intl/icu/source/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 /* set of property starts for UnicodeSet ------------------------------------ */
  48
  49 static UBool U_CALLCONV
  50 _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
  51     (void)end;
  52     (void)value;
  53     /* add the start code point to the USet */
  54     const USetAdder *sa=(const USetAdder *)context;
  55     sa->add(sa->set, start);
  56     return true;
  57 }
  58
  59 U_CFUNC void
  60 ubidi_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
  61     int32_t i, length;
  62     UChar32 c, start, limit;
  63
  64     const uint8_t *jgArray;
  65     uint8_t prev, jg;
  66
  67     if(U_FAILURE(*pErrorCode)) {
  68         return;
  69     }
  70
  71     /* add the start code point of each same-value range of the trie */
  72     utrie2_enum(&ubidi_props_singleton.trie, nullptr, _enumPropertyStartsRange, sa);
  73
  74     /* add the code points from the bidi mirroring table */
  75     length=ubidi_props_singleton.indexes[UBIDI_IX_MIRROR_LENGTH];
  76     for(i=0; i<length; ++i) {
  77         c=UBIDI_GET_MIRROR_CODE_POINT(ubidi_props_singleton.mirrors[i]);
  78         sa->addRange(sa->set, c, c+1);
  79     }
  80
  81     /* add the code points from the Joining_Group array where the value changes */
  82     start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START];
  83     limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT];
  84     jgArray=ubidi_props_singleton.jgArray;
  85     for(;;) {
  86         prev=0;
  87         while(start<limit) {
  88             jg=*jgArray++;
  89             if(jg!=prev) {
  90                 sa->add(sa->set, start);
  91                 prev=jg;
  92             }
  93             ++start;
  94         }
  95         if(prev!=0) {
  96             /* add the limit code point if the last value was not 0 (it is now start==limit) */
  97             sa->add(sa->set, limit);
  98         }
  99         if(limit==ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT]) {
 100             /* switch to the second Joining_Group range */
 101             start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START2];
 102             limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT2];
 103             jgArray=ubidi_props_singleton.jgArray2;
 104         } else {
 105             break;
 106         }
 107     }
 108
 109     /* add code points with hardcoded properties, plus the ones following them */
 110
 111     /* (none right now) */
 112 }
 113
 114 /* property access functions ------------------------------------------------ */
 115
 116 U_CFUNC int32_t
 117 ubidi_getMaxValue(UProperty which) {
 118     int32_t max=ubidi_props_singleton.indexes[UBIDI_MAX_VALUES_INDEX];
 119     switch(which) {
 120     case UCHAR_BIDI_CLASS:
 121         return (max&UBIDI_CLASS_MASK);
 122     case UCHAR_JOINING_GROUP:
 123         return (max&UBIDI_MAX_JG_MASK)>>UBIDI_MAX_JG_SHIFT;
 124     case UCHAR_JOINING_TYPE:
 125         return (max&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT;
 126     case UCHAR_BIDI_PAIRED_BRACKET_TYPE:
 127         return (max&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT;
 128     default:
 129         return -1; /* undefined */
 130     }
 131 }
 132
 133 U_CAPI UCharDirection
 134 ubidi_getClass(UChar32 c) {
 135     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);
 136     return (UCharDirection)UBIDI_GET_CLASS(props);
 137 }
 138
 139 U_CFUNC UBool
 140 ubidi_isMirrored(UChar32 c) {
 141     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);
 142     return (UBool)UBIDI_GET_FLAG(props, UBIDI_IS_MIRRORED_SHIFT);
 143 }
 144
 145 static UChar32
 146 getMirror(UChar32 c, uint16_t props) {
 147     int32_t delta=UBIDI_GET_MIRROR_DELTA(props);
 148     if(delta!=UBIDI_ESC_MIRROR_DELTA) {
 149         return c+delta;
 150     } else {
 151         /* look for mirror code point in the mirrors[] table */
 152         const uint32_t *mirrors;
 153         uint32_t m;
 154         int32_t i, length;
 155         UChar32 c2;
 156
 157         mirrors=ubidi_props_singleton.mirrors;
 158         length=ubidi_props_singleton.indexes[UBIDI_IX_MIRROR_LENGTH];
 159
 160         /* linear search */
 161         for(i=0; i<length; ++i) {
 162             m=mirrors[i];
 163             c2=UBIDI_GET_MIRROR_CODE_POINT(m);
 164             if(c==c2) {
 165                 /* found c, return its mirror code point using the index in m */
 166                 return UBIDI_GET_MIRROR_CODE_POINT(mirrors[UBIDI_GET_MIRROR_INDEX(m)]);
 167             } else if(c<c2) {
 168                 break;
 169             }
 170         }
 171
 172         /* c not found, return it itself */
 173         return c;
 174     }
 175 }
 176
 177 U_CFUNC UChar32
 178 ubidi_getMirror(UChar32 c) {
 179     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);
 180     return getMirror(c, props);
 181 }
 182
 183 U_CFUNC UBool
 184 ubidi_isBidiControl(UChar32 c) {
 185     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);
 186     return (UBool)UBIDI_GET_FLAG(props, UBIDI_BIDI_CONTROL_SHIFT);
 187 }
 188
 189 U_CFUNC UBool
 190 ubidi_isJoinControl(UChar32 c) {
 191     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);
 192     return (UBool)UBIDI_GET_FLAG(props, UBIDI_JOIN_CONTROL_SHIFT);
 193 }
 194
 195 U_CFUNC UJoiningType
 196 ubidi_getJoiningType(UChar32 c) {
 197     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);
 198     return (UJoiningType)((props&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT);
 199 }
 200
 201 U_CFUNC UJoiningGroup
 202 ubidi_getJoiningGroup(UChar32 c) {
 203     UChar32 start, limit;
 204
 205     start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START];
 206     limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT];
 207     if(start<=c && c<limit) {
 208         return (UJoiningGroup)ubidi_props_singleton.jgArray[c-start];
 209     }
 210     start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START2];
 211     limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT2];
 212     if(start<=c && c<limit) {
 213         return (UJoiningGroup)ubidi_props_singleton.jgArray2[c-start];
 214     }
 215     return U_JG_NO_JOINING_GROUP;
 216 }
 217
 218 U_CFUNC UBidiPairedBracketType
 219 ubidi_getPairedBracketType(UChar32 c) {
 220     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);
 221     return (UBidiPairedBracketType)((props&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT);
 222 }
 223
 224 U_CFUNC UChar32
 225 ubidi_getPairedBracket(UChar32 c) {
 226     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);
 227     if((props&UBIDI_BPT_MASK)==0) {
 228         return c;
 229     } else {
 230         return getMirror(c, props);
 231     }
 232 }
 233
 234 /* public API (see uchar.h) ------------------------------------------------- */
 235
 236 U_CFUNC UCharDirection
 237 u_charDirection(UChar32 c) {
 238     return ubidi_getClass(c);
 239 }
 240
 241 U_CFUNC UBool
 242 u_isMirrored(UChar32 c) {
 243     return ubidi_isMirrored(c);
 244 }
 245
 246 U_CFUNC UChar32
 247 u_charMirror(UChar32 c) {
 248     return ubidi_getMirror(c);
 249 }
 250
 251 U_CAPI UChar32 U_EXPORT2
 252 u_getBidiPairedBracket(UChar32 c) {
 253     return ubidi_getPairedBracket(c);
 254 }