xrdp/uirdesktop/mppc.c

381 lines
8.5 KiB
C
Raw Normal View History

2006-06-21 07:58:16 +04:00
/* -*- c-basic-offset: 8 -*-
rdesktop: A Remote Desktop Protocol client.
Protocol services - RDP decompression
Copyright (C) Matthew Chapman 1999-2005
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
#include <stdio.h>
#include <string.h>
#include "rdesktop.h"
/* mppc decompression */
/* http://www.faqs.org/rfcs/rfc2118.html */
/* Contacts: */
/* hifn contact mentioned in the faq is */
/* Robert Friend rfriend at hifn dot com */
/* if you have questions regarding MPPC */
/* our contact is */
/* Guus Dhaeze GDhaeze at hifn dot com */
/* Licensing: */
/* decompression is alright as long as we */
/* don't compress data */
/* Algorithm: */
/* as the rfc states the algorithm seems to */
/* be LZ77 with a sliding buffer */
/* that is empty at init. */
/* the algorithm is called LZS and is */
/* patented for another couple of years. */
/* more information is available in */
/* http://www.ietf.org/ietf/IPR/hifn-ipr-draft-friend-tls-lzs-compression.txt */
RDPCOMP g_mppc_dict;
int
mppc_expand(uint8 * data, uint32 clen, uint8 ctype, uint32 * roff, uint32 * rlen)
{
int k, walker_len = 0, walker;
uint32 i = 0;
int next_offset, match_off;
int match_len;
int old_offset, match_bits;
BOOL big = ctype & RDP_MPPC_BIG ? True : False;
uint8 *dict = g_mppc_dict.hist;
if ((ctype & RDP_MPPC_COMPRESSED) == 0)
{
*roff = 0;
*rlen = clen;
return 0;
}
if ((ctype & RDP_MPPC_RESET) != 0)
{
g_mppc_dict.roff = 0;
}
if ((ctype & RDP_MPPC_FLUSH) != 0)
{
memset(dict, 0, RDP_MPPC_DICT_SIZE);
g_mppc_dict.roff = 0;
}
*roff = 0;
*rlen = 0;
walker = g_mppc_dict.roff;
next_offset = walker;
old_offset = next_offset;
*roff = old_offset;
if (clen == 0)
return 0;
clen += i;
do
{
if (walker_len == 0)
{
if (i >= clen)
break;
walker = data[i++] << 24;
walker_len = 8;
}
if (walker >= 0)
{
if (walker_len < 8)
{
if (i >= clen)
{
if (walker != 0)
return -1;
break;
}
walker |= (data[i++] & 0xff) << (24 - walker_len);
walker_len += 8;
}
if (next_offset >= RDP_MPPC_DICT_SIZE)
return -1;
dict[next_offset++] = (((uint32) walker) >> ((uint32) 24));
walker <<= 8;
walker_len -= 8;
continue;
}
walker <<= 1;
/* fetch next 8-bits */
if (--walker_len == 0)
{
if (i >= clen)
return -1;
walker = data[i++] << 24;
walker_len = 8;
}
/* literal decoding */
if (walker >= 0)
{
if (walker_len < 8)
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
walker_len += 8;
}
if (next_offset >= RDP_MPPC_DICT_SIZE)
return -1;
dict[next_offset++] = (uint8) (walker >> 24 | 0x80);
walker <<= 8;
walker_len -= 8;
continue;
}
/* decode offset */
/* length pair */
walker <<= 1;
if (--walker_len < (big ? 3 : 2))
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
walker_len += 8;
}
if (big)
{
/* offset decoding where offset len is:
-63: 11111 followed by the lower 6 bits of the value
64-319: 11110 followed by the lower 8 bits of the value ( value - 64 )
320-2367: 1110 followed by lower 11 bits of the value ( value - 320 )
2368-65535: 110 followed by lower 16 bits of the value ( value - 2368 )
*/
switch (((uint32) walker) >> ((uint32) 29))
{
case 7: /* - 63 */
for (; walker_len < 9; walker_len += 8)
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
}
walker <<= 3;
match_off = ((uint32) walker) >> ((uint32) 26);
walker <<= 6;
walker_len -= 9;
break;
case 6: /* 64 - 319 */
for (; walker_len < 11; walker_len += 8)
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
}
walker <<= 3;
match_off = (((uint32) walker) >> ((uint32) 24)) + 64;
walker <<= 8;
walker_len -= 11;
break;
case 5:
case 4: /* 320 - 2367 */
for (; walker_len < 13; walker_len += 8)
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
}
walker <<= 2;
match_off = (((uint32) walker) >> ((uint32) 21)) + 320;
walker <<= 11;
walker_len -= 13;
break;
default: /* 2368 - 65535 */
for (; walker_len < 17; walker_len += 8)
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
}
walker <<= 1;
match_off = (((uint32) walker) >> ((uint32) 16)) + 2368;
walker <<= 16;
walker_len -= 17;
break;
}
}
else
{
/* offset decoding where offset len is:
-63: 1111 followed by the lower 6 bits of the value
64-319: 1110 followed by the lower 8 bits of the value ( value - 64 )
320-8191: 110 followed by the lower 13 bits of the value ( value - 320 )
*/
switch (((uint32) walker) >> ((uint32) 30))
{
case 3: /* - 63 */
if (walker_len < 8)
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
walker_len += 8;
}
walker <<= 2;
match_off = ((uint32) walker) >> ((uint32) 26);
walker <<= 6;
walker_len -= 8;
break;
case 2: /* 64 - 319 */
for (; walker_len < 10; walker_len += 8)
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
}
walker <<= 2;
match_off = (((uint32) walker) >> ((uint32) 24)) + 64;
walker <<= 8;
walker_len -= 10;
break;
default: /* 320 - 8191 */
for (; walker_len < 14; walker_len += 8)
{
if (i >= clen)
return -1;
walker |= (data[i++] & 0xff) << (24 - walker_len);
}
match_off = (walker >> 18) + 320;
walker <<= 14;
walker_len -= 14;
break;
}
}
if (walker_len == 0)
{
if (i >= clen)
return -1;
walker = data[i++] << 24;
walker_len = 8;
}
/* decode length of match */
match_len = 0;
if (walker >= 0)
{ /* special case - length of 3 is in bit 0 */
match_len = 3;
walker <<= 1;
walker_len--;
}
else
{
/* this is how it works len of:
4-7: 10 followed by 2 bits of the value
8-15: 110 followed by 3 bits of the value
16-31: 1110 followed by 4 bits of the value
32-63: .... and so forth
64-127:
128-255:
256-511:
512-1023:
1024-2047:
2048-4095:
4096-8191:
i.e. 4097 is encoded as: 111111111110 000000000001
meaning 4096 + 1...
*/
match_bits = big ? 14 : 11; /* 11 or 14 bits of value at most */
do
{
walker <<= 1;
if (--walker_len == 0)
{
if (i >= clen)
return -1;
walker = data[i++] << 24;
walker_len = 8;
}
if (walker >= 0)
break;
if (--match_bits == 0)
{
return -1;
}
}
while (1);
match_len = (big ? 16 : 13) - match_bits;
walker <<= 1;
if (--walker_len < match_len)
{
for (; walker_len < match_len; walker_len += 8)
{
if (i >= clen)
{
return -1;
}
walker |= (data[i++] & 0xff) << (24 - walker_len);
}
}
match_bits = match_len;
match_len =
((walker >> (32 - match_bits)) & (~(-1 << match_bits))) | (1 <<
match_bits);
walker <<= match_bits;
walker_len -= match_bits;
}
if (next_offset + match_len >= RDP_MPPC_DICT_SIZE)
{
return -1;
}
/* memory areas can overlap - meaning we can't use memXXX functions */
k = (next_offset - match_off) & (big ? 65535 : 8191);
do
{
dict[next_offset++] = dict[k++];
}
while (--match_len != 0);
}
while (1);
/* store history offset */
g_mppc_dict.roff = next_offset;
*roff = old_offset;
*rlen = next_offset - old_offset;
return 0;
}