Source code
Revision control
Copy as Markdown
Other Tools
/****************************************************************************
*
* ftzopen.c
*
* FreeType support for .Z compressed files.
*
* This optional component relies on NetBSD's zopen(). It should mainly
* be used to parse compressed PCF fonts, as found with many X11 server
* distributions.
*
* Copyright (C) 2005-2024 by
* David Turner.
*
* This file is part of the FreeType project, and may only be used,
* modified, and distributed under the terms of the FreeType project
* license, LICENSE.TXT. By continuing to use, modify, or distribute
* this file you indicate that you have read the license and
* understand and accept it fully.
*
*/
#include "ftzopen.h"
#include <freetype/internal/ftmemory.h>
#include <freetype/internal/ftstream.h>
#include <freetype/internal/ftdebug.h>
static int
ft_lzwstate_refill( FT_LzwState state )
{
FT_ULong count;
if ( state->in_eof )
return -1;
count = FT_Stream_TryRead( state->source,
state->buf_tab,
state->num_bits ); /* WHY? */
state->buf_size = (FT_UInt)count;
state->buf_total += count;
state->in_eof = FT_BOOL( count < state->num_bits );
state->buf_offset = 0;
state->buf_size <<= 3;
if ( state->buf_size > state->num_bits )
state->buf_size -= state->num_bits - 1;
else
return -1; /* not enough data */
if ( count == 0 ) /* end of file */
return -1;
return 0;
}
static FT_Int32
ft_lzwstate_get_code( FT_LzwState state )
{
FT_UInt num_bits = state->num_bits;
FT_UInt offset = state->buf_offset;
FT_Byte* p;
FT_Int result;
if ( state->buf_clear ||
offset >= state->buf_size ||
state->free_ent >= state->free_bits )
{
if ( state->free_ent >= state->free_bits )
{
state->num_bits = ++num_bits;
if ( num_bits > LZW_MAX_BITS )
return -1;
state->free_bits = state->num_bits < state->max_bits
? (FT_UInt)( ( 1UL << num_bits ) - 256 )
: state->max_free + 1;
}
if ( state->buf_clear )
{
state->num_bits = num_bits = LZW_INIT_BITS;
state->free_bits = (FT_UInt)( ( 1UL << num_bits ) - 256 );
state->buf_clear = 0;
}
if ( ft_lzwstate_refill( state ) < 0 )
return -1;
offset = 0;
}
state->buf_offset = offset + num_bits;
p = &state->buf_tab[offset >> 3];
offset &= 7;
result = *p++ >> offset;
offset = 8 - offset;
num_bits -= offset;
if ( num_bits >= 8 )
{
result |= *p++ << offset;
offset += 8;
num_bits -= 8;
}
if ( num_bits > 0 )
result |= ( *p & LZW_MASK( num_bits ) ) << offset;
return result;
}
/* grow the character stack */
static int
ft_lzwstate_stack_grow( FT_LzwState state )
{
if ( state->stack_top >= state->stack_size )
{
FT_Memory memory = state->memory;
FT_Error error;
FT_Offset old_size = state->stack_size;
FT_Offset new_size = old_size;
new_size = new_size + ( new_size >> 1 ) + 4;
/* if relocating to heap */
if ( state->stack == state->stack_0 )
{
state->stack = NULL;
old_size = 0;
}
/* requirement of the character stack larger than 1<<LZW_MAX_BITS */
/* implies bug in the decompression code */
if ( new_size > ( 1 << LZW_MAX_BITS ) )
{
new_size = 1 << LZW_MAX_BITS;
if ( new_size == old_size )
return -1;
}
if ( FT_QREALLOC( state->stack, old_size, new_size ) )
return -1;
/* if relocating to heap */
if ( old_size == 0 )
FT_MEM_COPY( state->stack, state->stack_0, FT_LZW_DEFAULT_STACK_SIZE );
state->stack_size = new_size;
}
return 0;
}
/* grow the prefix/suffix arrays */
static int
ft_lzwstate_prefix_grow( FT_LzwState state )
{
FT_UInt old_size = state->prefix_size;
FT_UInt new_size = old_size;
FT_Memory memory = state->memory;
FT_Error error;
if ( new_size == 0 ) /* first allocation -> 9 bits */
new_size = 512;
else
new_size += new_size >> 2; /* don't grow too fast */
/*
* Note that the `suffix' array is located in the same memory block
* pointed to by `prefix'.
*
* I know that sizeof(FT_Byte) == 1 by definition, but it is clearer
* to write it literally.
*
*/
if ( FT_REALLOC_MULT( state->prefix, old_size, new_size,
sizeof ( FT_UShort ) + sizeof ( FT_Byte ) ) )
return -1;
/* now adjust `suffix' and move the data accordingly */
state->suffix = (FT_Byte*)( state->prefix + new_size );
FT_MEM_MOVE( state->suffix,
state->prefix + old_size,
old_size * sizeof ( FT_Byte ) );
state->prefix_size = new_size;
return 0;
}
FT_LOCAL_DEF( void )
ft_lzwstate_reset( FT_LzwState state )
{
state->in_eof = 0;
state->buf_offset = 0;
state->buf_size = 0;
state->buf_clear = 0;
state->buf_total = 0;
state->stack_top = 0;
state->num_bits = LZW_INIT_BITS;
state->phase = FT_LZW_PHASE_START;
}
FT_LOCAL_DEF( void )
ft_lzwstate_init( FT_LzwState state,
FT_Stream source )
{
FT_ZERO( state );
state->source = source;
state->memory = source->memory;
state->prefix = NULL;
state->suffix = NULL;
state->prefix_size = 0;
state->stack = state->stack_0;
state->stack_size = sizeof ( state->stack_0 );
ft_lzwstate_reset( state );
}
FT_LOCAL_DEF( void )
ft_lzwstate_done( FT_LzwState state )
{
FT_Memory memory = state->memory;
ft_lzwstate_reset( state );
if ( state->stack != state->stack_0 )
FT_FREE( state->stack );
FT_FREE( state->prefix );
state->suffix = NULL;
FT_ZERO( state );
}
#define FTLZW_STACK_PUSH( c ) \
FT_BEGIN_STMNT \
if ( state->stack_top >= state->stack_size && \
ft_lzwstate_stack_grow( state ) < 0 ) \
goto Eof; \
\
state->stack[state->stack_top++] = (FT_Byte)(c); \
FT_END_STMNT
FT_LOCAL_DEF( FT_ULong )
ft_lzwstate_io( FT_LzwState state,
FT_Byte* buffer,
FT_ULong out_size )
{
FT_ULong result = 0;
FT_UInt old_char = state->old_char;
FT_UInt old_code = state->old_code;
FT_UInt in_code = state->in_code;
if ( out_size == 0 )
goto Exit;
switch ( state->phase )
{
case FT_LZW_PHASE_START:
{
FT_Byte max_bits;
FT_Int32 c;
/* skip magic bytes, and read max_bits + block_flag */
if ( FT_Stream_Seek( state->source, 2 ) != 0 ||
FT_Stream_TryRead( state->source, &max_bits, 1 ) != 1 )
goto Eof;
state->max_bits = max_bits & LZW_BIT_MASK;
state->block_mode = max_bits & LZW_BLOCK_MASK;
state->max_free = (FT_UInt)( ( 1UL << state->max_bits ) - 256 );
if ( state->max_bits > LZW_MAX_BITS )
goto Eof;
state->num_bits = LZW_INIT_BITS;
state->free_ent = ( state->block_mode ? LZW_FIRST
: LZW_CLEAR ) - 256;
in_code = 0;
state->free_bits = state->num_bits < state->max_bits
? (FT_UInt)( ( 1UL << state->num_bits ) - 256 )
: state->max_free + 1;
c = ft_lzwstate_get_code( state );
if ( c < 0 || c > 255 )
goto Eof;
old_code = old_char = (FT_UInt)c;
if ( buffer )
buffer[result] = (FT_Byte)old_char;
if ( ++result >= out_size )
goto Exit;
state->phase = FT_LZW_PHASE_CODE;
}
FALL_THROUGH;
case FT_LZW_PHASE_CODE:
{
FT_Int32 c;
FT_UInt code;
NextCode:
c = ft_lzwstate_get_code( state );
if ( c < 0 )
goto Eof;
code = (FT_UInt)c;
if ( code == LZW_CLEAR && state->block_mode )
{
/* why not LZW_FIRST-256 ? */
state->free_ent = ( LZW_FIRST - 1 ) - 256;
state->buf_clear = 1;
/* not quite right, but at least more predictable */
old_code = 0;
old_char = 0;
goto NextCode;
}
in_code = code; /* save code for later */
if ( code >= 256U )
{
/* special case for KwKwKwK */
if ( code - 256U >= state->free_ent )
{
/* corrupted LZW stream */
if ( code - 256U > state->free_ent )
goto Eof;
FTLZW_STACK_PUSH( old_char );
code = old_code;
}
while ( code >= 256U )
{
if ( !state->prefix )
goto Eof;
FTLZW_STACK_PUSH( state->suffix[code - 256] );
code = state->prefix[code - 256];
}
}
old_char = code;
FTLZW_STACK_PUSH( old_char );
state->phase = FT_LZW_PHASE_STACK;
}
FALL_THROUGH;
case FT_LZW_PHASE_STACK:
{
while ( state->stack_top > 0 )
{
state->stack_top--;
if ( buffer )
buffer[result] = state->stack[state->stack_top];
if ( ++result == out_size )
goto Exit;
}
/* now create new entry */
if ( state->free_ent < state->max_free )
{
if ( state->free_ent >= state->prefix_size &&
ft_lzwstate_prefix_grow( state ) < 0 )
goto Eof;
FT_ASSERT( state->free_ent < state->prefix_size );
state->prefix[state->free_ent] = (FT_UShort)old_code;
state->suffix[state->free_ent] = (FT_Byte) old_char;
state->free_ent += 1;
}
old_code = in_code;
state->phase = FT_LZW_PHASE_CODE;
goto NextCode;
}
default: /* state == EOF */
;
}
Exit:
state->old_code = old_code;
state->old_char = old_char;
state->in_code = in_code;
return result;
Eof:
state->phase = FT_LZW_PHASE_EOF;
goto Exit;
}
/* END */