Intravision/reactos
1
0
Fork 0
mirror of https://github.com/reactos/reactos.git synced 2024-12-28 01:55:19 +00:00
Code Issues Projects Releases Packages Wiki Activity

[MSPATCHA] Initial patch decoding + applying.

Browse source 
Some initial work is done, all testcases pass now.
However a lot of the patch format is still unknown / not working,
so do not expect anything other than the apitest to work yet.
The next thing to do now, is to gather input, output and patch files,
and analyze how the compression and patching works on bigger /
more complex files.
CORE-15417
This commit is contained in:
Mark Jansen 2018-12-08 23:49:57 +01:00
parent 33e3a69a7f
commit 97d7206f8b
No known key found for this signature in database
GPG key ID: B39240EE84BEAE8B
5 changed files with 1263 additions and 21 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
dll/win32/mspatcha/CMakeLists.txt
View file

@ -5,6 +5,8 @@ spec2def(mspatcha.dll mspatcha.spec ADD_IMPORTLIB)
list(APPEND SOURCE
mspatcha_main.c
lzx.c
lzx.h
${CMAKE_CURRENT_BINARY_DIR}/mspatcha_stubs.c
${CMAKE_CURRENT_BINARY_DIR}/mspatcha.def)

834
dll/win32/mspatcha/lzx.c Normal file
View file

@ -0,0 +1,834 @@
/***************************************************************************
* lzx.c - LZX decompression routines *
* ------------------- *
* *
* maintainer: Jed Wing <jedwin@ugcs.caltech.edu> *
* source: modified lzx.c from cabextract v0.5 *
* notes: This file was taken from cabextract v0.5, which was, *
* itself, a modified version of the lzx decompression code *
* from unlzx. *
* *
* platforms: In its current incarnation, this file has been tested on *
* two different Linux platforms (one, redhat-based, with a *
* 2.1.2 glibc and gcc 2.95.x, and the other, Debian, with *
* 2.2.4 glibc and both gcc 2.95.4 and gcc 3.0.2). Both were *
* Intel x86 compatible machines. *
***************************************************************************/
/***************************************************************************
*
* Copyright(C) Stuart Caie
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library 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
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
*
***************************************************************************/
#include "lzx.h"
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "windef.h"
#include "winbase.h"
/* sized types */
typedef unsigned char UBYTE; /* 8 bits exactly */
typedef unsigned short UWORD; /* 16 bits (or more) */
/* some constants defined by the LZX specification */
#define LZX_MIN_MATCH (2)
#define LZX_MAX_MATCH (257)
#define LZX_NUM_CHARS (256)
#define LZX_BLOCKTYPE_INVALID (0) /* also blocktypes 4-7 invalid */
#define LZX_BLOCKTYPE_VERBATIM (1)
#define LZX_BLOCKTYPE_ALIGNED (2)
#define LZX_BLOCKTYPE_UNCOMPRESSED (3)
#define LZX_PRETREE_NUM_ELEMENTS (20)
#define LZX_ALIGNED_NUM_ELEMENTS (8) /* aligned offset tree #elements */
#define LZX_NUM_PRIMARY_LENGTHS (7) /* this one missing from spec! */
#define LZX_NUM_SECONDARY_LENGTHS (249) /* length tree #elements */
/* LZX huffman defines: tweak tablebits as desired */
#define LZX_PRETREE_MAXSYMBOLS (LZX_PRETREE_NUM_ELEMENTS)
#define LZX_PRETREE_TABLEBITS (6)
#define LZX_MAINTREE_MAXSYMBOLS (LZX_NUM_CHARS + 50*8)
#define LZX_MAINTREE_TABLEBITS (12)
#define LZX_LENGTH_MAXSYMBOLS (LZX_NUM_SECONDARY_LENGTHS+1)
#define LZX_LENGTH_TABLEBITS (12)
#define LZX_ALIGNED_MAXSYMBOLS (LZX_ALIGNED_NUM_ELEMENTS)
#define LZX_ALIGNED_TABLEBITS (7)
#define LZX_LENTABLE_SAFETY (64) /* we allow length table decoding overruns */
#define LZX_DECLARE_TABLE(tbl) \
UWORD tbl##_table[(1<<LZX_##tbl##_TABLEBITS) + (LZX_##tbl##_MAXSYMBOLS<<1)];\
UBYTE tbl##_len [LZX_##tbl##_MAXSYMBOLS + LZX_LENTABLE_SAFETY]
struct LZXstate
{
UBYTE *window; /* the actual decoding window */
ULONG window_size; /* window size (32Kb through 2Mb) */
ULONG actual_size; /* window size when it was first allocated */
ULONG window_posn; /* current offset within the window */
ULONG R0, R1, R2; /* for the LRU offset system */
UWORD main_elements; /* number of main tree elements */
int header_read; /* have we started decoding at all yet? */
UWORD block_type; /* type of this block */
ULONG block_length; /* uncompressed length of this block */
ULONG block_remaining; /* uncompressed bytes still left to decode */
ULONG frames_read; /* the number of CFDATA blocks processed */
LONG intel_filesize; /* magic header value used for transform */
LONG intel_curpos; /* current offset in transform space */
int intel_started; /* have we seen any translatable data yet? */
LZX_DECLARE_TABLE(PRETREE);
LZX_DECLARE_TABLE(MAINTREE);
LZX_DECLARE_TABLE(LENGTH);
LZX_DECLARE_TABLE(ALIGNED);
};
/* LZX decruncher */
/* Microsoft's LZX document and their implementation of the
* com.ms.util.cab Java package do not concur.
*
* In the LZX document, there is a table showing the correlation between
* window size and the number of position slots. It states that the 1MB
* window = 40 slots and the 2MB window = 42 slots. In the implementation,
* 1MB = 42 slots, 2MB = 50 slots. The actual calculation is 'find the
* first slot whose position base is equal to or more than the required
* window size'. This would explain why other tables in the document refer
* to 50 slots rather than 42.
*
* The constant NUM_PRIMARY_LENGTHS used in the decompression pseudocode
* is not defined in the specification.
*
* The LZX document does not state the uncompressed block has an
* uncompressed length field. Where does this length field come from, so
* we can know how large the block is? The implementation has it as the 24
* bits following after the 3 blocktype bits, before the alignment
* padding.
*
* The LZX document states that aligned offset blocks have their aligned
* offset huffman tree AFTER the main and length trees. The implementation
* suggests that the aligned offset tree is BEFORE the main and length
* trees.
*
* The LZX document decoding algorithm states that, in an aligned offset
* block, if an extra_bits value is 1, 2 or 3, then that number of bits
* should be read and the result added to the match offset. This is
* correct for 1 and 2, but not 3, where just a huffman symbol (using the
* aligned tree) should be read.
*
* Regarding the E8 preprocessing, the LZX document states 'No translation
* may be performed on the last 6 bytes of the input block'. This is
* correct. However, the pseudocode provided checks for the *E8 leader*
* up to the last 6 bytes. If the leader appears between -10 and -7 bytes
* from the end, this would cause the next four bytes to be modified, at
* least one of which would be in the last 6 bytes, which is not allowed
* according to the spec.
*
* The specification states that the huffman trees must always contain at
* least one element. However, many CAB files contain blocks where the
* length tree is completely empty (because there are no matches), and
* this is expected to succeed.
*/
/* LZX uses what it calls 'position slots' to represent match offsets.
* What this means is that a small 'position slot' number and a small
* offset from that slot are encoded instead of one large offset for
* every match.
* - position_base is an index to the position slot bases
* - extra_bits states how many bits of offset-from-base data is needed.
*/
static const UBYTE extra_bits[51] = {
0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13, 14, 14,
15, 15, 16, 16, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17,
17, 17, 17
};
static const ULONG position_base[51] = {
0, 1, 2, 3, 4, 6, 8, 12, 16, 24, 32, 48, 64, 96, 128, 192,
256, 384, 512, 768, 1024, 1536, 2048, 3072, 4096, 6144, 8192, 12288, 16384, 24576, 32768, 49152,
65536, 98304, 131072, 196608, 262144, 393216, 524288, 655360, 786432, 917504, 1048576, 1179648, 1310720, 1441792, 1572864, 1703936,
1835008, 1966080, 2097152
};
struct LZXstate *LZXinit(int window)
{
struct LZXstate *pState=NULL;
ULONG wndsize = 1 << window;
int i, posn_slots;
/* LZX supports window sizes of 2^15 (32Kb) through 2^21 (2Mb) */
/* if a previously allocated window is big enough, keep it */
if (window < 15 || window > 21) return NULL;
/* allocate state and associated window */
pState = HeapAlloc(GetProcessHeap(), 0, sizeof(struct LZXstate));
if (!(pState->window = HeapAlloc(GetProcessHeap(), 0, wndsize)))
{
HeapFree(GetProcessHeap(), 0, pState);
return NULL;
}
pState->actual_size = wndsize;
pState->window_size = wndsize;
/* calculate required position slots */
if (window == 20) posn_slots = 42;
else if (window == 21) posn_slots = 50;
else posn_slots = window << 1;
/** alternatively **/
/* posn_slots=i=0; while (i < wndsize) i += 1 << extra_bits[posn_slots++]; */
/* initialize other state */
pState->R0 = pState->R1 = pState->R2 = 1;
pState->main_elements = LZX_NUM_CHARS + (posn_slots << 3);
pState->header_read = 0;
pState->frames_read = 0;
pState->block_remaining = 0;
pState->block_type = LZX_BLOCKTYPE_INVALID;
pState->intel_curpos = 0;
pState->intel_started = 0;
pState->window_posn = 0;
/* initialise tables to 0 (because deltas will be applied to them) */
for (i = 0; i < LZX_MAINTREE_MAXSYMBOLS; i++) pState->MAINTREE_len[i] = 0;
for (i = 0; i < LZX_LENGTH_MAXSYMBOLS; i++) pState->LENGTH_len[i] = 0;
return pState;
}
void LZXteardown(struct LZXstate *pState)
{
if (pState)
{
HeapFree(GetProcessHeap(), 0, pState->window);
HeapFree(GetProcessHeap(), 0, pState);
}
}
int LZXreset(struct LZXstate *pState)
{
int i;
pState->R0 = pState->R1 = pState->R2 = 1;
pState->header_read = 0;
pState->frames_read = 0;
pState->block_remaining = 0;
pState->block_type = LZX_BLOCKTYPE_INVALID;
pState->intel_curpos = 0;
pState->intel_started = 0;
pState->window_posn = 0;
for (i = 0; i < LZX_MAINTREE_MAXSYMBOLS + LZX_LENTABLE_SAFETY; i++) pState->MAINTREE_len[i] = 0;
for (i = 0; i < LZX_LENGTH_MAXSYMBOLS + LZX_LENTABLE_SAFETY; i++) pState->LENGTH_len[i] = 0;
return DECR_OK;
}
/* Bitstream reading macros:
*
* INIT_BITSTREAM should be used first to set up the system
* READ_BITS(var,n) takes N bits from the buffer and puts them in var
*
* ENSURE_BITS(n) ensures there are at least N bits in the bit buffer
* PEEK_BITS(n) extracts (without removing) N bits from the bit buffer
* REMOVE_BITS(n) removes N bits from the bit buffer
*
* These bit access routines work by using the area beyond the MSB and the
* LSB as a free source of zeroes. This avoids having to mask any bits.
* So we have to know the bit width of the bitbuffer variable. This is
* sizeof(ULONG) * 8, also defined as ULONG_BITS
*/
/* number of bits in ULONG. Note: This must be at multiple of 16, and at
* least 32 for the bitbuffer code to work (ie, it must be able to ensure
* up to 17 bits - that's adding 16 bits when there's one bit left, or
* adding 32 bits when there are no bits left. The code should work fine
* for machines where ULONG >= 32 bits.
*/
#define ULONG_BITS (sizeof(ULONG)<<3)
#define INIT_BITSTREAM do { bitsleft = 0; bitbuf = 0; } while (0)
#define ENSURE_BITS(n) \
while (bitsleft < (n)) { \
bitbuf |= ((inpos[1]<<8)|inpos[0]) << (ULONG_BITS-16 - bitsleft); \
bitsleft += 16; inpos+=2; \
}
#define PEEK_BITS(n) (bitbuf >> (ULONG_BITS - (n)))
#define REMOVE_BITS(n) ((bitbuf <<= (n)), (bitsleft -= (n)))
#define READ_BITS(v,n) do { \
ENSURE_BITS(n); \
(v) = PEEK_BITS(n); \
REMOVE_BITS(n); \
} while (0)
/* Huffman macros */
#define TABLEBITS(tbl) (LZX_##tbl##_TABLEBITS)
#define MAXSYMBOLS(tbl) (LZX_##tbl##_MAXSYMBOLS)
#define SYMTABLE(tbl) (pState->tbl##_table)
#define LENTABLE(tbl) (pState->tbl##_len)
/* BUILD_TABLE(tablename) builds a huffman lookup table from code lengths.
* In reality, it just calls make_decode_table() with the appropriate
* values - they're all fixed by some #defines anyway, so there's no point
* writing each call out in full by hand.
*/
#define BUILD_TABLE(tbl) \
if (make_decode_table( \
MAXSYMBOLS(tbl), TABLEBITS(tbl), LENTABLE(tbl), SYMTABLE(tbl) \
)) { return DECR_ILLEGALDATA; }
/* READ_HUFFSYM(tablename, var) decodes one huffman symbol from the
* bitstream using the stated table and puts it in var.
*/
#define READ_HUFFSYM(tbl,var) do { \
ENSURE_BITS(16); \
hufftbl = SYMTABLE(tbl); \
if ((i = hufftbl[PEEK_BITS(TABLEBITS(tbl))]) >= MAXSYMBOLS(tbl)) { \
j = 1 << (ULONG_BITS - TABLEBITS(tbl)); \
do { \
j >>= 1; i <<= 1; i |= (bitbuf & j) ? 1 : 0; \
if (!j) { return DECR_ILLEGALDATA; } \
} while ((i = hufftbl[i]) >= MAXSYMBOLS(tbl)); \
} \
j = LENTABLE(tbl)[(var) = i]; \
REMOVE_BITS(j); \
} while (0)
/* READ_LENGTHS(tablename, first, last) reads in code lengths for symbols
* first to last in the given table. The code lengths are stored in their
* own special LZX way.
*/
#define READ_LENGTHS(tbl,first,last) do { \
lb.bb = bitbuf; lb.bl = bitsleft; lb.ip = inpos; \
if (lzx_read_lens(pState, LENTABLE(tbl),(first),(last),&lb)) { \
return DECR_ILLEGALDATA; \
} \
bitbuf = lb.bb; bitsleft = lb.bl; inpos = lb.ip; \
} while (0)
/* make_decode_table(nsyms, nbits, length[], table[])
*
* This function was coded by David Tritscher. It builds a fast huffman
* decoding table out of just a canonical huffman code lengths table.
*
* nsyms = total number of symbols in this huffman tree.
* nbits = any symbols with a code length of nbits or less can be decoded
* in one lookup of the table.
* length = A table to get code lengths from [0 to syms-1]
* table = The table to fill up with decoded symbols and pointers.
*
* Returns 0 for OK or 1 for error
*/
static int make_decode_table(ULONG nsyms, ULONG nbits, UBYTE *length, UWORD *table) {
register UWORD sym;
register ULONG leaf;
register UBYTE bit_num = 1;
ULONG fill;
ULONG pos = 0; /* the current position in the decode table */
ULONG table_mask = 1 << nbits;
ULONG bit_mask = table_mask >> 1; /* don't do 0 length codes */
ULONG next_symbol = bit_mask; /* base of allocation for long codes */
/* fill entries for codes short enough for a direct mapping */
while (bit_num <= nbits) {
for (sym = 0; sym < nsyms; sym++) {
if (length[sym] == bit_num) {
leaf = pos;
if((pos += bit_mask) > table_mask) return 1; /* table overrun */
/* fill all possible lookups of this symbol with the symbol itself */
fill = bit_mask;
while (fill-- > 0) table[leaf++] = sym;
}
}
bit_mask >>= 1;
bit_num++;
}
/* if there are any codes longer than nbits */
if (pos != table_mask) {
/* clear the remainder of the table */
for (sym = pos; sym < table_mask; sym++) table[sym] = 0;
/* give ourselves room for codes to grow by up to 16 more bits */
pos <<= 16;
table_mask <<= 16;
bit_mask = 1 << 15;
while (bit_num <= 16) {
for (sym = 0; sym < nsyms; sym++) {
if (length[sym] == bit_num) {
leaf = pos >> 16;
for (fill = 0; fill < bit_num - nbits; fill++) {
/* if this path hasn't been taken yet, 'allocate' two entries */
if (table[leaf] == 0) {
table[(next_symbol << 1)] = 0;
table[(next_symbol << 1) + 1] = 0;
table[leaf] = next_symbol++;
}
/* follow the path and select either left or right for next bit */
leaf = table[leaf] << 1;
if ((pos >> (15-fill)) & 1) leaf++;
}
table[leaf] = sym;
if ((pos += bit_mask) > table_mask) return 1; /* table overflow */
}
}
bit_mask >>= 1;
bit_num++;
}
}
/* full table? */
if (pos == table_mask) return 0;
/* either erroneous table, or all elements are 0 - let's find out. */
for (sym = 0; sym < nsyms; sym++) if (length[sym]) return 1;
return 0;
}
struct lzx_bits {
ULONG bb;
int bl;
UBYTE *ip;
};
static int lzx_read_lens(struct LZXstate *pState, UBYTE *lens, ULONG first, ULONG last, struct lzx_bits *lb) {
ULONG i,j, x,y;
int z;
register ULONG bitbuf = lb->bb;
register int bitsleft = lb->bl;
UBYTE *inpos = lb->ip;
UWORD *hufftbl;
for (x = 0; x < 20; x++) {
READ_BITS(y, 4);
LENTABLE(PRETREE)[x] = y;
}
BUILD_TABLE(PRETREE);
for (x = first; x < last; ) {
READ_HUFFSYM(PRETREE, z);
if (z == 17) {
READ_BITS(y, 4); y += 4;
while (y--) lens[x++] = 0;
}
else if (z == 18) {
READ_BITS(y, 5); y += 20;
while (y--) lens[x++] = 0;
}
else if (z == 19) {
READ_BITS(y, 1); y += 4;
READ_HUFFSYM(PRETREE, z);
z = lens[x] - z; if (z < 0) z += 17;
while (y--) lens[x++] = z;
}
else {
z = lens[x] - z; if (z < 0) z += 17;
lens[x++] = z;
}
}
lb->bb = bitbuf;
lb->bl = bitsleft;
lb->ip = inpos;
return 0;
}
int LZXdecompress(struct LZXstate *pState, unsigned char *inpos, unsigned char *outpos, int inlen, int outlen) {
UBYTE *endinp = inpos + inlen;
UBYTE *window = pState->window;
UBYTE *runsrc, *rundest;
UWORD *hufftbl; /* used in READ_HUFFSYM macro as chosen decoding table */
ULONG window_posn = pState->window_posn;
ULONG window_size = pState->window_size;
ULONG R0 = pState->R0;
ULONG R1 = pState->R1;
ULONG R2 = pState->R2;
register ULONG bitbuf;
register int bitsleft;
ULONG match_offset, i,j,k; /* ijk used in READ_HUFFSYM macro */
struct lzx_bits lb; /* used in READ_LENGTHS macro */
int togo = outlen, this_run, main_element, aligned_bits;
int match_length, length_footer, extra, verbatim_bits;
int copy_length;
INIT_BITSTREAM;
/* read header if necessary */
if (!pState->header_read) {
i = j = 0;
READ_BITS(k, 1); if (k) { READ_BITS(i,16); READ_BITS(j,16); }
pState->intel_filesize = (i << 16) | j; /* or 0 if not encoded */
pState->header_read = 1;
}
/* main decoding loop */
while (togo > 0) {
/* last block finished, new block expected */
if (pState->block_remaining == 0) {
if (pState->block_type == LZX_BLOCKTYPE_UNCOMPRESSED) {
if (pState->block_length & 1) inpos++; /* realign bitstream to word */
INIT_BITSTREAM;
}
READ_BITS(pState->block_type, 3);
READ_BITS(i, 16);
READ_BITS(j, 8);
pState->block_remaining = pState->block_length = (i << 8) | j;
switch (pState->block_type) {
case LZX_BLOCKTYPE_ALIGNED:
for (i = 0; i < 8; i++) { READ_BITS(j, 3); LENTABLE(ALIGNED)[i] = j; }
BUILD_TABLE(ALIGNED);
/* rest of aligned header is same as verbatim */
case LZX_BLOCKTYPE_VERBATIM:
READ_LENGTHS(MAINTREE, 0, 256);
READ_LENGTHS(MAINTREE, 256, pState->main_elements);
BUILD_TABLE(MAINTREE);
if (LENTABLE(MAINTREE)[0xE8] != 0) pState->intel_started = 1;
READ_LENGTHS(LENGTH, 0, LZX_NUM_SECONDARY_LENGTHS);
BUILD_TABLE(LENGTH);
break;
case LZX_BLOCKTYPE_UNCOMPRESSED:
pState->intel_started = 1; /* because we can't assume otherwise */
ENSURE_BITS(16); /* get up to 16 pad bits into the buffer */
if (bitsleft > 16) inpos -= 2; /* and align the bitstream! */
R0 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
R1 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
R2 = inpos[0]|(inpos[1]<<8)|(inpos[2]<<16)|(inpos[3]<<24);inpos+=4;
break;
default:
return DECR_ILLEGALDATA;
}
}
/* buffer exhaustion check */
if (inpos > endinp) {
/* it's possible to have a file where the next run is less than
* 16 bits in size. In this case, the READ_HUFFSYM() macro used
* in building the tables will exhaust the buffer, so we should
* allow for this, but not allow those accidentally read bits to
* be used (so we check that there are at least 16 bits
* remaining - in this boundary case they aren't really part of
* the compressed data)
*/
if (inpos > (endinp+2) || bitsleft < 16) return DECR_ILLEGALDATA;
}
while ((this_run = pState->block_remaining) > 0 && togo > 0) {
if (this_run > togo) this_run = togo;
togo -= this_run;
pState->block_remaining -= this_run;
/* apply 2^x-1 mask */
window_posn &= window_size - 1;
/* runs can't straddle the window wraparound */
if ((window_posn + this_run) > window_size)
return DECR_DATAFORMAT;
switch (pState->block_type) {
case LZX_BLOCKTYPE_VERBATIM:
while (this_run > 0) {
READ_HUFFSYM(MAINTREE, main_element);
if (main_element < LZX_NUM_CHARS) {
/* literal: 0 to LZX_NUM_CHARS-1 */
window[window_posn++] = main_element;
this_run--;
}
else {
/* match: LZX_NUM_CHARS + ((slot<<3) | length_header (3 bits)) */
main_element -= LZX_NUM_CHARS;
match_length = main_element & LZX_NUM_PRIMARY_LENGTHS;
if (match_length == LZX_NUM_PRIMARY_LENGTHS) {
READ_HUFFSYM(LENGTH, length_footer);
match_length += length_footer;
}
match_length += LZX_MIN_MATCH;
match_offset = main_element >> 3;
if (match_offset > 2) {
/* not repeated offset */
if (match_offset != 3) {
extra = extra_bits[match_offset];
READ_BITS(verbatim_bits, extra);
match_offset = position_base[match_offset] - 2 + verbatim_bits;
}
else {
match_offset = 1;
}
/* update repeated offset LRU queue */
R2 = R1; R1 = R0; R0 = match_offset;
}
else if (match_offset == 0) {
match_offset = R0;
}
else if (match_offset == 1) {
match_offset = R1;
R1 = R0; R0 = match_offset;
}
else /* match_offset == 2 */ {
match_offset = R2;
R2 = R0; R0 = match_offset;
}
rundest = window + window_posn;
this_run -= match_length;
/* copy any wrapped around source data */
if (window_posn >= match_offset) {
/* no wrap */
runsrc = rundest - match_offset;
} else {
runsrc = rundest + (window_size - match_offset);
copy_length = match_offset - window_posn;
if (copy_length < match_length) {
match_length -= copy_length;
window_posn += copy_length;
while (copy_length-- > 0) *rundest++ = *runsrc++;
runsrc = window;
}
}
window_posn += match_length;
/* copy match data - no worries about destination wraps */
while (match_length-- > 0) *rundest++ = *runsrc++;
}
}
break;
case LZX_BLOCKTYPE_ALIGNED:
while (this_run > 0) {
READ_HUFFSYM(MAINTREE, main_element);
if (main_element < LZX_NUM_CHARS) {
/* literal: 0 to LZX_NUM_CHARS-1 */
window[window_posn++] = main_element;
this_run--;
}
else {
/* match: LZX_NUM_CHARS + ((slot<<3) | length_header (3 bits)) */
main_element -= LZX_NUM_CHARS;
match_length = main_element & LZX_NUM_PRIMARY_LENGTHS;
if (match_length == LZX_NUM_PRIMARY_LENGTHS) {
READ_HUFFSYM(LENGTH, length_footer);
match_length += length_footer;
}
match_length += LZX_MIN_MATCH;
match_offset = main_element >> 3;
if (match_offset > 2) {
/* not repeated offset */
extra = extra_bits[match_offset];
match_offset = position_base[match_offset] - 2;
if (extra > 3) {
/* verbatim and aligned bits */
extra -= 3;
READ_BITS(verbatim_bits, extra);
match_offset += (verbatim_bits << 3);
READ_HUFFSYM(ALIGNED, aligned_bits);
match_offset += aligned_bits;
}
else if (extra == 3) {
/* aligned bits only */
READ_HUFFSYM(ALIGNED, aligned_bits);
match_offset += aligned_bits;
}
else if (extra > 0) { /* extra==1, extra==2 */
/* verbatim bits only */
READ_BITS(verbatim_bits, extra);
match_offset += verbatim_bits;
}
else /* extra == 0 */ {
/* ??? */
match_offset = 1;
}
/* update repeated offset LRU queue */
R2 = R1; R1 = R0; R0 = match_offset;
}
else if (match_offset == 0) {
match_offset = R0;
}
else if (match_offset == 1) {
match_offset = R1;
R1 = R0; R0 = match_offset;
}
else /* match_offset == 2 */ {
match_offset = R2;
R2 = R0; R0 = match_offset;
}
rundest = window + window_posn;
this_run -= match_length;
/* copy any wrapped around source data */
if (window_posn >= match_offset) {
/* no wrap */
runsrc = rundest - match_offset;
} else {
runsrc = rundest + (window_size - match_offset);
copy_length = match_offset - window_posn;
if (copy_length < match_length) {
match_length -= copy_length;
window_posn += copy_length;
while (copy_length-- > 0) *rundest++ = *runsrc++;
runsrc = window;
}
}
window_posn += match_length;
/* copy match data - no worries about destination wraps */
while (match_length-- > 0) *rundest++ = *runsrc++;
}
}
break;
case LZX_BLOCKTYPE_UNCOMPRESSED:
if ((inpos + this_run) > endinp) return DECR_ILLEGALDATA;
memcpy(window + window_posn, inpos, (size_t) this_run);
inpos += this_run; window_posn += this_run;
break;
default:
return DECR_ILLEGALDATA; /* might as well */
}
}
}
if (togo != 0) return DECR_ILLEGALDATA;
memcpy(outpos, window + ((!window_posn) ? window_size : window_posn) - outlen, (size_t) outlen);
pState->window_posn = window_posn;
pState->R0 = R0;
pState->R1 = R1;
pState->R2 = R2;
/* intel E8 decoding */
if ((pState->frames_read++ < 32768) && pState->intel_filesize != 0) {
if (outlen <= 6 || !pState->intel_started) {
pState->intel_curpos += outlen;
}
else {
UBYTE *data = outpos;
UBYTE *dataend = data + outlen - 10;
LONG curpos = pState->intel_curpos;
LONG filesize = pState->intel_filesize;
LONG abs_off, rel_off;
pState->intel_curpos = curpos + outlen;
while (data < dataend) {
if (*data++ != 0xE8) { curpos++; continue; }
abs_off = data[0] | (data[1]<<8) | (data[2]<<16) | (data[3]<<24);
if ((abs_off >= -curpos) && (abs_off < filesize)) {
rel_off = (abs_off >= 0) ? abs_off - curpos : abs_off + filesize;
data[0] = (UBYTE) rel_off;
data[1] = (UBYTE) (rel_off >> 8);
data[2] = (UBYTE) (rel_off >> 16);
data[3] = (UBYTE) (rel_off >> 24);
}
data += 4;
curpos += 5;
}
}
}
return DECR_OK;
}
#ifdef LZX_CHM_TESTDRIVER
int main(int c, char **v)
{
FILE *fin, *fout;
struct LZXstate state;
UBYTE ibuf[16384];
UBYTE obuf[32768];
int ilen, olen;
int status;
int i;
int count=0;
int w = atoi(v[1]);
LZXinit(&state, w);
fout = fopen(v[2], "wb");
for (i=3; i<c; i++)
{
fin = fopen(v[i], "rb");
ilen = fread(ibuf, 1, 16384, fin);
status = LZXdecompress(&state, ibuf, obuf, ilen, 32768);
switch (status)
{
case DECR_OK:
printf("ok\n");
fwrite(obuf, 1, 32768, fout);
break;
case DECR_DATAFORMAT:
printf("bad format\n");
break;
case DECR_ILLEGALDATA:
printf("illegal data\n");
break;
case DECR_NOMEMORY:
printf("no memory\n");
break;
default:
break;
}
fclose(fin);
if (++count == 2)
{
count = 0;
LZXreset(&state);
}
}
fclose(fout);
}
#endif

60
dll/win32/mspatcha/lzx.h Normal file
View file

@ -0,0 +1,60 @@
/***************************************************************************
* lzx.h - LZX decompression routines *
* ------------------- *
* *
* maintainer: Jed Wing <jedwin@ugcs.caltech.edu> *
* source: modified lzx.c from cabextract v0.5 *
* notes: This file was taken from cabextract v0.5, which was, *
* itself, a modified version of the lzx decompression code *
* from unlzx. *
***************************************************************************/
/***************************************************************************
*
* Copyright(C) Stuart Caie
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library 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
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
*
***************************************************************************/
#ifndef INCLUDED_LZX_H
#define INCLUDED_LZX_H
/* return codes */
#define DECR_OK (0)
#define DECR_DATAFORMAT (1)
#define DECR_ILLEGALDATA (2)
#define DECR_NOMEMORY (3)
/* opaque state structure */
struct LZXstate;
/* create an lzx state object */
struct LZXstate *LZXinit(int window);
/* destroy an lzx state object */
void LZXteardown(struct LZXstate *pState);
/* reset an lzx stream */
int LZXreset(struct LZXstate *pState);
/* decompress an LZX compressed block */
int LZXdecompress(struct LZXstate *pState,
unsigned char *inpos,
unsigned char *outpos,
int inlen,
int outlen);
#endif /* INCLUDED_LZX_H */

383
dll/win32/mspatcha/mspatcha_main.c
View file

@ -2,6 +2,7 @@
* PatchAPI
*
* Copyright 2011 David Hedberg for CodeWeavers
* Copyright 2018 Mark Jansen (mark.jansen@reactos.org)
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
@ -24,27 +25,200 @@
#include "winnls.h"
#include "ndk/rtlfuncs.h"
#include "patchapi.h"
#include "lzx.h"
#include "wine/debug.h"
static const char szHexString[] = "0123456789abcdef";
#define SIGNATURE_MIN_SIZE 9
#define UNKNOWN_FLAGS_COMBINATION 0x00c40001
WINE_DEFAULT_DEBUG_CHANNEL(mspatcha);
typedef struct _SAFE_READ
{
PBYTE Root;
DWORD Size;
PBYTE Ptr;
} SAFE_READ, *PSAFE_READ;
/**
* @name ReadByte
* Read the next byte available from @param pRead
*
* @param pRead
* The input buffer
*
* @return The byte, or 0
*/
BYTE ReadByte(PSAFE_READ pRead)
{
if (pRead->Ptr + sizeof(BYTE) <= (pRead->Root + pRead->Size))
{
BYTE Value = *(PBYTE)pRead->Ptr;
pRead->Ptr += sizeof(BYTE);
return Value;
}
pRead->Ptr = pRead->Root + pRead->Size;
return 0;
}
/**
* @name ReadUShort
* Read the next unsigned short available from @param pRead
*
* @param pRead
* The input buffer
*
* @return The unsigned short, or 0
*/
USHORT ReadUShort(PSAFE_READ pRead)
{
if (pRead->Ptr + sizeof(USHORT) <= (pRead->Root + pRead->Size))
{
USHORT Value = *(PUSHORT)pRead->Ptr;
pRead->Ptr += sizeof(USHORT);
return Value;
}
pRead->Ptr = pRead->Root + pRead->Size;
return 0;
}
/**
* @name ReadDWord
* Read the next dword available from @param pRead
*
* @param pRead
* The input buffer
*
* @return The dword, or 0
*/
DWORD ReadDWord(PSAFE_READ pRead)
{
if (pRead->Ptr + sizeof(DWORD) <= (pRead->Root + pRead->Size))
{
DWORD Value = *(PDWORD)pRead->Ptr;
pRead->Ptr += sizeof(DWORD);
return Value;
}
pRead->Ptr = pRead->Root + pRead->Size;
return 0;
}
/**
* @name DecodeDWord
* Read the next variable length-encoded dword from @param pRead
*
* @param pRead
* The input buffer
*
* @return The dword, or 0
*/
DWORD DecodeDWord(PSAFE_READ pRead)
{
UINT Result = 0, offset;
for (offset = 0; offset < 32; offset += 7)
{
DWORD val = ReadByte(pRead);
Result |= ((val & 0x7f) << offset);
if (val & 0x80)
break;
}
return Result;
}
/**
* @name DecodeInt
* Read the next variable length-encoded int from @param pRead
*
* @param pRead
* The input buffer
*
* @return The int, or 0
*/
INT DecodeInt(PSAFE_READ pRead)
{
INT Result = 0, offset;
for (offset = 0; offset < 32; offset += 6)
{
INT val = (INT)(DWORD)ReadByte(pRead);
Result |= ((val & 0x3f) << offset);
if (val & 0x80)
{
if (val & 0x40)
Result *= -1;
break;
}
}
return Result;
}
typedef struct _PATCH_HEADER
{
DWORD Flags;
DWORD ImageBase;
DWORD ImageTimeStamp;
DWORD OutputSize;
DWORD OutputCrc;
DWORD OldSize;
DWORD OldCrc;
DWORD DataSize; // Payload after the patch header
} PATCH_HEADER;
/**
* @name MapFile
* Map a view of a file into readonly memory
*
* @param hFile
* The input file handle, readable
*
* @param dwSize
* Mapped file size (out)
*
* @return A Pointer to the start of the memory
*/
static PBYTE MapFile(HANDLE hFile, DWORD* dwSize)
{
HANDLE hMap;
PVOID pView;
*dwSize = GetFileSize(hFile, NULL);
hMap = CreateFileMappingW(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
if (hMap != INVALID_HANDLE_VALUE)
{
pView = MapViewOfFile(hMap, FILE_MAP_READ, 0, 0, 0);
CloseHandle(hMap);
return pView;
}
return NULL;
}
/*****************************************************
* DllMain (MSPATCHA.@)
*/
BOOL WINAPI DllMain(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpvReserved)
{
TRACE("(0x%p, %d, %p)\n", hinstDLL, fdwReason, lpvReserved);
switch (fdwReason)
{
case DLL_WINE_PREATTACH:
return FALSE; /* prefer native version */
case DLL_PROCESS_ATTACH:
DisableThreadLibraryCalls(hinstDLL);
break;
case DLL_PROCESS_ATTACH:
DisableThreadLibraryCalls(hinstDLL);
break;
}
return TRUE;
@ -81,15 +255,188 @@ BOOL WINAPI ApplyPatchToFileA(LPCSTR patch_file, LPCSTR old_file, LPCSTR new_fil
return ret;
}
/**
* @name ParseHeader
* Parse a Patch file header
* @note The current implementation is far from complete!
*
* @param Patch
* Buffer pointing to the raw patch data
*
* @param Header
* The result of the parsed header
*
* @return STATUS_SUCCESS on success, an error code otherwise
*/
DWORD ParseHeader(SAFE_READ* Patch, PATCH_HEADER* Header)
{
DWORD Crc, Unknown;
int Delta;
ZeroMemory(Header, sizeof(*Header));
/* Validate the patch */
Crc = RtlComputeCrc32(0, Patch->Root, Patch->Size);
if (Crc != ~0)
return ERROR_PATCH_CORRUPT;
if (ReadDWord(Patch) != '91AP')
return ERROR_PATCH_DECODE_FAILURE;
/* Read the flags, warn about an unknown combination */
Header->Flags = ReadDWord(Patch);
if (Header->Flags ^ UNKNOWN_FLAGS_COMBINATION)
ERR("Unknown flags: 0x%x, patch will most likely fail\n", Header->Flags ^ UNKNOWN_FLAGS_COMBINATION);
/* 0x5bb3284e, 0x5bb33562, 0x5bb357b1 */
Unknown = ReadDWord(Patch);
TRACE("Unknown: 0x%x\n", Unknown);
Header->OutputSize = DecodeDWord(Patch);
Header->OutputCrc = ReadDWord(Patch);
Unknown = ReadByte(Patch);
if (Unknown != 1)
ERR("Field after CRC is not 1 but %u\n", Unknown);
Delta = DecodeInt(Patch);
Header->OldSize = Header->OutputSize + Delta;
Header->OldCrc = ReadDWord(Patch);
Unknown = ReadUShort(Patch);
if (Unknown != 0)
ERR("Field1 after OldCrc is not 0 but %u\n", Unknown);
Unknown = DecodeDWord(Patch);
if (Unknown != 0)
ERR("Field2 after OldCrc is not 0 but %u\n", Unknown);
Header->DataSize = DecodeDWord(Patch);
/* Remaining data, minus the CRC appended */
if (Header->DataSize != (Patch->Size - (Patch->Ptr - Patch->Root) - sizeof(DWORD)))
{
ERR("Unable to read header, check previous logging!\n");
return ERROR_PATCH_DECODE_FAILURE;
}
return STATUS_SUCCESS;
}
/**
* @name CreateNewFileFromPatch
* Using the input @param Header and @param Patch, create a new file on @param new_file
*
* @param Header
* Parsed / preprocessed patch header
*
* @param Patch
* Memory buffer pointing to the patch payload
*
* @param new_file
* A handle to the output file. This file will be resized
*
* @return STATUS_SUCCESS on success, an error code otherwise
*/
DWORD CreateNewFileFromPatch(PATCH_HEADER* Header, SAFE_READ* Patch, HANDLE new_file)
{
SAFE_READ NewFile;
HANDLE hMap;
USHORT BlockSize;
DWORD dwStatus;
struct LZXstate* state;
int lzxResult;
hMap = CreateFileMappingW(new_file, NULL, PAGE_READWRITE, 0, Header->OutputSize, NULL);
if (hMap == INVALID_HANDLE_VALUE)
return ERROR_PATCH_NOT_AVAILABLE;
NewFile.Root = NewFile.Ptr = MapViewOfFile(hMap, FILE_MAP_WRITE, 0, 0, 0);
CloseHandle(hMap);
NewFile.Size = Header->OutputSize;
if (!NewFile.Root)
return ERROR_PATCH_NOT_AVAILABLE;
/* At this point Patch->Ptr should point to the payload */
BlockSize = ReadUShort(Patch);
/* This window size does not work on all files (for example, MS SQL Express 2008 setup) */
state = LZXinit(17);
if (state)
{
lzxResult = LZXdecompress(state, Patch->Ptr, NewFile.Ptr, BlockSize, NewFile.Size);
LZXteardown(state);
if (lzxResult == DECR_OK)
dwStatus = STATUS_SUCCESS;
else
dwStatus = ERROR_PATCH_DECODE_FAILURE;
}
else
{
dwStatus = ERROR_INSUFFICIENT_BUFFER;
}
UnmapViewOfFile(NewFile.Root);
return dwStatus;
}
/*****************************************************
* ApplyPatchToFileByHandles (MSPATCHA.2)
*/
BOOL WINAPI ApplyPatchToFileByHandles(HANDLE patch_file, HANDLE old_file, HANDLE new_file, ULONG apply_flags)
{
FIXME("stub - %p, %p, %p, %08x\n", patch_file, old_file, new_file, apply_flags);
SAFE_READ Patch, OldFile;
DWORD dwStatus;
PATCH_HEADER Header;
SetLastError(ERROR_CALL_NOT_IMPLEMENTED);
return FALSE;
Patch.Root = Patch.Ptr = MapFile(patch_file, &Patch.Size);
if (!Patch.Root)
{
SetLastError(ERROR_PATCH_CORRUPT);
return FALSE;
}
/* Let's decode the header */
dwStatus = ParseHeader(&Patch, &Header);
if (dwStatus != STATUS_SUCCESS)
{
UnmapViewOfFile(Patch.Root);
SetLastError(dwStatus);
return FALSE;
}
OldFile.Root = OldFile.Ptr = MapFile(old_file, &OldFile.Size);
if (OldFile.Root)
{
DWORD dwCrc;
/* Verify the input file */
dwCrc = RtlComputeCrc32(0, OldFile.Root, OldFile.Size);
if (OldFile.Size == Header.OldSize && dwCrc == Header.OldCrc)
{
if (apply_flags & APPLY_OPTION_TEST_ONLY)
dwStatus = STATUS_SUCCESS;
else
dwStatus = CreateNewFileFromPatch(&Header, &Patch, new_file);
}
else
{
dwStatus = ERROR_PATCH_WRONG_FILE;
}
UnmapViewOfFile(OldFile.Root);
}
else
{
dwStatus = GetLastError();
if (dwStatus == STATUS_SUCCESS)
dwStatus = ERROR_PATCH_NOT_AVAILABLE;
}
UnmapViewOfFile(Patch.Root);
SetLastError(dwStatus);
return dwStatus == STATUS_SUCCESS;
}
/*****************************************************
@ -153,9 +500,8 @@ BOOL WINAPI GetFilePatchSignatureByHandle(HANDLE hFile, ULONG flags, PVOID data,
PPATCH_RETAIN_RANGE retain_range, ULONG bufsize, PVOID buffer)
{
BOOL ret = FALSE;
HANDLE hMap;
DWORD dwSize, ulCrc;
PVOID pView;
PBYTE pData;
if (flags)
FIXME("Unhandled flags 0x%x\n", flags);
@ -164,19 +510,14 @@ BOOL WINAPI GetFilePatchSignatureByHandle(HANDLE hFile, ULONG flags, PVOID data,
if (retain_range_count)
FIXME("Unhandled ignore_range_count %u\n", retain_range_count);
dwSize = GetFileSize(hFile, NULL);
hMap = CreateFileMappingW(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
if (hMap != INVALID_HANDLE_VALUE)
if ((pData = MapFile(hFile, &dwSize)))
{
pView = MapViewOfFile(hMap, FILE_MAP_READ, 0, 0, 0);
CloseHandle(hMap);
if (dwSize >= 2 && *(PWORD)pView == IMAGE_DOS_SIGNATURE)
if (dwSize >= 2 && *(PWORD)pData == IMAGE_DOS_SIGNATURE)
{
FIXME("Potentially unimplemented case, normalized signature\n");
}
ulCrc = RtlComputeCrc32(0, pView, dwSize);
ulCrc = RtlComputeCrc32(0, pData, dwSize);
if (bufsize >= SIGNATURE_MIN_SIZE)
{
char *pBuffer = buffer;
@ -188,7 +529,7 @@ BOOL WINAPI GetFilePatchSignatureByHandle(HANDLE hFile, ULONG flags, PVOID data,
}
ret = TRUE;
}
UnmapViewOfFile(pView);
UnmapViewOfFile(pData);
if (bufsize < SIGNATURE_MIN_SIZE)
SetLastError(ERROR_INSUFFICIENT_BUFFER);

5
sdk/include/psdk/patchapi.h
View file

@ -32,8 +32,13 @@ extern "C" {
//
// apply error codes
//
#define ERROR_PATCH_DECODE_FAILURE 0xC00E4101
#define ERROR_PATCH_CORRUPT 0xC00E4102
#define ERROR_PATCH_NEWER_FORMAT 0xC00E4103
#define ERROR_PATCH_WRONG_FILE 0xC00E4104
#define ERROR_PATCH_NOT_NECESSARY 0xC00E4105
#define ERROR_PATCH_NOT_AVAILABLE 0xC00E4106
typedef struct _PATCH_IGNORE_RANGE
{