icuSources/common/ubidi_props.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 2004-2014, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  ubidi_props.c
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2004dec30
  16 *   created by: Markus W. Scherer
  17 *
  18 *   Low-level Unicode bidi/shaping properties access.
  19 */
  20
  21 #include "unicode/utypes.h"
  22 #include "unicode/uset.h"
  23 #include "unicode/udata.h" /* UDataInfo */
  24 #include "ucmndata.h" /* DataHeader */
  25 #include "udatamem.h"
  26 #include "uassert.h"
  27 #include "cmemory.h"
  28 #include "utrie2.h"
  29 #include "ubidi_props.h"
  30 #include "ucln_cmn.h"
  31
  32 struct UBiDiProps {
  33     UDataMemory *mem;
  34     const int32_t *indexes;
  35     const uint32_t *mirrors;
  36     const uint8_t *jgArray;
  37     const uint8_t *jgArray2;
  38
  39     UTrie2 trie;
  40     uint8_t formatVersion[4];
  41 };
  42
  43 /* ubidi_props_data.h is machine-generated by genbidi --csource */
  44 #define INCLUDED_FROM_UBIDI_PROPS_C
  45 #include "ubidi_props_data.h"
  46
  47 /* UBiDiProps singleton ----------------------------------------------------- */
  48
  49 U_CFUNC const UBiDiProps *
  50 ubidi_getSingleton() {
  51     return &ubidi_props_singleton;
  52 }
  53
  54 /* set of property starts for UnicodeSet ------------------------------------ */
  55
  56 static UBool U_CALLCONV
  57 _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
  58     (void)end;
  59     (void)value;
  60     /* add the start code point to the USet */
  61     const USetAdder *sa=(const USetAdder *)context;
  62     sa->add(sa->set, start);
  63     return TRUE;
  64 }
  65
  66 U_CFUNC void
  67 ubidi_addPropertyStarts(const UBiDiProps *bdp, const USetAdder *sa, UErrorCode *pErrorCode) {
  68     int32_t i, length;
  69     UChar32 c, start, limit;
  70
  71     const uint8_t *jgArray;
  72     uint8_t prev, jg;
  73
  74     if(U_FAILURE(*pErrorCode)) {
  75         return;
  76     }
  77
  78     /* add the start code point of each same-value range of the trie */
  79     utrie2_enum(&bdp->trie, NULL, _enumPropertyStartsRange, sa);
  80
  81     /* add the code points from the bidi mirroring table */
  82     length=bdp->indexes[UBIDI_IX_MIRROR_LENGTH];
  83     for(i=0; i<length; ++i) {
  84         c=UBIDI_GET_MIRROR_CODE_POINT(bdp->mirrors[i]);
  85         sa->addRange(sa->set, c, c+1);
  86     }
  87
  88     /* add the code points from the Joining_Group array where the value changes */
  89     start=bdp->indexes[UBIDI_IX_JG_START];
  90     limit=bdp->indexes[UBIDI_IX_JG_LIMIT];
  91     jgArray=bdp->jgArray;
  92     for(;;) {
  93         prev=0;
  94         while(start<limit) {
  95             jg=*jgArray++;
  96             if(jg!=prev) {
  97                 sa->add(sa->set, start);
  98                 prev=jg;
  99             }
 100             ++start;
 101         }
 102         if(prev!=0) {
 103             /* add the limit code point if the last value was not 0 (it is now start==limit) */
 104             sa->add(sa->set, limit);
 105         }
 106         if(limit==bdp->indexes[UBIDI_IX_JG_LIMIT]) {
 107             /* switch to the second Joining_Group range */
 108             start=bdp->indexes[UBIDI_IX_JG_START2];
 109             limit=bdp->indexes[UBIDI_IX_JG_LIMIT2];
 110             jgArray=bdp->jgArray2;
 111         } else {
 112             break;
 113         }
 114     }
 115
 116     /* add code points with hardcoded properties, plus the ones following them */
 117
 118     /* (none right now) */
 119 }
 120
 121 /* property access functions ------------------------------------------------ */
 122
 123 U_CFUNC int32_t
 124 ubidi_getMaxValue(const UBiDiProps *bdp, UProperty which) {
 125     int32_t max;
 126
 127     if(bdp==NULL) {
 128         return -1;
 129     }
 130
 131     max=bdp->indexes[UBIDI_MAX_VALUES_INDEX];
 132     switch(which) {
 133     case UCHAR_BIDI_CLASS:
 134         return (max&UBIDI_CLASS_MASK);
 135     case UCHAR_JOINING_GROUP:
 136         return (max&UBIDI_MAX_JG_MASK)>>UBIDI_MAX_JG_SHIFT;
 137     case UCHAR_JOINING_TYPE:
 138         return (max&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT;
 139     case UCHAR_BIDI_PAIRED_BRACKET_TYPE:
 140         return (max&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT;
 141     default:
 142         return -1; /* undefined */
 143     }
 144 }
 145
 146 U_CAPI UCharDirection
 147 ubidi_getClass(const UBiDiProps *bdp, UChar32 c) {
 148     uint16_t props=UTRIE2_GET16(&bdp->trie, c);
 149     return (UCharDirection)UBIDI_GET_CLASS(props);
 150 }
 151
 152 U_CFUNC UBool
 153 ubidi_isMirrored(const UBiDiProps *bdp, UChar32 c) {
 154     uint16_t props=UTRIE2_GET16(&bdp->trie, c);
 155     return (UBool)UBIDI_GET_FLAG(props, UBIDI_IS_MIRRORED_SHIFT);
 156 }
 157
 158 static UChar32
 159 getMirror(const UBiDiProps *bdp, UChar32 c, uint16_t props) {
 160     int32_t delta=UBIDI_GET_MIRROR_DELTA(props);
 161     if(delta!=UBIDI_ESC_MIRROR_DELTA) {
 162         return c+delta;
 163     } else {
 164         /* look for mirror code point in the mirrors[] table */
 165         const uint32_t *mirrors;
 166         uint32_t m;
 167         int32_t i, length;
 168         UChar32 c2;
 169
 170         mirrors=bdp->mirrors;
 171         length=bdp->indexes[UBIDI_IX_MIRROR_LENGTH];
 172
 173         /* linear search */
 174         for(i=0; i<length; ++i) {
 175             m=mirrors[i];
 176             c2=UBIDI_GET_MIRROR_CODE_POINT(m);
 177             if(c==c2) {
 178                 /* found c, return its mirror code point using the index in m */
 179                 return UBIDI_GET_MIRROR_CODE_POINT(mirrors[UBIDI_GET_MIRROR_INDEX(m)]);
 180             } else if(c<c2) {
 181                 break;
 182             }
 183         }
 184
 185         /* c not found, return it itself */
 186         return c;
 187     }
 188 }
 189
 190 U_CFUNC UChar32
 191 ubidi_getMirror(const UBiDiProps *bdp, UChar32 c) {
 192     uint16_t props=UTRIE2_GET16(&bdp->trie, c);
 193     return getMirror(bdp, c, props);
 194 }
 195
 196 U_CFUNC UBool
 197 ubidi_isBidiControl(const UBiDiProps *bdp, UChar32 c) {
 198     uint16_t props=UTRIE2_GET16(&bdp->trie, c);
 199     return (UBool)UBIDI_GET_FLAG(props, UBIDI_BIDI_CONTROL_SHIFT);
 200 }
 201
 202 U_CFUNC UBool
 203 ubidi_isJoinControl(const UBiDiProps *bdp, UChar32 c) {
 204     uint16_t props=UTRIE2_GET16(&bdp->trie, c);
 205     return (UBool)UBIDI_GET_FLAG(props, UBIDI_JOIN_CONTROL_SHIFT);
 206 }
 207
 208 U_CFUNC UJoiningType
 209 ubidi_getJoiningType(const UBiDiProps *bdp, UChar32 c) {
 210     uint16_t props=UTRIE2_GET16(&bdp->trie, c);
 211     return (UJoiningType)((props&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT);
 212 }
 213
 214 U_CFUNC UJoiningGroup
 215 ubidi_getJoiningGroup(const UBiDiProps *bdp, UChar32 c) {
 216     UChar32 start, limit;
 217
 218     start=bdp->indexes[UBIDI_IX_JG_START];
 219     limit=bdp->indexes[UBIDI_IX_JG_LIMIT];
 220     if(start<=c && c<limit) {
 221         return (UJoiningGroup)bdp->jgArray[c-start];
 222     }
 223     start=bdp->indexes[UBIDI_IX_JG_START2];
 224     limit=bdp->indexes[UBIDI_IX_JG_LIMIT2];
 225     if(start<=c && c<limit) {
 226         return (UJoiningGroup)bdp->jgArray2[c-start];
 227     }
 228     return U_JG_NO_JOINING_GROUP;
 229 }
 230
 231 U_CFUNC UBidiPairedBracketType
 232 ubidi_getPairedBracketType(const UBiDiProps *bdp, UChar32 c) {
 233     uint16_t props=UTRIE2_GET16(&bdp->trie, c);
 234     return (UBidiPairedBracketType)((props&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT);
 235 }
 236
 237 U_CFUNC UChar32
 238 ubidi_getPairedBracket(const UBiDiProps *bdp, UChar32 c) {
 239     uint16_t props=UTRIE2_GET16(&bdp->trie, c);
 240     if((props&UBIDI_BPT_MASK)==0) {
 241         return c;
 242     } else {
 243         return getMirror(bdp, c, props);
 244     }
 245 }
 246
 247 /* public API (see uchar.h) ------------------------------------------------- */
 248
 249 U_CFUNC UCharDirection
 250 u_charDirection(UChar32 c) {
 251     return ubidi_getClass(&ubidi_props_singleton, c);
 252 }
 253
 254 U_CFUNC UBool
 255 u_isMirrored(UChar32 c) {
 256     return ubidi_isMirrored(&ubidi_props_singleton, c);
 257 }
 258
 259 U_CFUNC UChar32
 260 u_charMirror(UChar32 c) {
 261     return ubidi_getMirror(&ubidi_props_singleton, c);
 262 }
 263
 264 U_STABLE UChar32 U_EXPORT2
 265 u_getBidiPairedBracket(UChar32 c) {
 266     return ubidi_getPairedBracket(&ubidi_props_singleton, c);
 267 }