65 +/**
1.66 + * Counts the trail bytes for a UTF-8 lead byte.
1.67 + * Returns 0 for 0..0xbf as well as for 0xfe and 0xff.
1.68 + *
1.69 + * This is internal since it is not meant to be called directly by external clients;
1.70 + * however it is called by public macros in this file and thus must remain stable.
1.71 + *
1.72 + * Note: Beginning with ICU 50, the implementation uses a multi-condition expression
1.73 + * which was shown in 2012 (on x86-64) to compile to fast, branch-free code.
1.74 + * leadByte is evaluated multiple times.
1.75 + *
1.76 + * The pre-ICU 50 implementation used the exported array utf8_countTrailBytes:
1.77 + * #define U8_COUNT_TRAIL_BYTES(leadByte) (utf8_countTrailBytes[leadByte])
1.78 + * leadByte was evaluated exactly once.
1.79 + *
1.80 + * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
1.81 + * @internal
1.82 + */
1.83 +#define U8_COUNT_TRAIL_BYTES(leadByte) \
1.84 + ((leadByte)<0xf0 ? \
1.85 + ((leadByte)>=0xc0)+((leadByte)>=0xe0) : \
1.86 + (leadByte)<0xfe ? 3+((leadByte)>=0xf8)+((leadByte)>=0xfc) : 0)
1.87 +
1.88 +/**
1.89 + * Counts the trail bytes for a UTF-8 lead byte of a valid UTF-8 sequence.
1.90 + * The maximum supported lead byte is 0xf4 corresponding to U+10FFFF.
1.91 + * leadByte might be evaluated multiple times.
1.92 + *
1.93 + * This is internal since it is not meant to be called directly by external clients;
1.94 + * however it is called by public macros in this file and thus must remain stable.
1.95 + *
1.96 + * @param leadByte The first byte of a UTF-8 sequence. Must be 0..0xff.
1.97 + * @internal
1.98 + */
1.99 +#define U8_COUNT_TRAIL_BYTES_UNSAFE(leadByte) \
1.100 + (((leadByte)>=0xc0)+((leadByte)>=0xe0)+((leadByte)>=0xf0))
1.101 +
1.102 +/**
1.103 + * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value.
1.104 + *
1.105 + * This is internal since it is not meant to be called directly by external clients;
1.106 + * however it is called by public macros in this file and thus must remain stable.
1.107 + * @internal
1.108 + */
1.109 +#define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
1.110 +
1.111 +/**
1.112 + * Function for handling "next code point" with error-checking.
1.113 + *
1.114 + * This is internal since it is not meant to be called directly by external clients;
1.115 + * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
1.116 + * file and thus must remain stable, and should not be hidden when other internal
1.117 + * functions are hidden (otherwise public macros would fail to compile).
1.118 + * @internal
1.119 + */
1.120 +U_STABLE UChar32 U_EXPORT2
1.121 +utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
1.122 +
1.123 +/**
1.124 + * Function for handling "append code point" with error-checking.
1.125 + *
1.126 + * This is internal since it is not meant to be called directly by external clients;
1.127 + * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
1.128 + * file and thus must remain stable, and should not be hidden when other internal
1.129 + * functions are hidden (otherwise public macros would fail to compile).
1.130 + * @internal
1.131 + */
1.132 +U_STABLE int32_t U_EXPORT2
1.133 +utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
1.134 +
1.135 +/**
1.136 + * Function for handling "previous code point" with error-checking.
1.137 + *
1.138 + * This is internal since it is not meant to be called directly by external clients;
1.139 + * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
1.140 + * file and thus must remain stable, and should not be hidden when other internal
1.141 + * functions are hidden (otherwise public macros would fail to compile).
1.142 + * @internal
1.143 + */
1.144 +U_STABLE UChar32 U_EXPORT2
1.145 +utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
1.146 +
1.147 +/**
1.148 + * Function for handling "skip backward one code point" with error-checking.
1.149 + *
1.150 + * This is internal since it is not meant to be called directly by external clients;
1.151 + * however it is U_STABLE (not U_INTERNAL) since it is called by public macros in this
1.152 + * file and thus must remain stable, and should not be hidden when other internal
1.153 + * functions are hidden (otherwise public macros would fail to compile).
1.154 + * @internal
1.155 + */
1.156 +U_STABLE int32_t U_EXPORT2
1.157 +utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
1.158 +
1.159 +/* single-code point definitions -------------------------------------------- */
1.160 +
1.161 +/**
1.162 + * Does this code unit (byte) encode a code point by itself (US-ASCII 0..0x7f)?
1.163 + * @param c 8-bit code unit (byte)
1.164 + * @return TRUE or FALSE
1.165 + * @stable ICU 2.4
1.166 + */
1.167 +#define U8_IS_SINGLE(c) (((c)&0x80)==0)
1.168 +
1.169 +/**
1.170 + * Is this code unit (byte) a UTF-8 lead byte?
1.171 + * @param c 8-bit code unit (byte)
1.172 + * @return TRUE or FALSE
1.173 + * @stable ICU 2.4
1.174 + */
1.175 +#define U8_IS_LEAD(c) ((uint8_t)((c)-0xc0)<0x3e)
1.176 +
1.177 +/**
1.178 + * Is this code unit (byte) a UTF-8 trail byte?
1.179 + * @param c 8-bit code unit (byte)
1.180 + * @return TRUE or FALSE
1.181 + * @stable ICU 2.4
1.182 + */
1.183 +#define U8_IS_TRAIL(c) (((c)&0xc0)==0x80)
1.184 +
1.185 +/**
1.186 + * How many code units (bytes) are used for the UTF-8 encoding
1.187 + * of this Unicode code point?
1.188 + * @param c 32-bit code point
1.189 + * @return 1..4, or 0 if c is a surrogate or not a Unicode code point
1.190 + * @stable ICU 2.4
1.191 + */
1.192 +#define U8_LENGTH(c) \
1.193 + ((uint32_t)(c)<=0x7f ? 1 : \
1.194 + ((uint32_t)(c)<=0x7ff ? 2 : \
1.195 + ((uint32_t)(c)<=0xd7ff ? 3 : \
1.196 + ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
1.197 + ((uint32_t)(c)<=0xffff ? 3 : 4)\
1.198 + ) \
1.199 + ) \
1.200 + ) \
1.201 + )
1.202 +
1.203 +/**
1.204 + * The maximum number of UTF-8 code units (bytes) per Unicode code point (U+0000..U+10ffff).
1.205 + * @return 4
1.206 + * @stable ICU 2.4
1.207 + */
1.208 +#define U8_MAX_LENGTH 4
1.209 +
1.210 +/**
1.211 + * Get a code point from a string at a random-access offset,
1.212 + * without changing the offset.
1.213 + * The offset may point to either the lead byte or one of the trail bytes
1.214 + * for a code point, in which case the macro will read all of the bytes
1.215 + * for the code point.
1.216 + * The result is undefined if the offset points to an illegal UTF-8
1.217 + * byte sequence.
1.218 + * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
1.219 + *
1.220 + * @param s const uint8_t * string
1.221 + * @param i string offset
1.222 + * @param c output UChar32 variable
1.223 + * @see U8_GET
1.224 + * @stable ICU 2.4
1.225 + */
1.226 +#define U8_GET_UNSAFE(s, i, c) { \
1.227 + int32_t _u8_get_unsafe_index=(int32_t)(i); \
1.228 + U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
1.229 + U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
1.230 +}
1.231 +
1.232 +/**
1.233 + * Get a code point from a string at a random-access offset,
1.234 + * without changing the offset.
1.235 + * The offset may point to either the lead byte or one of the trail bytes
1.236 + * for a code point, in which case the macro will read all of the bytes
1.237 + * for the code point.
1.238 + *
1.239 + * The length can be negative for a NUL-terminated string.
1.240 + *
1.241 + * If the offset points to an illegal UTF-8 byte sequence, then
1.242 + * c is set to a negative value.
1.243 + * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
1.244 + *
1.245 + * @param s const uint8_t * string
1.246 + * @param start int32_t starting string offset
1.247 + * @param i int32_t string offset, must be start<=i<length
1.248 + * @param length int32_t string length
1.249 + * @param c output UChar32 variable, set to <0 in case of an error
1.250 + * @see U8_GET_UNSAFE
1.251 + * @stable ICU 2.4
1.252 + */
1.253 +#define U8_GET(s, start, i, length, c) { \
1.254 + int32_t _u8_get_index=(i); \
1.255 + U8_SET_CP_START(s, start, _u8_get_index); \
1.256 + U8_NEXT(s, _u8_get_index, length, c); \
1.257 +}
1.258 +
1.259 +#ifndef U_HIDE_DRAFT_API
1.260 +/**
1.261 + * Get a code point from a string at a random-access offset,
1.262 + * without changing the offset.
1.263 + * The offset may point to either the lead byte or one of the trail bytes
1.264 + * for a code point, in which case the macro will read all of the bytes
1.265 + * for the code point.
1.266 + *
1.267 + * The length can be negative for a NUL-terminated string.
1.268 + *
1.269 + * If the offset points to an illegal UTF-8 byte sequence, then
1.270 + * c is set to U+FFFD.
1.271 + * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT_OR_FFFD.
1.272 + *
1.273 + * This macro does not distinguish between a real U+FFFD in the text
1.274 + * and U+FFFD returned for an ill-formed sequence.
1.275 + * Use U8_GET() if that distinction is important.
1.276 + *
1.277 + * @param s const uint8_t * string
1.278 + * @param start int32_t starting string offset
1.279 + * @param i int32_t string offset, must be start<=i<length
1.280 + * @param length int32_t string length
1.281 + * @param c output UChar32 variable, set to U+FFFD in case of an error
1.282 + * @see U8_GET
1.283 + * @draft ICU 51
1.284 + */
1.285 +#define U8_GET_OR_FFFD(s, start, i, length, c) { \
1.286 + int32_t _u8_get_index=(i); \
1.287 + U8_SET_CP_START(s, start, _u8_get_index); \
1.288 + U8_NEXT_OR_FFFD(s, _u8_get_index, length, c); \
1.289 +}
1.290 +#endif /* U_HIDE_DRAFT_API */
1.291 +
1.292 +/* definitions with forward iteration --------------------------------------- */
1.293 +
1.294 +/**
1.295 + * Get a code point from a string at a code point boundary offset,
1.296 + * and advance the offset to the next code point boundary.
1.297 + * (Post-incrementing forward iteration.)
1.298 + * "Unsafe" macro, assumes well-formed UTF-8.
1.299 + *
1.300 + * The offset may point to the lead byte of a multi-byte sequence,
1.301 + * in which case the macro will read the whole sequence.
1.302 + * The result is undefined if the offset points to a trail byte
1.303 + * or an illegal UTF-8 sequence.
1.304 + *
1.305 + * @param s const uint8_t * string
1.306 + * @param i string offset
1.307 + * @param c output UChar32 variable
1.308 + * @see U8_NEXT
1.309 + * @stable ICU 2.4
1.310 + */
1.311 +#define U8_NEXT_UNSAFE(s, i, c) { \
1.312 + (c)=(uint8_t)(s)[(i)++]; \
1.313 + if((c)>=0x80) { \
1.314 + if((c)<0xe0) { \
1.315 + (c)=(((c)&0x1f)<<6)|((s)[(i)++]&0x3f); \
1.316 + } else if((c)<0xf0) { \
1.317 + /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
1.318 + (c)=(UChar)(((c)<<12)|(((s)[i]&0x3f)<<6)|((s)[(i)+1]&0x3f)); \
1.319 + (i)+=2; \
1.320 + } else { \
1.321 + (c)=(((c)&7)<<18)|(((s)[i]&0x3f)<<12)|(((s)[(i)+1]&0x3f)<<6)|((s)[(i)+2]&0x3f); \
1.322 + (i)+=3; \
1.323 + } \
1.324 + } \
1.325 +}
1.326 +
1.327 +/**
1.328 + * Get a code point from a string at a code point boundary offset,
1.329 + * and advance the offset to the next code point boundary.
1.330 + * (Post-incrementing forward iteration.)
1.331 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.332 + *
1.333 + * The length can be negative for a NUL-terminated string.
1.334 + *
1.335 + * The offset may point to the lead byte of a multi-byte sequence,
1.336 + * in which case the macro will read the whole sequence.
1.337 + * If the offset points to a trail byte or an illegal UTF-8 sequence, then
1.338 + * c is set to a negative value.
1.339 + *
1.340 + * @param s const uint8_t * string
1.341 + * @param i int32_t string offset, must be i<length
1.342 + * @param length int32_t string length
1.343 + * @param c output UChar32 variable, set to <0 in case of an error
1.344 + * @see U8_NEXT_UNSAFE
1.345 + * @stable ICU 2.4
1.346 + */
1.347 +#define U8_NEXT(s, i, length, c) { \
1.348 + (c)=(uint8_t)(s)[(i)++]; \
1.349 + if((c)>=0x80) { \
1.350 + uint8_t __t1, __t2; \
1.351 + if( /* handle U+1000..U+CFFF inline */ \
1.352 + (0xe0<(c) && (c)<=0xec) && \
1.353 + (((i)+1)<(length) || (length)<0) && \
1.354 + (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \
1.355 + (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \
1.356 + ) { \
1.357 + /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
1.358 + (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \
1.359 + (i)+=2; \
1.360 + } else if( /* handle U+0080..U+07FF inline */ \
1.361 + ((c)<0xe0 && (c)>=0xc2) && \
1.362 + ((i)!=(length)) && \
1.363 + (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \
1.364 + ) { \
1.365 + (c)=(((c)&0x1f)<<6)|__t1; \
1.366 + ++(i); \
1.367 + } else { \
1.368 + /* function call for "complicated" and error cases */ \
1.369 + (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -1); \
1.370 + } \
1.371 + } \
1.372 +}
1.373 +
1.374 +#ifndef U_HIDE_DRAFT_API
1.375 +/**
1.376 + * Get a code point from a string at a code point boundary offset,
1.377 + * and advance the offset to the next code point boundary.
1.378 + * (Post-incrementing forward iteration.)
1.379 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.380 + *
1.381 + * The length can be negative for a NUL-terminated string.
1.382 + *
1.383 + * The offset may point to the lead byte of a multi-byte sequence,
1.384 + * in which case the macro will read the whole sequence.
1.385 + * If the offset points to a trail byte or an illegal UTF-8 sequence, then
1.386 + * c is set to U+FFFD.
1.387 + *
1.388 + * This macro does not distinguish between a real U+FFFD in the text
1.389 + * and U+FFFD returned for an ill-formed sequence.
1.390 + * Use U8_NEXT() if that distinction is important.
1.391 + *
1.392 + * @param s const uint8_t * string
1.393 + * @param i int32_t string offset, must be i<length
1.394 + * @param length int32_t string length
1.395 + * @param c output UChar32 variable, set to U+FFFD in case of an error
1.396 + * @see U8_NEXT
1.397 + * @draft ICU 51
1.398 + */
1.399 +#define U8_NEXT_OR_FFFD(s, i, length, c) { \
1.400 + (c)=(uint8_t)(s)[(i)++]; \
1.401 + if((c)>=0x80) { \
1.402 + uint8_t __t1, __t2; \
1.403 + if( /* handle U+1000..U+CFFF inline */ \
1.404 + (0xe0<(c) && (c)<=0xec) && \
1.405 + (((i)+1)<(length) || (length)<0) && \
1.406 + (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \
1.407 + (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \
1.408 + ) { \
1.409 + /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
1.410 + (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \
1.411 + (i)+=2; \
1.412 + } else if( /* handle U+0080..U+07FF inline */ \
1.413 + ((c)<0xe0 && (c)>=0xc2) && \
1.414 + ((i)!=(length)) && \
1.415 + (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \
1.416 + ) { \
1.417 + (c)=(((c)&0x1f)<<6)|__t1; \
1.418 + ++(i); \
1.419 + } else { \
1.420 + /* function call for "complicated" and error cases */ \
1.421 + (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -3); \
1.422 + } \
1.423 + } \
1.424 +}
1.425 +#endif /* U_HIDE_DRAFT_API */
1.426 +
1.427 +/**
1.428 + * Append a code point to a string, overwriting 1 to 4 bytes.
1.429 + * The offset points to the current end of the string contents
1.430 + * and is advanced (post-increment).
1.431 + * "Unsafe" macro, assumes a valid code point and sufficient space in the string.
1.432 + * Otherwise, the result is undefined.
1.433 + *
1.434 + * @param s const uint8_t * string buffer
1.435 + * @param i string offset
1.436 + * @param c code point to append
1.437 + * @see U8_APPEND
1.438 + * @stable ICU 2.4
1.439 + */
1.440 +#define U8_APPEND_UNSAFE(s, i, c) { \
1.441 + if((uint32_t)(c)<=0x7f) { \
1.442 + (s)[(i)++]=(uint8_t)(c); \
1.443 + } else { \
1.444 + if((uint32_t)(c)<=0x7ff) { \
1.445 + (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
1.446 + } else { \
1.447 + if((uint32_t)(c)<=0xffff) { \
1.448 + (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
1.449 + } else { \
1.450 + (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
1.451 + (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
1.452 + } \
1.453 + (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
1.454 + } \
1.455 + (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
1.456 + } \
1.457 +}
1.458 +
1.459 +/**
1.460 + * Append a code point to a string, overwriting 1 to 4 bytes.
1.461 + * The offset points to the current end of the string contents
1.462 + * and is advanced (post-increment).
1.463 + * "Safe" macro, checks for a valid code point.
1.464 + * If a non-ASCII code point is written, checks for sufficient space in the string.
1.465 + * If the code point is not valid or trail bytes do not fit,
1.466 + * then isError is set to TRUE.
1.467 + *
1.468 + * @param s const uint8_t * string buffer
1.469 + * @param i int32_t string offset, must be i<capacity
1.470 + * @param capacity int32_t size of the string buffer
1.471 + * @param c UChar32 code point to append
1.472 + * @param isError output UBool set to TRUE if an error occurs, otherwise not modified
1.473 + * @see U8_APPEND_UNSAFE
1.474 + * @stable ICU 2.4
1.475 + */
1.476 +#define U8_APPEND(s, i, capacity, c, isError) { \
1.477 + if((uint32_t)(c)<=0x7f) { \
1.478 + (s)[(i)++]=(uint8_t)(c); \
1.479 + } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \
1.480 + (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
1.481 + (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
1.482 + } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \
1.483 + (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
1.484 + (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
1.485 + (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
1.486 + } else { \
1.487 + (i)=utf8_appendCharSafeBody(s, (i), (capacity), c, &(isError)); \
1.488 + } \
1.489 +}
1.490 +
1.491 +/**
1.492 + * Advance the string offset from one code point boundary to the next.
1.493 + * (Post-incrementing iteration.)
1.494 + * "Unsafe" macro, assumes well-formed UTF-8.
1.495 + *
1.496 + * @param s const uint8_t * string
1.497 + * @param i string offset
1.498 + * @see U8_FWD_1
1.499 + * @stable ICU 2.4
1.500 + */
1.501 +#define U8_FWD_1_UNSAFE(s, i) { \
1.502 + (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((uint8_t)(s)[i]); \
1.503 +}
1.504 +
1.505 +/**
1.506 + * Advance the string offset from one code point boundary to the next.
1.507 + * (Post-incrementing iteration.)
1.508 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.509 + *
1.510 + * The length can be negative for a NUL-terminated string.
1.511 + *
1.512 + * @param s const uint8_t * string
1.513 + * @param i int32_t string offset, must be i<length
1.514 + * @param length int32_t string length
1.515 + * @see U8_FWD_1_UNSAFE
1.516 + * @stable ICU 2.4
1.517 + */
1.518 +#define U8_FWD_1(s, i, length) { \
1.519 + uint8_t __b=(uint8_t)(s)[(i)++]; \
1.520 + if(U8_IS_LEAD(__b)) { \
1.521 + uint8_t __count=U8_COUNT_TRAIL_BYTES(__b); \
1.522 + if((i)+__count>(length) && (length)>=0) { \
1.523 + __count=(uint8_t)((length)-(i)); \
1.524 + } \
1.525 + while(__count>0 && U8_IS_TRAIL((s)[i])) { \
1.526 + ++(i); \
1.527 + --__count; \
1.528 + } \
1.529 + } \
1.530 +}
1.531 +
1.532 +/**
1.533 + * Advance the string offset from one code point boundary to the n-th next one,
1.534 + * i.e., move forward by n code points.
1.535 + * (Post-incrementing iteration.)
1.536 + * "Unsafe" macro, assumes well-formed UTF-8.
1.537 + *
1.538 + * @param s const uint8_t * string
1.539 + * @param i string offset
1.540 + * @param n number of code points to skip
1.541 + * @see U8_FWD_N
1.542 + * @stable ICU 2.4
1.543 + */
1.544 +#define U8_FWD_N_UNSAFE(s, i, n) { \
1.545 + int32_t __N=(n); \
1.546 + while(__N>0) { \
1.547 + U8_FWD_1_UNSAFE(s, i); \
1.548 + --__N; \
1.549 + } \
1.550 +}
1.551 +
1.552 +/**
1.553 + * Advance the string offset from one code point boundary to the n-th next one,
1.554 + * i.e., move forward by n code points.
1.555 + * (Post-incrementing iteration.)
1.556 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.557 + *
1.558 + * The length can be negative for a NUL-terminated string.
1.559 + *
1.560 + * @param s const uint8_t * string
1.561 + * @param i int32_t string offset, must be i<length
1.562 + * @param length int32_t string length
1.563 + * @param n number of code points to skip
1.564 + * @see U8_FWD_N_UNSAFE
1.565 + * @stable ICU 2.4
1.566 + */
1.567 +#define U8_FWD_N(s, i, length, n) { \
1.568 + int32_t __N=(n); \
1.569 + while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
1.570 + U8_FWD_1(s, i, length); \
1.571 + --__N; \
1.572 + } \
1.573 +}
1.574 +
1.575 +/**
1.576 + * Adjust a random-access offset to a code point boundary
1.577 + * at the start of a code point.
1.578 + * If the offset points to a UTF-8 trail byte,
1.579 + * then the offset is moved backward to the corresponding lead byte.
1.580 + * Otherwise, it is not modified.
1.581 + * "Unsafe" macro, assumes well-formed UTF-8.
1.582 + *
1.583 + * @param s const uint8_t * string
1.584 + * @param i string offset
1.585 + * @see U8_SET_CP_START
1.586 + * @stable ICU 2.4
1.587 + */
1.588 +#define U8_SET_CP_START_UNSAFE(s, i) { \
1.589 + while(U8_IS_TRAIL((s)[i])) { --(i); } \
1.590 +}
1.591 +
1.592 +/**
1.593 + * Adjust a random-access offset to a code point boundary
1.594 + * at the start of a code point.
1.595 + * If the offset points to a UTF-8 trail byte,
1.596 + * then the offset is moved backward to the corresponding lead byte.
1.597 + * Otherwise, it is not modified.
1.598 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.599 + *
1.600 + * @param s const uint8_t * string
1.601 + * @param start int32_t starting string offset (usually 0)
1.602 + * @param i int32_t string offset, must be start<=i
1.603 + * @see U8_SET_CP_START_UNSAFE
1.604 + * @stable ICU 2.4
1.605 + */
1.606 +#define U8_SET_CP_START(s, start, i) { \
1.607 + if(U8_IS_TRAIL((s)[(i)])) { \
1.608 + (i)=utf8_back1SafeBody(s, start, (i)); \
1.609 + } \
1.610 +}
1.611 +
1.612 +/* definitions with backward iteration -------------------------------------- */
1.613 +
1.614 +/**
1.615 + * Move the string offset from one code point boundary to the previous one
1.616 + * and get the code point between them.
1.617 + * (Pre-decrementing backward iteration.)
1.618 + * "Unsafe" macro, assumes well-formed UTF-8.
1.619 + *
1.620 + * The input offset may be the same as the string length.
1.621 + * If the offset is behind a multi-byte sequence, then the macro will read
1.622 + * the whole sequence.
1.623 + * If the offset is behind a lead byte, then that itself
1.624 + * will be returned as the code point.
1.625 + * The result is undefined if the offset is behind an illegal UTF-8 sequence.
1.626 + *
1.627 + * @param s const uint8_t * string
1.628 + * @param i string offset
1.629 + * @param c output UChar32 variable
1.630 + * @see U8_PREV
1.631 + * @stable ICU 2.4
1.632 + */
1.633 +#define U8_PREV_UNSAFE(s, i, c) { \
1.634 + (c)=(uint8_t)(s)[--(i)]; \
1.635 + if(U8_IS_TRAIL(c)) { \
1.636 + uint8_t __b, __count=1, __shift=6; \
1.637 +\
1.638 + /* c is a trail byte */ \
1.639 + (c)&=0x3f; \
1.640 + for(;;) { \
1.641 + __b=(uint8_t)(s)[--(i)]; \
1.642 + if(__b>=0xc0) { \
1.643 + U8_MASK_LEAD_BYTE(__b, __count); \
1.644 + (c)|=(UChar32)__b<<__shift; \
1.645 + break; \
1.646 + } else { \
1.647 + (c)|=(UChar32)(__b&0x3f)<<__shift; \
1.648 + ++__count; \
1.649 + __shift+=6; \
1.650 + } \
1.651 + } \
1.652 + } \
1.653 +}
1.654 +
1.655 +/**
1.656 + * Move the string offset from one code point boundary to the previous one
1.657 + * and get the code point between them.
1.658 + * (Pre-decrementing backward iteration.)
1.659 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.660 + *
1.661 + * The input offset may be the same as the string length.
1.662 + * If the offset is behind a multi-byte sequence, then the macro will read
1.663 + * the whole sequence.
1.664 + * If the offset is behind a lead byte, then that itself
1.665 + * will be returned as the code point.
1.666 + * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value.
1.667 + *
1.668 + * @param s const uint8_t * string
1.669 + * @param start int32_t starting string offset (usually 0)
1.670 + * @param i int32_t string offset, must be start<i
1.671 + * @param c output UChar32 variable, set to <0 in case of an error
1.672 + * @see U8_PREV_UNSAFE
1.673 + * @stable ICU 2.4
1.674 + */
1.675 +#define U8_PREV(s, start, i, c) { \
1.676 + (c)=(uint8_t)(s)[--(i)]; \
1.677 + if((c)>=0x80) { \
1.678 + (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
1.679 + } \
1.680 +}
1.681 +
1.682 +#ifndef U_HIDE_DRAFT_API
1.683 +/**
1.684 + * Move the string offset from one code point boundary to the previous one
1.685 + * and get the code point between them.
1.686 + * (Pre-decrementing backward iteration.)
1.687 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.688 + *
1.689 + * The input offset may be the same as the string length.
1.690 + * If the offset is behind a multi-byte sequence, then the macro will read
1.691 + * the whole sequence.
1.692 + * If the offset is behind a lead byte, then that itself
1.693 + * will be returned as the code point.
1.694 + * If the offset is behind an illegal UTF-8 sequence, then c is set to U+FFFD.
1.695 + *
1.696 + * This macro does not distinguish between a real U+FFFD in the text
1.697 + * and U+FFFD returned for an ill-formed sequence.
1.698 + * Use U8_PREV() if that distinction is important.
1.699 + *
1.700 + * @param s const uint8_t * string
1.701 + * @param start int32_t starting string offset (usually 0)
1.702 + * @param i int32_t string offset, must be start<i
1.703 + * @param c output UChar32 variable, set to U+FFFD in case of an error
1.704 + * @see U8_PREV
1.705 + * @draft ICU 51
1.706 + */
1.707 +#define U8_PREV_OR_FFFD(s, start, i, c) { \
1.708 + (c)=(uint8_t)(s)[--(i)]; \
1.709 + if((c)>=0x80) { \
1.710 + (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -3); \
1.711 + } \
1.712 +}
1.713 +#endif /* U_HIDE_DRAFT_API */
1.714 +
1.715 +/**
1.716 + * Move the string offset from one code point boundary to the previous one.
1.717 + * (Pre-decrementing backward iteration.)
1.718 + * The input offset may be the same as the string length.
1.719 + * "Unsafe" macro, assumes well-formed UTF-8.
1.720 + *
1.721 + * @param s const uint8_t * string
1.722 + * @param i string offset
1.723 + * @see U8_BACK_1
1.724 + * @stable ICU 2.4
1.725 + */
1.726 +#define U8_BACK_1_UNSAFE(s, i) { \
1.727 + while(U8_IS_TRAIL((s)[--(i)])) {} \
1.728 +}
1.729 +
1.730 +/**
1.731 + * Move the string offset from one code point boundary to the previous one.
1.732 + * (Pre-decrementing backward iteration.)
1.733 + * The input offset may be the same as the string length.
1.734 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.735 + *
1.736 + * @param s const uint8_t * string
1.737 + * @param start int32_t starting string offset (usually 0)
1.738 + * @param i int32_t string offset, must be start<i
1.739 + * @see U8_BACK_1_UNSAFE
1.740 + * @stable ICU 2.4
1.741 + */
1.742 +#define U8_BACK_1(s, start, i) { \
1.743 + if(U8_IS_TRAIL((s)[--(i)])) { \
1.744 + (i)=utf8_back1SafeBody(s, start, (i)); \
1.745 + } \
1.746 +}
1.747 +
1.748 +/**
1.749 + * Move the string offset from one code point boundary to the n-th one before it,
1.750 + * i.e., move backward by n code points.
1.751 + * (Pre-decrementing backward iteration.)
1.752 + * The input offset may be the same as the string length.
1.753 + * "Unsafe" macro, assumes well-formed UTF-8.
1.754 + *
1.755 + * @param s const uint8_t * string
1.756 + * @param i string offset
1.757 + * @param n number of code points to skip
1.758 + * @see U8_BACK_N
1.759 + * @stable ICU 2.4
1.760 + */
1.761 +#define U8_BACK_N_UNSAFE(s, i, n) { \
1.762 + int32_t __N=(n); \
1.763 + while(__N>0) { \
1.764 + U8_BACK_1_UNSAFE(s, i); \
1.765 + --__N; \
1.766 + } \
1.767 +}
1.768 +
1.769 +/**
1.770 + * Move the string offset from one code point boundary to the n-th one before it,
1.771 + * i.e., move backward by n code points.
1.772 + * (Pre-decrementing backward iteration.)
1.773 + * The input offset may be the same as the string length.
1.774 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.775 + *
1.776 + * @param s const uint8_t * string
1.777 + * @param start int32_t index of the start of the string
1.778 + * @param i int32_t string offset, must be start<i
1.779 + * @param n number of code points to skip
1.780 + * @see U8_BACK_N_UNSAFE
1.781 + * @stable ICU 2.4
1.782 + */
1.783 +#define U8_BACK_N(s, start, i, n) { \
1.784 + int32_t __N=(n); \
1.785 + while(__N>0 && (i)>(start)) { \
1.786 + U8_BACK_1(s, start, i); \
1.787 + --__N; \
1.788 + } \
1.789 +}
1.790 +
1.791 +/**
1.792 + * Adjust a random-access offset to a code point boundary after a code point.
1.793 + * If the offset is behind a partial multi-byte sequence,
1.794 + * then the offset is incremented to behind the whole sequence.
1.795 + * Otherwise, it is not modified.
1.796 + * The input offset may be the same as the string length.
1.797 + * "Unsafe" macro, assumes well-formed UTF-8.
1.798 + *
1.799 + * @param s const uint8_t * string
1.800 + * @param i string offset
1.801 + * @see U8_SET_CP_LIMIT
1.802 + * @stable ICU 2.4
1.803 + */
1.804 +#define U8_SET_CP_LIMIT_UNSAFE(s, i) { \
1.805 + U8_BACK_1_UNSAFE(s, i); \
1.806 + U8_FWD_1_UNSAFE(s, i); \
1.807 +}
1.808 +
1.809 +/**
1.810 + * Adjust a random-access offset to a code point boundary after a code point.
1.811 + * If the offset is behind a partial multi-byte sequence,
1.812 + * then the offset is incremented to behind the whole sequence.
1.813 + * Otherwise, it is not modified.
1.814 + * The input offset may be the same as the string length.
1.815 + * "Safe" macro, checks for illegal sequences and for string boundaries.
1.816 + *
1.817 + * The length can be negative for a NUL-terminated string.
1.818 + *
1.819 + * @param s const uint8_t * string
1.820 + * @param start int32_t starting string offset (usually 0)
1.821 + * @param i int32_t string offset, must be start<=i<=length
1.822 + * @param length int32_t string length
1.823 + * @see U8_SET_CP_LIMIT_UNSAFE
1.824 + * @stable ICU 2.4
1.825 + */
1.826 +#define U8_SET_CP_LIMIT(s, start, i, length) { \
1.827 + if((start)<(i) && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
1.828 + U8_BACK_1(s, start, i); \
1.829 + U8_FWD_1(s, i, length); \
1.830 + } \
1.831 +}
1.832 +
1.833 +#endif
