ICU 57.1: utf8.h Source File

00001 /*
00002 *******************************************************************************
00003 *
00004 *   Copyright (C) 1999-2015, International Business Machines
00005 *   Corporation and others.  All Rights Reserved.
00006 *
00007 *******************************************************************************
00008 *   file name:  utf8.h
00009 *   encoding:   US-ASCII
00010 *   tab size:   8 (not used)
00011 *   indentation:4
00012 *
00013 *   created on: 1999sep13
00014 *   created by: Markus W. Scherer
00015 */
00016 
00032 #ifndef __UTF8_H__
00033 #define __UTF8_H__
00034 
00035 #include "unicode/umachine.h"
00036 #ifndef __UTF_H__
00037 #   include "unicode/utf.h"
00038 #endif
00039 
00040 /* internal definitions ----------------------------------------------------- */
00041 
00053 #ifdef U_UTF8_IMPL
00054 U_EXPORT const uint8_t 
00055 #elif defined(U_STATIC_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION)
00056 U_CFUNC const uint8_t
00057 #else
00058 U_CFUNC U_IMPORT const uint8_t /* U_IMPORT2? */ /*U_IMPORT*/ 
00059 #endif
00060 utf8_countTrailBytes[256];
00061 
00080 #define U8_COUNT_TRAIL_BYTES(leadByte) \
00081     ((uint8_t)(leadByte)<0xf0 ? \
00082         ((uint8_t)(leadByte)>=0xc0)+((uint8_t)(leadByte)>=0xe0) : \
00083         (uint8_t)(leadByte)<0xfe ? 3+((uint8_t)(leadByte)>=0xf8)+((uint8_t)(leadByte)>=0xfc) : 0)
00084 
00096 #define U8_COUNT_TRAIL_BYTES_UNSAFE(leadByte) \
00097     (((leadByte)>=0xc0)+((leadByte)>=0xe0)+((leadByte)>=0xf0))
00098 
00106 #define U8_MASK_LEAD_BYTE(leadByte, countTrailBytes) ((leadByte)&=(1<<(6-(countTrailBytes)))-1)
00107 
00117 U_STABLE UChar32 U_EXPORT2
00118 utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
00119 
00129 U_STABLE int32_t U_EXPORT2
00130 utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError);
00131 
00141 U_STABLE UChar32 U_EXPORT2
00142 utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
00143 
00153 U_STABLE int32_t U_EXPORT2
00154 utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i);
00155 
00156 /* single-code point definitions -------------------------------------------- */
00157 
00164 #define U8_IS_SINGLE(c) (((c)&0x80)==0)
00165 
00172 #define U8_IS_LEAD(c) ((uint8_t)((c)-0xc0)<0x3e)
00173 
00180 #define U8_IS_TRAIL(c) (((c)&0xc0)==0x80)
00181 
00189 #define U8_LENGTH(c) \
00190     ((uint32_t)(c)<=0x7f ? 1 : \
00191         ((uint32_t)(c)<=0x7ff ? 2 : \
00192             ((uint32_t)(c)<=0xd7ff ? 3 : \
00193                 ((uint32_t)(c)<=0xdfff || (uint32_t)(c)>0x10ffff ? 0 : \
00194                     ((uint32_t)(c)<=0xffff ? 3 : 4)\
00195                 ) \
00196             ) \
00197         ) \
00198     )
00199 
00205 #define U8_MAX_LENGTH 4
00206 
00223 #define U8_GET_UNSAFE(s, i, c) { \
00224     int32_t _u8_get_unsafe_index=(int32_t)(i); \
00225     U8_SET_CP_START_UNSAFE(s, _u8_get_unsafe_index); \
00226     U8_NEXT_UNSAFE(s, _u8_get_unsafe_index, c); \
00227 }
00228 
00250 #define U8_GET(s, start, i, length, c) { \
00251     int32_t _u8_get_index=(i); \
00252     U8_SET_CP_START(s, start, _u8_get_index); \
00253     U8_NEXT(s, _u8_get_index, length, c); \
00254 }
00255 
00281 #define U8_GET_OR_FFFD(s, start, i, length, c) { \
00282     int32_t _u8_get_index=(i); \
00283     U8_SET_CP_START(s, start, _u8_get_index); \
00284     U8_NEXT_OR_FFFD(s, _u8_get_index, length, c); \
00285 }
00286 
00287 /* definitions with forward iteration --------------------------------------- */
00288 
00306 #define U8_NEXT_UNSAFE(s, i, c) { \
00307     (c)=(uint8_t)(s)[(i)++]; \
00308     if((c)>=0x80) { \
00309         if((c)<0xe0) { \
00310             (c)=(((c)&0x1f)<<6)|((s)[(i)++]&0x3f); \
00311         } else if((c)<0xf0) { \
00312             /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
00313             (c)=(UChar)(((c)<<12)|(((s)[i]&0x3f)<<6)|((s)[(i)+1]&0x3f)); \
00314             (i)+=2; \
00315         } else { \
00316             (c)=(((c)&7)<<18)|(((s)[i]&0x3f)<<12)|(((s)[(i)+1]&0x3f)<<6)|((s)[(i)+2]&0x3f); \
00317             (i)+=3; \
00318         } \
00319     } \
00320 }
00321 
00342 #define U8_NEXT(s, i, length, c) { \
00343     (c)=(uint8_t)(s)[(i)++]; \
00344     if((c)>=0x80) { \
00345         uint8_t __t1, __t2; \
00346         if( /* handle U+1000..U+CFFF inline */ \
00347             (0xe0<(c) && (c)<=0xec) && \
00348             (((i)+1)<(length) || (length)<0) && \
00349             (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \
00350             (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \
00351         ) { \
00352             /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
00353             (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \
00354             (i)+=2; \
00355         } else if( /* handle U+0080..U+07FF inline */ \
00356             ((c)<0xe0 && (c)>=0xc2) && \
00357             ((i)!=(length)) && \
00358             (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \
00359         ) { \
00360             (c)=(((c)&0x1f)<<6)|__t1; \
00361             ++(i); \
00362         } else { \
00363             /* function call for "complicated" and error cases */ \
00364             (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -1); \
00365         } \
00366     } \
00367 }
00368 
00393 #define U8_NEXT_OR_FFFD(s, i, length, c) { \
00394     (c)=(uint8_t)(s)[(i)++]; \
00395     if((c)>=0x80) { \
00396         uint8_t __t1, __t2; \
00397         if( /* handle U+1000..U+CFFF inline */ \
00398             (0xe0<(c) && (c)<=0xec) && \
00399             (((i)+1)<(length) || (length)<0) && \
00400             (__t1=(uint8_t)((s)[i]-0x80))<=0x3f && \
00401             (__t2=(uint8_t)((s)[(i)+1]-0x80))<= 0x3f \
00402         ) { \
00403             /* no need for (c&0xf) because the upper bits are truncated after <<12 in the cast to (UChar) */ \
00404             (c)=(UChar)(((c)<<12)|(__t1<<6)|__t2); \
00405             (i)+=2; \
00406         } else if( /* handle U+0080..U+07FF inline */ \
00407             ((c)<0xe0 && (c)>=0xc2) && \
00408             ((i)!=(length)) && \
00409             (__t1=(uint8_t)((s)[i]-0x80))<=0x3f \
00410         ) { \
00411             (c)=(((c)&0x1f)<<6)|__t1; \
00412             ++(i); \
00413         } else { \
00414             /* function call for "complicated" and error cases */ \
00415             (c)=utf8_nextCharSafeBody((const uint8_t *)s, &(i), (length), c, -3); \
00416         } \
00417     } \
00418 }
00419 
00433 #define U8_APPEND_UNSAFE(s, i, c) { \
00434     if((uint32_t)(c)<=0x7f) { \
00435         (s)[(i)++]=(uint8_t)(c); \
00436     } else { \
00437         if((uint32_t)(c)<=0x7ff) { \
00438             (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
00439         } else { \
00440             if((uint32_t)(c)<=0xffff) { \
00441                 (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
00442             } else { \
00443                 (s)[(i)++]=(uint8_t)(((c)>>18)|0xf0); \
00444                 (s)[(i)++]=(uint8_t)((((c)>>12)&0x3f)|0x80); \
00445             } \
00446             (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
00447         } \
00448         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00449     } \
00450 }
00451 
00469 #define U8_APPEND(s, i, capacity, c, isError) { \
00470     if((uint32_t)(c)<=0x7f) { \
00471         (s)[(i)++]=(uint8_t)(c); \
00472     } else if((uint32_t)(c)<=0x7ff && (i)+1<(capacity)) { \
00473         (s)[(i)++]=(uint8_t)(((c)>>6)|0xc0); \
00474         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00475     } else if((uint32_t)(c)<=0xd7ff && (i)+2<(capacity)) { \
00476         (s)[(i)++]=(uint8_t)(((c)>>12)|0xe0); \
00477         (s)[(i)++]=(uint8_t)((((c)>>6)&0x3f)|0x80); \
00478         (s)[(i)++]=(uint8_t)(((c)&0x3f)|0x80); \
00479     } else { \
00480         (i)=utf8_appendCharSafeBody(s, (i), (capacity), c, &(isError)); \
00481     } \
00482 }
00483 
00494 #define U8_FWD_1_UNSAFE(s, i) { \
00495     (i)+=1+U8_COUNT_TRAIL_BYTES_UNSAFE((uint8_t)(s)[i]); \
00496 }
00497 
00511 #define U8_FWD_1(s, i, length) { \
00512     uint8_t __b=(uint8_t)(s)[(i)++]; \
00513     if(U8_IS_LEAD(__b)) { \
00514         uint8_t __count=U8_COUNT_TRAIL_BYTES(__b); \
00515         if((i)+__count>(length) && (length)>=0) { \
00516             __count=(uint8_t)((length)-(i)); \
00517         } \
00518         while(__count>0 && U8_IS_TRAIL((s)[i])) { \
00519             ++(i); \
00520             --__count; \
00521         } \
00522     } \
00523 }
00524 
00537 #define U8_FWD_N_UNSAFE(s, i, n) { \
00538     int32_t __N=(n); \
00539     while(__N>0) { \
00540         U8_FWD_1_UNSAFE(s, i); \
00541         --__N; \
00542     } \
00543 }
00544 
00560 #define U8_FWD_N(s, i, length, n) { \
00561     int32_t __N=(n); \
00562     while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \
00563         U8_FWD_1(s, i, length); \
00564         --__N; \
00565     } \
00566 }
00567 
00581 #define U8_SET_CP_START_UNSAFE(s, i) { \
00582     while(U8_IS_TRAIL((s)[i])) { --(i); } \
00583 }
00584 
00599 #define U8_SET_CP_START(s, start, i) { \
00600     if(U8_IS_TRAIL((s)[(i)])) { \
00601         (i)=utf8_back1SafeBody(s, start, (i)); \
00602     } \
00603 }
00604 
00605 /* definitions with backward iteration -------------------------------------- */
00606 
00626 #define U8_PREV_UNSAFE(s, i, c) { \
00627     (c)=(uint8_t)(s)[--(i)]; \
00628     if(U8_IS_TRAIL(c)) { \
00629         uint8_t __b, __count=1, __shift=6; \
00630 \
00631         /* c is a trail byte */ \
00632         (c)&=0x3f; \
00633         for(;;) { \
00634             __b=(uint8_t)(s)[--(i)]; \
00635             if(__b>=0xc0) { \
00636                 U8_MASK_LEAD_BYTE(__b, __count); \
00637                 (c)|=(UChar32)__b<<__shift; \
00638                 break; \
00639             } else { \
00640                 (c)|=(UChar32)(__b&0x3f)<<__shift; \
00641                 ++__count; \
00642                 __shift+=6; \
00643             } \
00644         } \
00645     } \
00646 }
00647 
00668 #define U8_PREV(s, start, i, c) { \
00669     (c)=(uint8_t)(s)[--(i)]; \
00670     if((c)>=0x80) { \
00671         (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -1); \
00672     } \
00673 }
00674 
00699 #define U8_PREV_OR_FFFD(s, start, i, c) { \
00700     (c)=(uint8_t)(s)[--(i)]; \
00701     if((c)>=0x80) { \
00702         (c)=utf8_prevCharSafeBody((const uint8_t *)s, start, &(i), c, -3); \
00703     } \
00704 }
00705 
00717 #define U8_BACK_1_UNSAFE(s, i) { \
00718     while(U8_IS_TRAIL((s)[--(i)])) {} \
00719 }
00720 
00733 #define U8_BACK_1(s, start, i) { \
00734     if(U8_IS_TRAIL((s)[--(i)])) { \
00735         (i)=utf8_back1SafeBody(s, start, (i)); \
00736     } \
00737 }
00738 
00752 #define U8_BACK_N_UNSAFE(s, i, n) { \
00753     int32_t __N=(n); \
00754     while(__N>0) { \
00755         U8_BACK_1_UNSAFE(s, i); \
00756         --__N; \
00757     } \
00758 }
00759 
00774 #define U8_BACK_N(s, start, i, n) { \
00775     int32_t __N=(n); \
00776     while(__N>0 && (i)>(start)) { \
00777         U8_BACK_1(s, start, i); \
00778         --__N; \
00779     } \
00780 }
00781 
00795 #define U8_SET_CP_LIMIT_UNSAFE(s, i) { \
00796     U8_BACK_1_UNSAFE(s, i); \
00797     U8_FWD_1_UNSAFE(s, i); \
00798 }
00799 
00817 #define U8_SET_CP_LIMIT(s, start, i, length) { \
00818     if((start)<(i) && ((i)<(length) || (length)<0)) { \
00819         U8_BACK_1(s, start, i); \
00820         U8_FWD_1(s, i, length); \
00821     } \
00822 }
00823 
00824 #endif