icuSources/test/perf/leperf/cmaps.cpp

   1 /**************************************************************************
   2 *
   3 *   © 2016 and later: Unicode, Inc. and others.
   4 *   License & terms of use: http://www.unicode.org/copyright.html#License
   5 *
   6 ***************************************************************************
   7 ***************************************************************************
   8 *
   9 *   Copyright (C) 1998-2013, International Business Machines
  10 *   Corporation and others.  All Rights Reserved.
  11 *
  12 ************************************************************************/
  13
  14 #include "layout/LETypes.h"
  15 #include "layout/LESwaps.h"
  16
  17 #include "sfnt.h"
  18 #include "cmaps.h"
  19 #include <stdio.h>
  20
  21 #define SWAPU16(code) ((LEUnicode16) SWAPW(code))
  22 #define SWAPU32(code) ((LEUnicode32) SWAPL(code))
  23
  24 //
  25 // Finds the high bit by binary searching
  26 // through the bits in value.
  27 //
  28 le_int8 highBit(le_uint32 value)
  29 {
  30     le_uint8 bit = 0;
  31
  32     if (value >= 1 << 16) {
  33         value >>= 16;
  34         bit += 16;
  35     }
  36
  37     if (value >= 1 << 8) {
  38         value >>= 8;
  39         bit += 8;
  40     }
  41
  42     if (value >= 1 << 4) {
  43         value >>= 4;
  44         bit += 4;
  45     }
  46
  47     if (value >= 1 << 2) {
  48         value >>= 2;
  49         bit += 2;
  50     }
  51
  52     if (value >= 1 << 1) {
  53         value >>= 1;
  54         bit += 1;
  55     }
  56
  57     return bit;
  58 }
  59
  60 CMAPMapper *CMAPMapper::createUnicodeMapper(const CMAPTable *cmap)
  61 {
  62     le_uint16 i;
  63     le_uint16 nSubtables = SWAPW(cmap->numberSubtables);
  64     const CMAPEncodingSubtable *subtable = NULL;
  65     le_bool found = FALSE;
  66     le_uint16 foundPlatformID = 0xFFFF;
  67     le_uint16 foundPlatformSpecificID = 0xFFFF;
  68     le_uint32 foundOffset = 0;
  69     le_uint16 foundTable = 0xFFFF;
  70     // first pass, look for MS table. (preferred?)
  71     for (i = 0; i < nSubtables && !found; i += 1) {
  72         const CMAPEncodingSubtableHeader *esh = &cmap->encodingSubtableHeaders[i];
  73
  74         le_uint16 platformID = SWAPW(esh->platformID);
  75         le_uint16 platformSpecificID = SWAPW(esh->platformSpecificID);
  76         if (platformID == 3) { // microsoft
  77           switch (platformSpecificID) {
  78             case 1: // Unicode BMP (UCS-2)
  79             case 10: // Unicode UCS-4
  80                 foundOffset = SWAPL(esh->encodingOffset);
  81                 foundPlatformID = platformID;
  82                 foundPlatformSpecificID = platformSpecificID;
  83                 found = TRUE;
  84                 foundTable = i;
  85                 break;
  86
  87                 //default:
  88               //              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);
  89             }
  90         } else {
  91           //printf("%s:%d: platform  ID %d (wanted 3, microsoft) for subtable %d/%d\n", __FILE__, __LINE__, (SWAPW(esh->platformID)), i, nSubtables);
  92         }
  93     }
  94
  95     // second pass, allow non MS table
  96     // first pass, look for MS table. (preferred?)
  97       for (i = 0; i < nSubtables && !found; i += 1) {
  98         const CMAPEncodingSubtableHeader *esh = &cmap->encodingSubtableHeaders[i];
  99         le_uint16 platformID = SWAPW(esh->platformID);
 100         le_uint16 platformSpecificID = SWAPW(esh->platformSpecificID);
 101         //printf("%s:%d: table %d/%d has platform:specific %d:%d\n", __FILE__, __LINE__, i, nSubtables, platformID, platformSpecificID);
 102         switch(platformID) {
 103         case 0: // Unicode platform
 104           switch(platformSpecificID) {
 105           case 0:
 106           case 1:
 107           case 2:
 108           case 3:
 109             foundOffset = SWAPL(esh->encodingOffset);
 110             foundPlatformID = platformID;
 111             foundPlatformSpecificID = platformSpecificID;
 112             foundTable = i;
 113             found = TRUE;
 114             break;
 115
 116           default: printf("Error: table %d (psid %d) is unknown. Skipping.\n", i, platformSpecificID); break;
 117           }
 118           break;
 119
 120           //default:
 121           //printf("Skipping platform id %d\n", platformID);
 122         }
 123     }
 124
 125
 126     if (found)
 127     {
 128       subtable = (const CMAPEncodingSubtable *) ((const char *) cmap + foundOffset);
 129       //printf("%s:%d: using subtable #%d/%d type %d:%d\n", __FILE__, __LINE__, foundTable, nSubtables, foundPlatformID, foundPlatformSpecificID);
 130     } else {
 131       printf("%s:%d: could not find subtable.\n", __FILE__, __LINE__);
 132       return NULL;
 133     }
 134
 135     le_uint16 tableFormat = SWAPW(subtable->format);
 136     //printf("%s:%d: table format %d\n", __FILE__, __LINE__, tableFormat);
 137
 138     switch (tableFormat) {
 139     case 4:
 140         return new CMAPFormat4Mapper(cmap, (const CMAPFormat4Encoding *) subtable);
 141
 142     case 12:
 143     {
 144         const CMAPFormat12Encoding *encoding = (const CMAPFormat12Encoding *) subtable;
 145
 146         return new CMAPGroupMapper(cmap, encoding->groups, SWAPL(encoding->nGroups));
 147     }
 148
 149     default:
 150         break;
 151     }
 152
 153     printf("%s:%d: Unknown format %x.\n", __FILE__, __LINE__, (SWAPW(subtable->format)));
 154     return NULL;
 155 }
 156
 157 CMAPFormat4Mapper::CMAPFormat4Mapper(const CMAPTable *cmap, const CMAPFormat4Encoding *header)
 158     : CMAPMapper(cmap)
 159 {
 160     le_uint16 segCount = SWAPW(header->segCountX2) / 2;
 161
 162     fEntrySelector = SWAPW(header->entrySelector);
 163     fRangeShift = SWAPW(header->rangeShift) / 2;
 164     fEndCodes = &header->endCodes[0];
 165     fStartCodes = &header->endCodes[segCount + 1]; // + 1 for reservedPad...
 166     fIdDelta = &fStartCodes[segCount];
 167     fIdRangeOffset = &fIdDelta[segCount];
 168 }
 169
 170 LEGlyphID CMAPFormat4Mapper::unicodeToGlyph(LEUnicode32 unicode32) const
 171 {
 172     if (unicode32 >= 0x10000) {
 173         return 0;
 174     }
 175
 176     LEUnicode16 unicode = (LEUnicode16) unicode32;
 177     le_uint16 index = 0;
 178     le_uint16 probe = 1 << fEntrySelector;
 179     TTGlyphID result = 0;
 180
 181     if (SWAPU16(fStartCodes[fRangeShift]) <= unicode) {
 182         index = fRangeShift;
 183     }
 184
 185     while (probe > (1 << 0)) {
 186         probe >>= 1;
 187
 188         if (SWAPU16(fStartCodes[index + probe]) <= unicode) {
 189             index += probe;
 190         }
 191     }
 192
 193     if (unicode >= SWAPU16(fStartCodes[index]) && unicode <= SWAPU16(fEndCodes[index])) {
 194         if (fIdRangeOffset[index] == 0) {
 195             result = (TTGlyphID) unicode;
 196         } else {
 197             le_uint16 offset = unicode - SWAPU16(fStartCodes[index]);
 198             le_uint16 rangeOffset = SWAPW(fIdRangeOffset[index]);
 199             le_uint16 *glyphIndexTable = (le_uint16 *) ((char *) &fIdRangeOffset[index] + rangeOffset);
 200
 201             result = SWAPW(glyphIndexTable[offset]);
 202         }
 203
 204         result += SWAPW(fIdDelta[index]);
 205     } else {
 206         result = 0;
 207     }
 208
 209     return LE_SET_GLYPH(0, result);
 210 }
 211
 212 CMAPFormat4Mapper::~CMAPFormat4Mapper()
 213 {
 214     // parent destructor does it all
 215 }
 216
 217 CMAPGroupMapper::CMAPGroupMapper(const CMAPTable *cmap, const CMAPGroup *groups, le_uint32 nGroups)
 218     : CMAPMapper(cmap), fGroups(groups)
 219 {
 220     le_uint8 bit = highBit(nGroups);
 221     fPower = 1 << bit;
 222     fRangeOffset = nGroups - fPower;
 223 }
 224
 225 LEGlyphID CMAPGroupMapper::unicodeToGlyph(LEUnicode32 unicode32) const
 226 {
 227     le_int32 probe = fPower;
 228     le_int32 range = 0;
 229
 230     if (SWAPU32(fGroups[fRangeOffset].startCharCode) <= unicode32) {
 231         range = fRangeOffset;
 232     }
 233
 234     while (probe > (1 << 0)) {
 235         probe >>= 1;
 236
 237         if (SWAPU32(fGroups[range + probe].startCharCode) <= unicode32) {
 238             range += probe;
 239         }
 240     }
 241
 242     if (SWAPU32(fGroups[range].startCharCode) <= unicode32 && SWAPU32(fGroups[range].endCharCode) >= unicode32) {
 243         return (LEGlyphID) (SWAPU32(fGroups[range].startGlyphCode) + unicode32 - SWAPU32(fGroups[range].startCharCode));
 244     }
 245
 246     return 0;
 247 }
 248
 249 CMAPGroupMapper::~CMAPGroupMapper()
 250 {
 251     // parent destructor does it all
 252 }
 253