cmap.c

ttfdump源代码

/* cmap.c -- Load and Print the content of cmap table
 * Copyright (C) 1996 Li-Da Lho, All right reserved 
 * 
 * The structure of a cmap is really complex, it depends on which format the 
 * cmap is and loading a cmap is a kind of confusing thing. 
 * The structure of CMAP:
 *   A cmap begins with a general description of the table
 *     USHORT version indicates the version of this cmap.
 *     USHORT numEncodings is the number of encodings in the cmap table.
 *   Then follow 'numEncodings' of encoding tables, each of which contains
 *   the following informations for each subtable. The subtable contains the 
 *   true data to map character code to glyph index
 *     USHORT PlatformID shows which platform the subtable is to be used 
 *            (e.g. Mac, PC)
 *     USHORT Platfrom specofic EncodingID shows what kind of encoding scheme
 *            this table uses (e.g. Big5, iso8859, Unicode etc.)
 *     ULONG  offset is the offset from the beginning of the whole cmap table
 *            to the beginning of the corresponding subtable
 * -----------------------         ---         ---
 * | USHORT version      |          ^           ^
 * -----------------------          |           |
 * | USHORT numEncodings |          |           |
 * -----------------------          |           |
 * | Encoding Table 1    |          |           |
 * |  USHORT PlatformID  |          |           |
 * |  USHORT EncodingID  |          |           |
 * |  ULONG  offset 1    |------    |           |
 * -----------------------     |    |  offset 1 | offset 2
 * | Encoding Table 2    |     |    |           |
 * |  USHORT PlatformID  |     |    |           |
 * |  USHORT EncodingID  |     |    |           |
 * |  ULONG  offset 2    |---  |    v           |
 * -----------------------  |  |   ---          |
 * | Subtable 1          |<-----                |
 * | with format 0       |  |                   v
 * -----------------------  |                  ---
 * | Subtable 2          |<--                
 * | with format 2       |
 * ----------------------|
 *
 * Problem with Character Code vs. Byte Endianess.
 * 
 */
#include <stdio.h>
#include <stdlib.h>
#include "config.h"
#include "ttf.h"
#include "ttfutil.h"

#ifdef MEMCHECK
#include <dmalloc.h>
#endif

/* 	$Id: cmap.c,v 1.1.1.1 1998/06/05 07:47:52 robert Exp $	 */

#ifndef lint
static char vcid[] = "$Id: cmap.c,v 1.1.1.1 1998/06/05 07:47:52 robert Exp $";
#endif /* lint */

static CMAPPtr ttfAllocCMAP(TTFontPtr font);
static void ttfLoadCMAP(FILE *fp,CMAPPtr cmap,ULONG offset);

static void ttfLoadEncodingTable(FILE *fp,SubTablePtr subTable,ULONG offset);
static SubTablePtr ttfAllocSubTable(CMAPPtr cmap);
static void ttfLoadSubTable(FILE *fp,SubTablePtr subTable,ULONG offset);
static void ttfPrintSubTable(FILE* fp,SubTablePtr ptable);
static void ttfFreeSubTable(SubTablePtr ptable);

static CMAP0 * ttfAllocCMAP0(SubTablePtr subTable);
static void ttfLoadCMAP0(FILE *fp,SubTablePtr subTable,ULONG offset);
static void ttfPrintCMAP0(FILE *fp,SubTablePtr subTable);
static USHORT ttfLookUpCMAP0(SubTablePtr subTable,USHORT cc);
static void ttfFreeCMAP0(SubTablePtr subTable);

static void ttfLoadCMAP2(FILE *fp,SubTablePtr subTable,ULONG offset);
static void ttfPrintCMAP2(FILE *fp,SubTablePtr subTable);
static USHORT ttfLookUpCMAP2(SubTablePtr subTable,USHORT cc);
static void ttfFreeCMAP2(SubTablePtr subTable);

static void ttfLoadCMAP4(FILE *fp,SubTablePtr subTable,ULONG offset);
static void ttfPrintCMAP4(FILE *fp,SubTablePtr subTable);
static USHORT ttfLookUpCMAP4(SubTablePtr subTable,USHORT cc);
static void ttfFreeCMAP4(SubTablePtr subTable);

static void ttfLoadCMAP6(FILE *fp,SubTablePtr subTable,ULONG offset);
static void ttfPrintCMAP6(FILE *fp, SubTablePtr sbuTable);
static USHORT ttfLookUpCMAP6(SubTablePtr subTable,USHORT cc);
static void ttfFreeCMAP6(SubTablePtr subTable);

void ttfInitCMAP(TTFontPtr font)
{
    ULONG tag = 'c' | 'm' << 8 | 'a' << 16 | 'p' << 24;
    TableDirPtr ptd;

    if ((ptd = ttfLookUpTableDir(tag,font)) != NULL)
	{
	    font->cmap = ttfAllocCMAP(font);
	    ttfLoadCMAP(font->fp,font->cmap,ptd->offset);
	}
}
static CMAPPtr ttfAllocCMAP(TTFontPtr font)
{
    CMAPPtr cmap;
    if ((cmap = (CMAPPtr) calloc(1,sizeof(CMAP))) == NULL)
	{
	    ttfError("Out Of Memory in  __FILE_ : __LINE__ \n");
	    return NULL;
	}
    return cmap;
}
static void ttfLoadCMAP(FILE *fp,CMAPPtr cmap,ULONG offset)
{
    USHORT i,n;
    ULONG posEnc;            /* beginning of the Encoding Table */
    ULONG baseSub = offset;  /* base of SubTable offset */

    if (fseek(fp,offset,SEEK_SET) !=0)
	ttfError("Fseek Failed in ttfLOADCMAP \n");	

    cmap->version = ttfGetUSHORT(fp);
    cmap->numberOfEncodings = n = ttfGetUSHORT(fp);
    cmap->subTables = ttfAllocSubTable(cmap);

    posEnc = baseSub + sizeof(USHORT)*2; /* step over the beginning of encoding
					  * table */
    
    /* for each encoding scheme, load the encoding table (EncodingTable) and 
     * the real cmap data (SubTable) */
    for (i=0;i<n;i++,posEnc += 8)
	{   /* 8 == ushort*2 + ulong*1 */
	    ttfLoadEncodingTable(fp,cmap->subTables+i,posEnc);
	    ttfLoadSubTable(fp,cmap->subTables+i,baseSub);
	}
}
void ttfPrintCMAP(FILE *fp,CMAPPtr cmap)
{
    USHORT i;

    fprintf(fp,"'cmap' Table - Character to Glyph Index Mapping Table\n");
    fprintf(fp,"-----------------------------------------------------\n");
    fprintf(fp,"\t 'cmap' version: %d\n",cmap->version);
    fprintf(fp,"\t number of encoding: %d\n\n",cmap->numberOfEncodings);

    for (i=0;i<cmap->numberOfEncodings;i++)
	{
	    fprintf(fp,"Subtable %3d.\t",i);
	    ttfPrintSubTable(fp,cmap->subTables+i);
	    fprintf(fp,"\n");
	}
}
USHORT ttfLookUpCMAP(SubTablePtr subTable,USHORT cc)
{
    USHORT idx,format = subTable->format;

    switch (format)
	{
	case 0:
	    idx = ttfLookUpCMAP0(subTable,cc);
	    break;
	case 2:
	    idx = ttfLookUpCMAP2(subTable,cc);
	    break;
	case 4:
	    idx = ttfLookUpCMAP4(subTable,cc);
	    break;
	case 6:
	    idx = ttfLookUpCMAP6(subTable,cc);
	    break;
	default:
	    ttfError("Unrecognized CMAP format\n");
	    return 0;
	}
    return idx;
}
SubTablePtr ttfSelectCMAP(CMAPPtr cmap, USHORT PlatformID, USHORT EncodingID)
{
    USHORT i;
    
    for (i=0;i<cmap->numberOfEncodings;i++)
	{
	    if ((cmap->subTables+i)->PlatformID == PlatformID &&
		(cmap->subTables+i)->EncodingID == EncodingID)
		return cmap->subTables+i;
	}
    return NULL;
}
void ttfFreeCMAP(CMAPPtr cmap)
{
    USHORT i;
    
    for (i=0;i<cmap->numberOfEncodings;i++)
	ttfFreeSubTable(cmap->subTables+i);

    free(cmap->subTables);
    free(cmap);
}

/* actually, EncodingTable is a part of SubTable */
static void ttfLoadEncodingTable(FILE *fp,SubTablePtr subTable,ULONG offset)
{
    if (fseek(fp,offset,SEEK_SET) !=0)
	ttfError("Fseek Failed in ttfLoadEncodingTable \n");
    
    subTable->PlatformID = ttfGetUSHORT(fp);
    subTable->EncodingID = ttfGetUSHORT(fp);
    subTable->offset = ttfGetULONG(fp);
}
static SubTablePtr ttfAllocSubTable(CMAPPtr cmap)
{
    SubTablePtr subTable;
    
    if ((subTable = (SubTablePtr) calloc(cmap->numberOfEncodings,sizeof(SubTable))) == NULL)
	{
	    ttfError("Out Of Memory in  __FILE__ : __LINE__ \n");
	    return NULL;
	}
    return subTable;
}
/* should this one be static ? */
static void ttfLoadSubTable(FILE *fp,SubTablePtr subTable,ULONG base)
{
    ULONG pos;
    USHORT format;

    /* seek to the actuall position for this subtable
     * base: beginning of cmap
     * offset: offset field of each encoding table */
    pos =  base  +  subTable->offset;
    if (fseek(fp,pos,SEEK_SET) !=0)
	ttfError("Fseek Failed in ttfLoadSubTable\n");

    subTable->format = format = ttfGetUSHORT(fp); 
    subTable->length = ttfGetUSHORT(fp);
    subTable->version = ttfGetUSHORT(fp);
   
    pos += 6;/* step over format independent data,USHORT*3 */

    switch(format)
	{
	case 0:
	    ttfLoadCMAP0(fp,subTable,pos);
	    break;
	case 2:
	    ttfLoadCMAP2(fp,subTable,pos);
	    break;
	case 4:
	    ttfLoadCMAP4(fp,subTable,pos);
	    break;
	case 6:
	    ttfLoadCMAP6(fp,subTable,pos);
	    break;
	default:
	    ttfError("Unrecognized CMAP format\n");
	}
}
static void ttfPrintSubTable(FILE* fp,SubTablePtr ptable)
{
    USHORT format = ptable->format;
    
    /* print encoding table */
    fprintf(fp,    " PlatformID: %2d\n",ptable->PlatformID);
    fprintf(fp,"\t\t EcodingID:  %2d\n",ptable->EncodingID);
    fprintf(fp,"\t\t 'cmap' Offset: 0x%08x\n",ptable->offset);
    
    /* print SubTable part */
    fprintf(fp,"\t\t Length:  %6d\n",ptable->length);
    fprintf(fp,"\t\t Version: %6d\n",ptable->version);

    switch(format)
	{
	case 0:
	    fprintf(fp,"\t\t Format 0 - Byte encoding table\n");
	    ttfPrintCMAP0(fp,ptable);
	    break;
	case 2:
	    fprintf(fp,"\t\t Format 2 - High-byte mapping through table\n");
	    ttfPrintCMAP2(fp,ptable);
	    break;
	case 4:
	    fprintf(fp,"\t\t Format 4 - Segment mapping to delta values\n");
	    ttfPrintCMAP4(fp,ptable);
	    break;
	case 6:
	    fprintf(fp,"\t\t Format 6 - Trimmed table mapping\n");
	    ttfPrintCMAP6(fp,ptable);
	    break;
	default:
	    ttfError("Unrecognized CMAP format\n");
	}
}
static void ttfFreeSubTable(SubTablePtr ptable)
{
    USHORT format = ptable->format;
    
    switch(format)
	{
	case 0:
	    ttfFreeCMAP0(ptable);
	    break;
	case 2:
	    ttfFreeCMAP2(ptable);
	    break;
	case 4:
	    ttfFreeCMAP4(ptable);
 	    break;
	case 6:
	    ttfFreeCMAP6(ptable);
	    break;
	}
}

static CMAP0 * ttfAllocCMAP0(SubTablePtr subTable)
{
    CMAP0 * cmap0;
  
    if ((cmap0 = (CMAP0 *) calloc(1,sizeof(CMAP0))) == NULL)
	{
	    ttfError("Out Of Memory in  __FILE__ : __LINE__ \n");
	    return NULL;
	}
    return cmap0;
}
static void ttfLoadCMAP0(FILE *fp,SubTablePtr subTable,ULONG offset)
{
    BYTE * array;
    
    subTable->map.cmap0 = ttfAllocCMAP0(subTable);
    array = subTable->map.cmap0->glyphIndexArray; 

    if (fseek(fp,offset,SEEK_SET) != 0)
	ttfError("Fseek Failed in ttfLoadCMAP0 \n");

    /* Attention: we get lots of bytes at once as a work around of the
     * usual ttfGet*, this cause byte sex trouble as will be seen
     * in the fellowing procedures */
    if (fread(array,sizeof(BYTE),256,fp) != 256)
	ttfError("Error when getting glyphIndexArray \n");
}
static void ttfPrintCMAP0(FILE *fp,SubTablePtr subTable)
{
    USHORT index;
    int i;
    
    for (i=0;i<256;i++)
	{
	    index = ttfLookUpCMAP(subTable,i);
	    fprintf(fp,"\t\t Char %3d -> Index %4d \n",i,index);
	}
}
static USHORT ttfLookUpCMAP0(SubTablePtr subTable,USHORT cc)
{
    return subTable->map.cmap0->glyphIndexArray[cc & 0x00ff];
}
static void ttfFreeCMAP0(SubTablePtr subTable)
{
    free(subTable->map.cmap0);
}

static void ttfLoadCMAP2(FILE *fp,SubTablePtr subTable,ULONG offset)
{
    USHORT * array,i,n = 0;
    USHORT numGlyphId;
    SubHeaderPtr header;

    if (fseek(fp,offset,SEEK_SET) != 0)
	ttfError("Fseek Failed in ttfLoadCMAP2 \n");

    subTable->map.cmap2 = (CMAP2 *) calloc(1,sizeof(CMAP2));
    array = subTable->map.cmap2->subHeaderKeys;
    
    if (fread(array,sizeof(USHORT),256,fp) != 256)
	ttfError("Error when getting subHeaderKeys \n");
#ifndef WORDS_BIGENDIAN
    TwoByteSwap((unsigned char *) array,256*sizeof(USHORT));
#endif