sl@0: /*
sl@0: *******************************************************************************
sl@0: *
sl@0: *   Copyright (C) 1999-2005, International Business Machines
sl@0: *   Corporation and others.  All Rights Reserved.
sl@0: *
sl@0: *******************************************************************************
sl@0: *   file name:  utf8.h
sl@0: *   encoding:   US-ASCII
sl@0: *   tab size:   8 (not used)
sl@0: *   indentation:4
sl@0: *
sl@0: *   created on: 1999sep13
sl@0: *   created by: Markus W. Scherer
sl@0: */
sl@0: 
sl@0: /**
sl@0:  * \file
sl@0:  * \brief C API: 8-bit Unicode handling macros
sl@0:  * 
sl@0:  * This file defines macros to deal with 8-bit Unicode (UTF-8) code units (bytes) and strings.
sl@0:  * utf8.h is included by utf.h after unicode/umachine.h
sl@0:  * and some common definitions.
sl@0:  *
sl@0:  * For more information see utf.h and the ICU User Guide Strings chapter
sl@0:  * (http://icu.sourceforge.net/userguide/strings.html).
sl@0:  *
sl@0:  * <em>Usage:</em>
sl@0:  * ICU coding guidelines for if() statements should be followed when using these macros.
sl@0:  * Compound statements (curly braces {}) must be used  for if-else-while... 
sl@0:  * bodies and all macro statements should be terminated with semicolon.
sl@0:  */
sl@0: 
sl@0: #ifndef __UTF8_H__
sl@0: #define __UTF8_H__
sl@0: 
sl@0: /* utf.h must be included first. */
sl@0: #ifndef __UTF_H__
sl@0: #   include "unicode/utf.h"
sl@0: #endif
sl@0: 
sl@0: /* internal definitions ----------------------------------------------------- */
sl@0: 
sl@0: /**
sl@0:  * \var utf8_countTrailBytes
sl@0:  * Internal array with numbers of trail bytes for any given byte used in
sl@0:  * lead byte position.
sl@0:  * @internal
sl@0:  */
sl@0: #ifdef U_UTF8_IMPL
sl@0: U_INTERNAL const uint8_t 
sl@0: #elif defined(U_STATIC_IMPLEMENTATION)
sl@0: U_CFUNC const uint8_t
sl@0: #else
sl@0: U_CFUNC U_IMPORT const uint8_t /* U_IMPORT2? */ /*U_IMPORT*/ 
sl@0: #endif
sl@0: utf8_countTrailBytes[256];
sl@0: 
sl@0: /**
sl@0:  * Count the trail bytes for a UTF-8 lead byte.
sl@0:  * @internal
sl@0:  */
sl@0: #define U8_COUNT_TRAIL_BYTES(leadByte) (utf8_countTrailBytes[(uint8_t)leadByte])
sl@0: 
sl@0: /**
sl@0:  * Mask a UTF-8 lead byte, leave only the lower bits that form part of the code point value.
sl@0:  * @internal
sl@0:  */
sl@0: #define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
sl@0: 
sl@0: /**
sl@0:  * Function for handling "next code point" with error-checking.
sl@0:  * @internal
sl@0:  */
sl@0: U_INTERNAL UChar32 U_EXPORT2
sl@0: utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
sl@0: 
sl@0: /**
sl@0:  * Function for handling "append code point" with error-checking.
sl@0:  * @internal
sl@0:  */
sl@0: U_INTERNAL int32_t U_EXPORT2
sl@0: utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
sl@0: 
sl@0: /**
sl@0:  * Function for handling "previous code point" with error-checking.
sl@0:  * @internal
sl@0:  */
sl@0: U_INTERNAL UChar32 U_EXPORT2
sl@0: utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
sl@0: 
sl@0: /**
sl@0:  * Function for handling "skip backward one code point" with error-checking.
sl@0:  * @internal
sl@0:  */
sl@0: U_INTERNAL int32_t U_EXPORT2
sl@0: utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
sl@0: 
sl@0: /* single-code point definitions -------------------------------------------- */
sl@0: 
sl@0: /**
sl@0:  * Does this code unit (byte) encode a code point by itself (US-ASCII 0..0x7f)?
sl@0:  * @param c 8-bit code unit (byte)
sl@0:  * @return TRUE or FALSE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_IS_SINGLE(c) (((c)&0x80)==0)
sl@0: 
sl@0: /**
sl@0:  * Is this code unit (byte) a UTF-8 lead byte?
sl@0:  * @param c 8-bit code unit (byte)
sl@0:  * @return TRUE or FALSE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_IS_LEAD(c) ((uint8_t)((c)-0xc0)<0x3e)
sl@0: 
sl@0: /**
sl@0:  * Is this code unit (byte) a UTF-8 trail byte?
sl@0:  * @param c 8-bit code unit (byte)
sl@0:  * @return TRUE or FALSE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_IS_TRAIL(c) (((c)&0xc0)==0x80)
sl@0: 
sl@0: /**
sl@0:  * How many code units (bytes) are used for the UTF-8 encoding
sl@0:  * of this Unicode code point?
sl@0:  * @param c 32-bit code point
sl@0:  * @return 1..4, or 0 if c is a surrogate or not a Unicode code point
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_LENGTH(c) \
sl@0:     ((uint32_t)(c)<=0x7f ? 1 : \
sl@0:         ((uint32_t)(c)<=0x7ff ? 2 : \
sl@0:             ((uint32_t)(c)<=0xd7ff ? 3 : \
sl@0:                 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
sl@0:                     ((uint32_t)(c)<=0xffff ? 3 : 4)\
sl@0:                 ) \
sl@0:             ) \
sl@0:         ) \
sl@0:     )
sl@0: 
sl@0: /**
sl@0:  * The maximum number of UTF-8 code units (bytes) per Unicode code point (U+0000..U+10ffff).
sl@0:  * @return 4
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_MAX_LENGTH 4
sl@0: 
sl@0: /**
sl@0:  * Get a code point from a string at a random-access offset,
sl@0:  * without changing the offset.
sl@0:  * The offset may point to either the lead byte or one of the trail bytes
sl@0:  * for a code point, in which case the macro will read all of the bytes
sl@0:  * for the code point.
sl@0:  * The result is undefined if the offset points to an illegal UTF-8
sl@0:  * byte sequence.
sl@0:  * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @param c output UChar32 variable
sl@0:  * @see U8_GET
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_GET_UNSAFE(s, i, c) { \
sl@0:     int32_t _u8_get_unsafe_index=(int32_t)(i); \
sl@0:     U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
sl@0:     U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Get a code point from a string at a random-access offset,
sl@0:  * without changing the offset.
sl@0:  * The offset may point to either the lead byte or one of the trail bytes
sl@0:  * for a code point, in which case the macro will read all of the bytes
sl@0:  * for the code point.
sl@0:  * If the offset points to an illegal UTF-8 byte sequence, then
sl@0:  * c is set to a negative value.
sl@0:  * Iteration through a string is more efficient with U8_NEXT_UNSAFE or U8_NEXT.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param start starting string offset
sl@0:  * @param i string offset, start<=i<length
sl@0:  * @param length string length
sl@0:  * @param c output UChar32 variable, set to <0 in case of an error
sl@0:  * @see U8_GET_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_GET(s, start, i, length, c) { \
sl@0:     int32_t _u8_get_index=(int32_t)(i); \
sl@0:     U8_SET_CP_START(s, start, _u8_get_index); \
sl@0:     U8_NEXT(s, _u8_get_index, length, c); \
sl@0: }
sl@0: 
sl@0: /* definitions with forward iteration --------------------------------------- */
sl@0: 
sl@0: /**
sl@0:  * Get a code point from a string at a code point boundary offset,
sl@0:  * and advance the offset to the next code point boundary.
sl@0:  * (Post-incrementing forward iteration.)
sl@0:  * "Unsafe" macro, assumes well-formed UTF-8.
sl@0:  *
sl@0:  * The offset may point to the lead byte of a multi-byte sequence,
sl@0:  * in which case the macro will read the whole sequence.
sl@0:  * The result is undefined if the offset points to a trail byte
sl@0:  * or an illegal UTF-8 sequence.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @param c output UChar32 variable
sl@0:  * @see U8_NEXT
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_NEXT_UNSAFE(s, i, c) { \
sl@0:     (c)=(s)[(i)++]; \
sl@0:     if((uint8_t)((c)-0xc0)<0x35) { \
sl@0:         uint8_t __count=U8_COUNT_TRAIL_BYTES(c); \
sl@0:         U8_MASK_LEAD_BYTE(c, __count); \
sl@0:         switch(__count) { \
sl@0:         /* each following branch falls through to the next one */ \
sl@0:         case 3: \
sl@0:             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
sl@0:         case 2: \
sl@0:             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
sl@0:         case 1: \
sl@0:             (c)=((c)<<6)|((s)[(i)++]&0x3f); \
sl@0:         /* no other branches to optimize switch() */ \
sl@0:             break; \
sl@0:         } \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Get a code point from a string at a code point boundary offset,
sl@0:  * and advance the offset to the next code point boundary.
sl@0:  * (Post-incrementing forward iteration.)
sl@0:  * "Safe" macro, checks for illegal sequences and for string boundaries.
sl@0:  *
sl@0:  * The offset may point to the lead byte of a multi-byte sequence,
sl@0:  * in which case the macro will read the whole sequence.
sl@0:  * If the offset points to a trail byte or an illegal UTF-8 sequence, then
sl@0:  * c is set to a negative value.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset, i<length
sl@0:  * @param length string length
sl@0:  * @param c output UChar32 variable, set to <0 in case of an error
sl@0:  * @see U8_NEXT_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_NEXT(s, i, length, c) { \
sl@0:     (c)=(s)[(i)++]; \
sl@0:     if(((uint8_t)(c))>=0x80) { \
sl@0:         if(U8_IS_LEAD(c)) { \
sl@0:             (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (int32_t)(length), c, -1); \
sl@0:         } else { \
sl@0:             (c)=U_SENTINEL; \
sl@0:         } \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Append a code point to a string, overwriting 1 to 4 bytes.
sl@0:  * The offset points to the current end of the string contents
sl@0:  * and is advanced (post-increment).
sl@0:  * "Unsafe" macro, assumes a valid code point and sufficient space in the string.
sl@0:  * Otherwise, the result is undefined.
sl@0:  *
sl@0:  * @param s const uint8_t * string buffer
sl@0:  * @param i string offset
sl@0:  * @param c code point to append
sl@0:  * @see U8_APPEND
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_APPEND_UNSAFE(s, i, c) { \
sl@0:     if((uint32_t)(c)<=0x7f) { \
sl@0:         (s)[(i)++]=(uint8_t)(c); \
sl@0:     } else { \
sl@0:         if((uint32_t)(c)<=0x7ff) { \
sl@0:             (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
sl@0:         } else { \
sl@0:             if((uint32_t)(c)<=0xffff) { \
sl@0:                 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
sl@0:             } else { \
sl@0:                 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
sl@0:                 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
sl@0:             } \
sl@0:             (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
sl@0:         } \
sl@0:         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Append a code point to a string, overwriting 1 or 2 code units.
sl@0:  * The offset points to the current end of the string contents
sl@0:  * and is advanced (post-increment).
sl@0:  * "Safe" macro, checks for a valid code point.
sl@0:  * If a non-ASCII code point is written, checks for sufficient space in the string.
sl@0:  * If the code point is not valid or trail bytes do not fit,
sl@0:  * then isError is set to TRUE.
sl@0:  *
sl@0:  * @param s const uint8_t * string buffer
sl@0:  * @param i string offset, i<length
sl@0:  * @param length size of the string buffer
sl@0:  * @param c code point to append
sl@0:  * @param isError output UBool set to TRUE if an error occurs, otherwise not modified
sl@0:  * @see U8_APPEND_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_APPEND(s, i, length, c, isError) { \
sl@0:     if((uint32_t)(c)<=0x7f) { \
sl@0:         (s)[(i)++]=(uint8_t)(c); \
sl@0:     } else { \
sl@0:         (i)=utf8_appendCharSafeBody(s, (int32_t)(i), (int32_t)(length), c, &(isError)); \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Advance the string offset from one code point boundary to the next.
sl@0:  * (Post-incrementing iteration.)
sl@0:  * "Unsafe" macro, assumes well-formed UTF-8.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @see U8_FWD_1
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_FWD_1_UNSAFE(s, i) { \
sl@0:     (i)+=1+U8_COUNT_TRAIL_BYTES((s)[i]); \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Advance the string offset from one code point boundary to the next.
sl@0:  * (Post-incrementing iteration.)
sl@0:  * "Safe" macro, checks for illegal sequences and for string boundaries.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset, i<length
sl@0:  * @param length string length
sl@0:  * @see U8_FWD_1_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_FWD_1(s, i, length) { \
sl@0:     uint8_t __b=(s)[(i)++]; \
sl@0:     if(U8_IS_LEAD(__b)) { \
sl@0:         uint8_t __count=U8_COUNT_TRAIL_BYTES(__b); \
sl@0:         if((i)+__count>(length)) { \
sl@0:             __count=(uint8_t)((length)-(i)); \
sl@0:         } \
sl@0:         while(__count>0 && U8_IS_TRAIL((s)[i])) { \
sl@0:             ++(i); \
sl@0:             --__count; \
sl@0:         } \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Advance the string offset from one code point boundary to the n-th next one,
sl@0:  * i.e., move forward by n code points.
sl@0:  * (Post-incrementing iteration.)
sl@0:  * "Unsafe" macro, assumes well-formed UTF-8.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @param n number of code points to skip
sl@0:  * @see U8_FWD_N
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_FWD_N_UNSAFE(s, i, n) { \
sl@0:     int32_t __N=(n); \
sl@0:     while(__N>0) { \
sl@0:         U8_FWD_1_UNSAFE(s, i); \
sl@0:         --__N; \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Advance the string offset from one code point boundary to the n-th next one,
sl@0:  * i.e., move forward by n code points.
sl@0:  * (Post-incrementing iteration.)
sl@0:  * "Safe" macro, checks for illegal sequences and for string boundaries.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset, i<length
sl@0:  * @param length string length
sl@0:  * @param n number of code points to skip
sl@0:  * @see U8_FWD_N_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_FWD_N(s, i, length, n) { \
sl@0:     int32_t __N=(n); \
sl@0:     while(__N>0 && (i)<(length)) { \
sl@0:         U8_FWD_1(s, i, length); \
sl@0:         --__N; \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Adjust a random-access offset to a code point boundary
sl@0:  * at the start of a code point.
sl@0:  * If the offset points to a UTF-8 trail byte,
sl@0:  * then the offset is moved backward to the corresponding lead byte.
sl@0:  * Otherwise, it is not modified.
sl@0:  * "Unsafe" macro, assumes well-formed UTF-8.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @see U8_SET_CP_START
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_SET_CP_START_UNSAFE(s, i) { \
sl@0:     while(U8_IS_TRAIL((s)[i])) { --(i); } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Adjust a random-access offset to a code point boundary
sl@0:  * at the start of a code point.
sl@0:  * If the offset points to a UTF-8 trail byte,
sl@0:  * then the offset is moved backward to the corresponding lead byte.
sl@0:  * Otherwise, it is not modified.
sl@0:  * "Safe" macro, checks for illegal sequences and for string boundaries.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param start starting string offset (usually 0)
sl@0:  * @param i string offset, start<=i
sl@0:  * @see U8_SET_CP_START_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_SET_CP_START(s, start, i) { \
sl@0:     if(U8_IS_TRAIL((s)[(i)])) { \
sl@0:         (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /* definitions with backward iteration -------------------------------------- */
sl@0: 
sl@0: /**
sl@0:  * Move the string offset from one code point boundary to the previous one
sl@0:  * and get the code point between them.
sl@0:  * (Pre-decrementing backward iteration.)
sl@0:  * "Unsafe" macro, assumes well-formed UTF-8.
sl@0:  *
sl@0:  * The input offset may be the same as the string length.
sl@0:  * If the offset is behind a multi-byte sequence, then the macro will read
sl@0:  * the whole sequence.
sl@0:  * If the offset is behind a lead byte, then that itself
sl@0:  * will be returned as the code point.
sl@0:  * The result is undefined if the offset is behind an illegal UTF-8 sequence.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @param c output UChar32 variable
sl@0:  * @see U8_PREV
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_PREV_UNSAFE(s, i, c) { \
sl@0:     (c)=(s)[--(i)]; \
sl@0:     if(U8_IS_TRAIL(c)) { \
sl@0:         uint8_t __b, __count=1, __shift=6; \
sl@0: \
sl@0:         /* c is a trail byte */ \
sl@0:         (c)&=0x3f; \
sl@0:         for(;;) { \
sl@0:             __b=(s)[--(i)]; \
sl@0:             if(__b>=0xc0) { \
sl@0:                 U8_MASK_LEAD_BYTE(__b, __count); \
sl@0:                 (c)|=(UChar32)__b<<__shift; \
sl@0:                 break; \
sl@0:             } else { \
sl@0:                 (c)|=(UChar32)(__b&0x3f)<<__shift; \
sl@0:                 ++__count; \
sl@0:                 __shift+=6; \
sl@0:             } \
sl@0:         } \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Move the string offset from one code point boundary to the previous one
sl@0:  * and get the code point between them.
sl@0:  * (Pre-decrementing backward iteration.)
sl@0:  * "Safe" macro, checks for illegal sequences and for string boundaries.
sl@0:  *
sl@0:  * The input offset may be the same as the string length.
sl@0:  * If the offset is behind a multi-byte sequence, then the macro will read
sl@0:  * the whole sequence.
sl@0:  * If the offset is behind a lead byte, then that itself
sl@0:  * will be returned as the code point.
sl@0:  * If the offset is behind an illegal UTF-8 sequence, then c is set to a negative value.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param start starting string offset (usually 0)
sl@0:  * @param i string offset, start<=i
sl@0:  * @param c output UChar32 variable, set to <0 in case of an error
sl@0:  * @see U8_PREV_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_PREV(s, start, i, c) { \
sl@0:     (c)=(s)[--(i)]; \
sl@0:     if((c)>=0x80) { \
sl@0:         if((c)<=0xbf) { \
sl@0:             (c)=utf8_prevCharSafeBody(s, start, &(i), c, -1); \
sl@0:         } else { \
sl@0:             (c)=U_SENTINEL; \
sl@0:         } \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Move the string offset from one code point boundary to the previous one.
sl@0:  * (Pre-decrementing backward iteration.)
sl@0:  * The input offset may be the same as the string length.
sl@0:  * "Unsafe" macro, assumes well-formed UTF-8.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @see U8_BACK_1
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_BACK_1_UNSAFE(s, i) { \
sl@0:     while(U8_IS_TRAIL((s)[--(i)])) {} \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Move the string offset from one code point boundary to the previous one.
sl@0:  * (Pre-decrementing backward iteration.)
sl@0:  * The input offset may be the same as the string length.
sl@0:  * "Safe" macro, checks for illegal sequences and for string boundaries.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param start starting string offset (usually 0)
sl@0:  * @param i string offset, start<=i
sl@0:  * @see U8_BACK_1_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_BACK_1(s, start, i) { \
sl@0:     if(U8_IS_TRAIL((s)[--(i)])) { \
sl@0:         (i)=utf8_back1SafeBody(s, start, (int32_t)(i)); \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Move the string offset from one code point boundary to the n-th one before it,
sl@0:  * i.e., move backward by n code points.
sl@0:  * (Pre-decrementing backward iteration.)
sl@0:  * The input offset may be the same as the string length.
sl@0:  * "Unsafe" macro, assumes well-formed UTF-8.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @param n number of code points to skip
sl@0:  * @see U8_BACK_N
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_BACK_N_UNSAFE(s, i, n) { \
sl@0:     int32_t __N=(n); \
sl@0:     while(__N>0) { \
sl@0:         U8_BACK_1_UNSAFE(s, i); \
sl@0:         --__N; \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Move the string offset from one code point boundary to the n-th one before it,
sl@0:  * i.e., move backward by n code points.
sl@0:  * (Pre-decrementing backward iteration.)
sl@0:  * The input offset may be the same as the string length.
sl@0:  * "Safe" macro, checks for illegal sequences and for string boundaries.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param start index of the start of the string
sl@0:  * @param i string offset, i<length
sl@0:  * @param n number of code points to skip
sl@0:  * @see U8_BACK_N_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_BACK_N(s, start, i, n) { \
sl@0:     int32_t __N=(n); \
sl@0:     while(__N>0 && (i)>(start)) { \
sl@0:         U8_BACK_1(s, start, i); \
sl@0:         --__N; \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Adjust a random-access offset to a code point boundary after a code point.
sl@0:  * If the offset is behind a partial multi-byte sequence,
sl@0:  * then the offset is incremented to behind the whole sequence.
sl@0:  * Otherwise, it is not modified.
sl@0:  * The input offset may be the same as the string length.
sl@0:  * "Unsafe" macro, assumes well-formed UTF-8.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param i string offset
sl@0:  * @see U8_SET_CP_LIMIT
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_SET_CP_LIMIT_UNSAFE(s, i) { \
sl@0:     U8_BACK_1_UNSAFE(s, i); \
sl@0:     U8_FWD_1_UNSAFE(s, i); \
sl@0: }
sl@0: 
sl@0: /**
sl@0:  * Adjust a random-access offset to a code point boundary after a code point.
sl@0:  * If the offset is behind a partial multi-byte sequence,
sl@0:  * then the offset is incremented to behind the whole sequence.
sl@0:  * Otherwise, it is not modified.
sl@0:  * The input offset may be the same as the string length.
sl@0:  * "Safe" macro, checks for illegal sequences and for string boundaries.
sl@0:  *
sl@0:  * @param s const uint8_t * string
sl@0:  * @param start starting string offset (usually 0)
sl@0:  * @param i string offset, start<=i<=length
sl@0:  * @param length string length
sl@0:  * @see U8_SET_CP_LIMIT_UNSAFE
sl@0:  * @stable ICU 2.4
sl@0:  */
sl@0: #define U8_SET_CP_LIMIT(s, start, i, length) { \
sl@0:     if((start)<(i) && (i)<(length)) { \
sl@0:         U8_BACK_1(s, start, i); \
sl@0:         U8_FWD_1(s, i, length); \
sl@0:     } \
sl@0: }
sl@0: 
sl@0: #endif