icuSources/common/ubidi_props.c

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