icuSources/test/letest/cmaps.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /***************************************************************************
   4 *
   5 *   Copyright (C) 1998-2014, International Business Machines
   6 *   Corporation and others.  All Rights Reserved.
   7 *
   8 ************************************************************************/
   9
  10 #include "layout/LETypes.h"
  11 #include "layout/LESwaps.h"
  12
  13 #include "sfnt.h"
  14 #include "cmaps.h"
  15 #include <stdio.h>
  16
  17 #define SWAPU16(code) ((LEUnicode16) SWAPW(code))
  18 #define SWAPU32(code) ((LEUnicode32) SWAPL(code))
  19
  20 //
  21 // Finds the high bit by binary searching
  22 // through the bits in value.
  23 //
  24 le_int8 highBit(le_uint32 value)
  25 {
  26     le_uint8 bit = 0;
  27
  28     if (value >= 1 << 16) {
  29         value >>= 16;
  30         bit += 16;
  31     }
  32
  33     if (value >= 1 << 8) {
  34         value >>= 8;
  35         bit += 8;
  36     }
  37
  38     if (value >= 1 << 4) {
  39         value >>= 4;
  40         bit += 4;
  41     }
  42
  43     if (value >= 1 << 2) {
  44         value >>= 2;
  45         bit += 2;
  46     }
  47
  48     if (value >= 1 << 1) {
  49         value >>= 1;
  50         bit += 1;
  51     }
  52
  53     return bit;
  54 }
  55
  56 CMAPMapper *CMAPMapper::createUnicodeMapper(const CMAPTable *cmap)
  57 {
  58     le_uint16 i;
  59     le_uint16 nSubtables = SWAPW(cmap->numberSubtables);
  60     const CMAPEncodingSubtable *subtable = NULL;
  61     le_bool found = FALSE;
  62     le_uint16 foundPlatformID = 0xFFFF;
  63     le_uint16 foundPlatformSpecificID = 0xFFFF;
  64     le_uint32 foundOffset = 0;
  65     le_uint16 foundTable = 0xFFFF;
  66     // first pass, look for MS table. (preferred?)
  67     for (i = 0; i < nSubtables && !found; i += 1) {
  68         const CMAPEncodingSubtableHeader *esh = &cmap->encodingSubtableHeaders[i];
  69
  70         le_uint16 platformID = SWAPW(esh->platformID);
  71         le_uint16 platformSpecificID = SWAPW(esh->platformSpecificID);
  72         if (platformID == 3) { // microsoft
  73           switch (platformSpecificID) {
  74             case 1: // Unicode BMP (UCS-2)
  75             case 10: // Unicode UCS-4
  76                 foundOffset = SWAPL(esh->encodingOffset);
  77                 foundPlatformID = platformID;
  78                 foundPlatformSpecificID = platformSpecificID;
  79                 found = TRUE;
  80                 foundTable = i;
  81                 break;
  82
  83                 //default:
  84               //              printf("%s:%d: microsoft (3) platform specific ID %d (wanted 1 or 10) for subtable %d/%d\n", __FILE__, __LINE__, (SWAPW(esh->platformSpecificID)), i, nSubtables);
  85             }
  86         } else {
  87           //printf("%s:%d: platform  ID %d (wanted 3, microsoft) for subtable %d/%d\n", __FILE__, __LINE__, (SWAPW(esh->platformID)), i, nSubtables);
  88         }
  89     }
  90
  91     // second pass, allow non MS table
  92     // first pass, look for MS table. (preferred?)
  93       for (i = 0; i < nSubtables && !found; i += 1) {
  94         const CMAPEncodingSubtableHeader *esh = &cmap->encodingSubtableHeaders[i];
  95         le_uint16 platformID = SWAPW(esh->platformID);
  96         le_uint16 platformSpecificID = SWAPW(esh->platformSpecificID);
  97         //printf("%s:%d: table %d/%d has platform:specific %d:%d\n", __FILE__, __LINE__, i, nSubtables, platformID, platformSpecificID);
  98         switch(platformID) {
  99         case 0: // Unicode platform
 100           switch(platformSpecificID) {
 101           case 0:
 102           case 1:
 103           case 2:
 104           case 3:
 105             foundOffset = SWAPL(esh->encodingOffset);
 106             foundPlatformID = platformID;
 107             foundPlatformSpecificID = platformSpecificID;
 108             foundTable = i;
 109             found = TRUE;
 110             break;
 111
 112           default: printf("Error: table %d (psid %d) is unknown. Skipping.\n", i, platformSpecificID); break;
 113           }
 114           break;
 115
 116           //default:
 117           //printf("Skipping platform id %d\n", platformID);
 118         }
 119     }
 120
 121
 122     if (found)
 123     {
 124       subtable = (const CMAPEncodingSubtable *) ((const char *) cmap + foundOffset);
 125       //printf("%s:%d: using subtable #%d/%d type %d:%d\n", __FILE__, __LINE__, foundTable, nSubtables, foundPlatformID, foundPlatformSpecificID);
 126       (void)foundPlatformID;   // Suppress unused variable compiler warnings.
 127       (void)foundTable;
 128       (void)foundPlatformSpecificID;
 129     } else {
 130       printf("%s:%d: could not find subtable.\n", __FILE__, __LINE__);
 131       return NULL;
 132     }
 133
 134     le_uint16 tableFormat = SWAPW(subtable->format);
 135     //printf("%s:%d: table format %d\n", __FILE__, __LINE__, tableFormat);
 136
 137     switch (tableFormat) {
 138     case 4:
 139         return new CMAPFormat4Mapper(cmap, (const CMAPFormat4Encoding *) subtable);
 140
 141     case 12:
 142     {
 143         const CMAPFormat12Encoding *encoding = (const CMAPFormat12Encoding *) subtable;
 144
 145         return new CMAPGroupMapper(cmap, encoding->groups, SWAPL(encoding->nGroups));
 146     }
 147
 148     default:
 149         break;
 150     }
 151
 152     printf("%s:%d: Unknown format %x.\n", __FILE__, __LINE__, (SWAPW(subtable->format)));
 153     return NULL;
 154 }
 155
 156 CMAPFormat4Mapper::CMAPFormat4Mapper(const CMAPTable *cmap, const CMAPFormat4Encoding *header)
 157     : CMAPMapper(cmap)
 158 {
 159     le_uint16 segCount = SWAPW(header->segCountX2) / 2;
 160
 161     fEntrySelector = SWAPW(header->entrySelector);
 162     fRangeShift = SWAPW(header->rangeShift) / 2;
 163     fEndCodes = &header->endCodes[0];
 164     fStartCodes = &header->endCodes[segCount + 1]; // + 1 for reservedPad...
 165     fIdDelta = &fStartCodes[segCount];
 166     fIdRangeOffset = &fIdDelta[segCount];
 167 }
 168
 169 LEGlyphID CMAPFormat4Mapper::unicodeToGlyph(LEUnicode32 unicode32) const
 170 {
 171     if (unicode32 >= 0x10000) {
 172         return 0;
 173     }
 174
 175     LEUnicode16 unicode = (LEUnicode16) unicode32;
 176     le_uint16 index = 0;
 177     le_uint16 probe = 1 << fEntrySelector;
 178     TTGlyphID result = 0;
 179
 180     if (SWAPU16(fStartCodes[fRangeShift]) <= unicode) {
 181         index = fRangeShift;
 182     }
 183
 184     while (probe > (1 << 0)) {
 185         probe >>= 1;
 186
 187         if (SWAPU16(fStartCodes[index + probe]) <= unicode) {
 188             index += probe;
 189         }
 190     }
 191
 192     if (unicode >= SWAPU16(fStartCodes[index]) && unicode <= SWAPU16(fEndCodes[index])) {
 193         if (fIdRangeOffset[index] == 0) {
 194             result = (TTGlyphID) unicode;
 195         } else {
 196             le_uint16 offset = unicode - SWAPU16(fStartCodes[index]);
 197             le_uint16 rangeOffset = SWAPW(fIdRangeOffset[index]);
 198             le_uint16 *glyphIndexTable = (le_uint16 *) ((char *) &fIdRangeOffset[index] + rangeOffset);
 199
 200             result = SWAPW(glyphIndexTable[offset]);
 201         }
 202
 203         result += SWAPW(fIdDelta[index]);
 204     } else {
 205         result = 0;
 206     }
 207
 208     return LE_SET_GLYPH(0, result);
 209 }
 210
 211 CMAPFormat4Mapper::~CMAPFormat4Mapper()
 212 {
 213     // parent destructor does it all
 214 }
 215
 216 CMAPGroupMapper::CMAPGroupMapper(const CMAPTable *cmap, const CMAPGroup *groups, le_uint32 nGroups)
 217     : CMAPMapper(cmap), fGroups(groups)
 218 {
 219     le_uint8 bit = highBit(nGroups);
 220     fPower = 1 << bit;
 221     fRangeOffset = nGroups - fPower;
 222 }
 223
 224 LEGlyphID CMAPGroupMapper::unicodeToGlyph(LEUnicode32 unicode32) const
 225 {
 226     le_int32 probe = fPower;
 227     le_int32 range = 0;
 228
 229     if (SWAPU32(fGroups[fRangeOffset].startCharCode) <= unicode32) {
 230         range = fRangeOffset;
 231     }
 232
 233     while (probe > (1 << 0)) {
 234         probe >>= 1;
 235
 236         if (SWAPU32(fGroups[range + probe].startCharCode) <= unicode32) {
 237             range += probe;
 238         }
 239     }
 240
 241     if (SWAPU32(fGroups[range].startCharCode) <= unicode32 && SWAPU32(fGroups[range].endCharCode) >= unicode32) {
 242         return (LEGlyphID) (SWAPU32(fGroups[range].startGlyphCode) + unicode32 - SWAPU32(fGroups[range].startCharCode));
 243     }
 244
 245     return 0;
 246 }
 247
 248 CMAPGroupMapper::~CMAPGroupMapper()
 249 {
 250     // parent destructor does it all
 251 }
 252