1 /***************************************************************************
2 * chm_lib.c - CHM archive manipulation routines *
3 * ------------------- *
5 * author: Jed Wing <jedwin@ugcs.caltech.edu> *
7 * notes: These routines are meant for the manipulation of microsoft *
8 * .chm (compiled html help) files, but may likely be used *
9 * for the manipulation of any ITSS archive, if ever ITSS *
10 * archives are used for any other purpose. *
12 * Note also that the section names are statically handled. *
13 * To be entirely correct, the section names should be read *
14 * from the section names meta-file, and then the various *
15 * content sections and the "transforms" to apply to the data *
16 * they contain should be inferred from the section name and *
17 * the meta-files referenced using that name; however, all of *
18 * the files I've been able to get my hands on appear to have *
19 * only two sections: Uncompressed and MSCompressed. *
20 * Additionally, the ITSS.DLL file included with Windows does *
21 * not appear to handle any different transforms than the *
22 * simple LZX-transform. Furthermore, the list of transforms *
23 * to apply is broken, in that only half the required space *
24 * is allocated for the list. (It appears as though the *
25 * space is allocated for ASCII strings, but the strings are *
26 * written as unicode. As a result, only the first half of *
27 * the string appears.) So this is probably not too big of *
28 * a deal, at least until CHM v4 (MS .lit files), which also *
29 * incorporate encryption, of some description. *
31 ***************************************************************************/
33 /***************************************************************************
35 * This library is free software; you can redistribute it and/or
36 * modify it under the terms of the GNU Lesser General Public
37 * License as published by the Free Software Foundation; either
38 * version 2.1 of the License, or (at your option) any later version.
40 * This library is distributed in the hope that it will be useful,
41 * but WITHOUT ANY WARRANTY; without even the implied warranty of
42 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
43 * Lesser General Public License for more details.
45 * You should have received a copy of the GNU Lesser General Public
46 * License along with this library; if not, write to the Free Software
47 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
49 ***************************************************************************/
51 /***************************************************************************
53 * Adapted for Wine by Mike McCormack *
55 ***************************************************************************/
58 #include "wine/port.h"
67 #include "wine/unicode.h"
72 #define CHM_ACQUIRE_LOCK(a) do { \
73 EnterCriticalSection(&(a)); \
75 #define CHM_RELEASE_LOCK(a) do { \
76 LeaveCriticalSection(&(a)); \
79 #define CHM_NULL_FD (INVALID_HANDLE_VALUE)
80 #define CHM_CLOSE_FILE(fd) CloseHandle((fd))
83 * defines related to tuning
85 #ifndef CHM_MAX_BLOCKS_CACHED
86 #define CHM_MAX_BLOCKS_CACHED 5
88 #define CHM_PARAM_MAX_BLOCKS_CACHED 0
91 * architecture specific defines
93 * Note: as soon as C99 is more widespread, the below defines should
94 * probably just use the C99 sized-int types.
96 * The following settings will probably work for many platforms. The sizes
97 * don't have to be exactly correct, but the types must accommodate at least as
98 * many bits as they specify.
101 /* i386, 32-bit, Windows */
104 typedef USHORT UInt16
;
106 typedef DWORD UInt32
;
107 typedef LONGLONG Int64
;
108 typedef ULONGLONG UInt64
;
110 /* utilities for unmarshalling data */
111 static int _unmarshal_char_array(unsigned char **pData
,
112 unsigned int *pLenRemain
,
116 if (count
<= 0 || (unsigned int)count
> *pLenRemain
)
118 memcpy(dest
, (*pData
), count
);
120 *pLenRemain
-= count
;
124 static int _unmarshal_uchar_array(unsigned char **pData
,
125 unsigned int *pLenRemain
,
129 if (count
<= 0 || (unsigned int)count
> *pLenRemain
)
131 memcpy(dest
, (*pData
), count
);
133 *pLenRemain
-= count
;
137 static int _unmarshal_int32(unsigned char **pData
,
138 unsigned int *pLenRemain
,
143 *dest
= (*pData
)[0] | (*pData
)[1]<<8 | (*pData
)[2]<<16 | (*pData
)[3]<<24;
149 static int _unmarshal_uint32(unsigned char **pData
,
150 unsigned int *pLenRemain
,
155 *dest
= (*pData
)[0] | (*pData
)[1]<<8 | (*pData
)[2]<<16 | (*pData
)[3]<<24;
161 static int _unmarshal_int64(unsigned char **pData
,
162 unsigned int *pLenRemain
,
173 temp
|= (*pData
)[i
-1];
181 static int _unmarshal_uint64(unsigned char **pData
,
182 unsigned int *pLenRemain
,
193 temp
|= (*pData
)[i
-1];
201 static int _unmarshal_uuid(unsigned char **pData
,
202 unsigned int *pDataLen
,
205 return _unmarshal_uchar_array(pData
, pDataLen
, dest
, 16);
208 /* names of sections essential to decompression */
209 static const WCHAR _CHMU_RESET_TABLE
[] = {
210 ':',':','D','a','t','a','S','p','a','c','e','/',
211 'S','t','o','r','a','g','e','/',
212 'M','S','C','o','m','p','r','e','s','s','e','d','/',
213 'T','r','a','n','s','f','o','r','m','/',
214 '{','7','F','C','2','8','9','4','0','-','9','D','3','1',
215 '-','1','1','D','0','-','9','B','2','7','-',
216 '0','0','A','0','C','9','1','E','9','C','7','C','}','/',
217 'I','n','s','t','a','n','c','e','D','a','t','a','/',
218 'R','e','s','e','t','T','a','b','l','e',0
220 static const WCHAR _CHMU_LZXC_CONTROLDATA
[] = {
221 ':',':','D','a','t','a','S','p','a','c','e','/',
222 'S','t','o','r','a','g','e','/',
223 'M','S','C','o','m','p','r','e','s','s','e','d','/',
224 'C','o','n','t','r','o','l','D','a','t','a',0
226 static const WCHAR _CHMU_CONTENT
[] = {
227 ':',':','D','a','t','a','S','p','a','c','e','/',
228 'S','t','o','r','a','g','e','/',
229 'M','S','C','o','m','p','r','e','s','s','e','d','/',
230 'C','o','n','t','e','n','t',0
234 * structures local to this module
237 /* structure of ITSF headers */
238 #define _CHM_ITSF_V2_LEN (0x58)
239 #define _CHM_ITSF_V3_LEN (0x60)
242 char signature
[4]; /* 0 (ITSF) */
243 Int32 version
; /* 4 */
244 Int32 header_len
; /* 8 */
245 Int32 unknown_000c
; /* c */
246 UInt32 last_modified
; /* 10 */
247 UInt32 lang_id
; /* 14 */
248 UChar dir_uuid
[16]; /* 18 */
249 UChar stream_uuid
[16]; /* 28 */
250 UInt64 unknown_offset
; /* 38 */
251 UInt64 unknown_len
; /* 40 */
252 UInt64 dir_offset
; /* 48 */
253 UInt64 dir_len
; /* 50 */
254 UInt64 data_offset
; /* 58 (Not present before V3) */
255 }; /* __attribute__ ((aligned (1))); */
257 static int _unmarshal_itsf_header(unsigned char **pData
,
258 unsigned int *pDataLen
,
259 struct chmItsfHeader
*dest
)
261 /* we only know how to deal with the 0x58 and 0x60 byte structures */
262 if (*pDataLen
!= _CHM_ITSF_V2_LEN
&& *pDataLen
!= _CHM_ITSF_V3_LEN
)
265 /* unmarshal common fields */
266 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
267 _unmarshal_int32 (pData
, pDataLen
, &dest
->version
);
268 _unmarshal_int32 (pData
, pDataLen
, &dest
->header_len
);
269 _unmarshal_int32 (pData
, pDataLen
, &dest
->unknown_000c
);
270 _unmarshal_uint32 (pData
, pDataLen
, &dest
->last_modified
);
271 _unmarshal_uint32 (pData
, pDataLen
, &dest
->lang_id
);
272 _unmarshal_uuid (pData
, pDataLen
, dest
->dir_uuid
);
273 _unmarshal_uuid (pData
, pDataLen
, dest
->stream_uuid
);
274 _unmarshal_uint64 (pData
, pDataLen
, &dest
->unknown_offset
);
275 _unmarshal_uint64 (pData
, pDataLen
, &dest
->unknown_len
);
276 _unmarshal_uint64 (pData
, pDataLen
, &dest
->dir_offset
);
277 _unmarshal_uint64 (pData
, pDataLen
, &dest
->dir_len
);
279 /* error check the data */
280 /* XXX: should also check UUIDs, probably, though with a version 3 file,
281 * current MS tools do not seem to use them.
283 if (memcmp(dest
->signature
, "ITSF", 4) != 0)
285 if (dest
->version
== 2)
287 if (dest
->header_len
< _CHM_ITSF_V2_LEN
)
290 else if (dest
->version
== 3)
292 if (dest
->header_len
< _CHM_ITSF_V3_LEN
)
298 /* now, if we have a V3 structure, unmarshal the rest.
299 * otherwise, compute it
301 if (dest
->version
== 3)
304 _unmarshal_uint64(pData
, pDataLen
, &dest
->data_offset
);
309 dest
->data_offset
= dest
->dir_offset
+ dest
->dir_len
;
314 /* structure of ITSP headers */
315 #define _CHM_ITSP_V1_LEN (0x54)
318 char signature
[4]; /* 0 (ITSP) */
319 Int32 version
; /* 4 */
320 Int32 header_len
; /* 8 */
321 Int32 unknown_000c
; /* c */
322 UInt32 block_len
; /* 10 */
323 Int32 blockidx_intvl
; /* 14 */
324 Int32 index_depth
; /* 18 */
325 Int32 index_root
; /* 1c */
326 Int32 index_head
; /* 20 */
327 Int32 unknown_0024
; /* 24 */
328 UInt32 num_blocks
; /* 28 */
329 Int32 unknown_002c
; /* 2c */
330 UInt32 lang_id
; /* 30 */
331 UChar system_uuid
[16]; /* 34 */
332 UChar unknown_0044
[16]; /* 44 */
333 }; /* __attribute__ ((aligned (1))); */
335 static int _unmarshal_itsp_header(unsigned char **pData
,
336 unsigned int *pDataLen
,
337 struct chmItspHeader
*dest
)
339 /* we only know how to deal with a 0x54 byte structures */
340 if (*pDataLen
!= _CHM_ITSP_V1_LEN
)
343 /* unmarshal fields */
344 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
345 _unmarshal_int32 (pData
, pDataLen
, &dest
->version
);
346 _unmarshal_int32 (pData
, pDataLen
, &dest
->header_len
);
347 _unmarshal_int32 (pData
, pDataLen
, &dest
->unknown_000c
);
348 _unmarshal_uint32 (pData
, pDataLen
, &dest
->block_len
);
349 _unmarshal_int32 (pData
, pDataLen
, &dest
->blockidx_intvl
);
350 _unmarshal_int32 (pData
, pDataLen
, &dest
->index_depth
);
351 _unmarshal_int32 (pData
, pDataLen
, &dest
->index_root
);
352 _unmarshal_int32 (pData
, pDataLen
, &dest
->index_head
);
353 _unmarshal_int32 (pData
, pDataLen
, &dest
->unknown_0024
);
354 _unmarshal_uint32 (pData
, pDataLen
, &dest
->num_blocks
);
355 _unmarshal_int32 (pData
, pDataLen
, &dest
->unknown_002c
);
356 _unmarshal_uint32 (pData
, pDataLen
, &dest
->lang_id
);
357 _unmarshal_uuid (pData
, pDataLen
, dest
->system_uuid
);
358 _unmarshal_uchar_array(pData
, pDataLen
, dest
->unknown_0044
, 16);
360 /* error check the data */
361 if (memcmp(dest
->signature
, "ITSP", 4) != 0)
363 if (dest
->version
!= 1)
365 if (dest
->header_len
!= _CHM_ITSP_V1_LEN
)
371 /* structure of PMGL headers */
372 static const char _chm_pmgl_marker
[4] = "PMGL";
373 #define _CHM_PMGL_LEN (0x14)
376 char signature
[4]; /* 0 (PMGL) */
377 UInt32 free_space
; /* 4 */
378 UInt32 unknown_0008
; /* 8 */
379 Int32 block_prev
; /* c */
380 Int32 block_next
; /* 10 */
381 }; /* __attribute__ ((aligned (1))); */
383 static int _unmarshal_pmgl_header(unsigned char **pData
,
384 unsigned int *pDataLen
,
385 struct chmPmglHeader
*dest
)
387 /* we only know how to deal with a 0x14 byte structures */
388 if (*pDataLen
!= _CHM_PMGL_LEN
)
391 /* unmarshal fields */
392 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
393 _unmarshal_uint32 (pData
, pDataLen
, &dest
->free_space
);
394 _unmarshal_uint32 (pData
, pDataLen
, &dest
->unknown_0008
);
395 _unmarshal_int32 (pData
, pDataLen
, &dest
->block_prev
);
396 _unmarshal_int32 (pData
, pDataLen
, &dest
->block_next
);
398 /* check structure */
399 if (memcmp(dest
->signature
, _chm_pmgl_marker
, 4) != 0)
405 /* structure of PMGI headers */
406 static const char _chm_pmgi_marker
[4] = "PMGI";
407 #define _CHM_PMGI_LEN (0x08)
410 char signature
[4]; /* 0 (PMGI) */
411 UInt32 free_space
; /* 4 */
412 }; /* __attribute__ ((aligned (1))); */
414 static int _unmarshal_pmgi_header(unsigned char **pData
,
415 unsigned int *pDataLen
,
416 struct chmPmgiHeader
*dest
)
418 /* we only know how to deal with a 0x8 byte structures */
419 if (*pDataLen
!= _CHM_PMGI_LEN
)
422 /* unmarshal fields */
423 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
424 _unmarshal_uint32 (pData
, pDataLen
, &dest
->free_space
);
426 /* check structure */
427 if (memcmp(dest
->signature
, _chm_pmgi_marker
, 4) != 0)
433 /* structure of LZXC reset table */
434 #define _CHM_LZXC_RESETTABLE_V1_LEN (0x28)
435 struct chmLzxcResetTable
441 UInt64 uncompressed_len
;
442 UInt64 compressed_len
;
444 }; /* __attribute__ ((aligned (1))); */
446 static int _unmarshal_lzxc_reset_table(unsigned char **pData
,
447 unsigned int *pDataLen
,
448 struct chmLzxcResetTable
*dest
)
450 /* we only know how to deal with a 0x28 byte structures */
451 if (*pDataLen
!= _CHM_LZXC_RESETTABLE_V1_LEN
)
454 /* unmarshal fields */
455 _unmarshal_uint32 (pData
, pDataLen
, &dest
->version
);
456 _unmarshal_uint32 (pData
, pDataLen
, &dest
->block_count
);
457 _unmarshal_uint32 (pData
, pDataLen
, &dest
->unknown
);
458 _unmarshal_uint32 (pData
, pDataLen
, &dest
->table_offset
);
459 _unmarshal_uint64 (pData
, pDataLen
, &dest
->uncompressed_len
);
460 _unmarshal_uint64 (pData
, pDataLen
, &dest
->compressed_len
);
461 _unmarshal_uint64 (pData
, pDataLen
, &dest
->block_len
);
463 /* check structure */
464 if (dest
->version
!= 2)
470 /* structure of LZXC control data block */
471 #define _CHM_LZXC_MIN_LEN (0x18)
472 #define _CHM_LZXC_V2_LEN (0x1c)
473 struct chmLzxcControlData
476 char signature
[4]; /* 4 (LZXC) */
477 UInt32 version
; /* 8 */
478 UInt32 resetInterval
; /* c */
479 UInt32 windowSize
; /* 10 */
480 UInt32 windowsPerReset
; /* 14 */
481 UInt32 unknown_18
; /* 18 */
484 static int _unmarshal_lzxc_control_data(unsigned char **pData
,
485 unsigned int *pDataLen
,
486 struct chmLzxcControlData
*dest
)
488 /* we want at least 0x18 bytes */
489 if (*pDataLen
< _CHM_LZXC_MIN_LEN
)
492 /* unmarshal fields */
493 _unmarshal_uint32 (pData
, pDataLen
, &dest
->size
);
494 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
495 _unmarshal_uint32 (pData
, pDataLen
, &dest
->version
);
496 _unmarshal_uint32 (pData
, pDataLen
, &dest
->resetInterval
);
497 _unmarshal_uint32 (pData
, pDataLen
, &dest
->windowSize
);
498 _unmarshal_uint32 (pData
, pDataLen
, &dest
->windowsPerReset
);
500 if (*pDataLen
>= _CHM_LZXC_V2_LEN
)
501 _unmarshal_uint32 (pData
, pDataLen
, &dest
->unknown_18
);
503 dest
->unknown_18
= 0;
505 if (dest
->version
== 2)
507 dest
->resetInterval
*= 0x8000;
508 dest
->windowSize
*= 0x8000;
510 if (dest
->windowSize
== 0 || dest
->resetInterval
== 0)
513 /* for now, only support resetInterval a multiple of windowSize/2 */
514 if (dest
->windowSize
== 1)
516 if ((dest
->resetInterval
% (dest
->windowSize
/2)) != 0)
519 /* check structure */
520 if (memcmp(dest
->signature
, "LZXC", 4) != 0)
526 /* the structure used for chm file handles */
531 CRITICAL_SECTION mutex
;
532 CRITICAL_SECTION lzx_mutex
;
533 CRITICAL_SECTION cache_mutex
;
543 struct chmUnitInfo rt_unit
;
544 struct chmUnitInfo cn_unit
;
545 struct chmLzxcResetTable reset_table
;
547 /* LZX control data */
548 int compression_enabled
;
550 UInt32 reset_interval
;
551 UInt32 reset_blkcount
;
553 /* decompressor state */
554 struct LZXstate
*lzx_state
;
557 /* cache for decompressed blocks */
558 UChar
**cache_blocks
;
559 Int64
*cache_block_indices
;
560 Int32 cache_num_blocks
;
564 * utility functions local to this module
567 /* utility function to handle differences between {pread,read}(64)? */
568 static Int64
_chm_fetch_bytes(struct chmFile
*h
,
574 if (h
->fd
== CHM_NULL_FD
)
577 CHM_ACQUIRE_LOCK(h
->mutex
);
578 /* NOTE: this might be better done with CreateFileMapping, et cetera... */
580 LARGE_INTEGER old_pos
, new_pos
;
583 /* awkward Win32 Seek/Tell */
584 new_pos
.QuadPart
= 0;
585 SetFilePointerEx( h
->fd
, new_pos
, &old_pos
, FILE_CURRENT
);
586 new_pos
.QuadPart
= os
;
587 SetFilePointerEx( h
->fd
, new_pos
, NULL
, FILE_BEGIN
);
599 /* restore original position */
600 SetFilePointerEx( h
->fd
, old_pos
, NULL
, FILE_BEGIN
);
602 CHM_RELEASE_LOCK(h
->mutex
);
607 * set a parameter on the file handle.
608 * valid parameter types:
609 * CHM_PARAM_MAX_BLOCKS_CACHED:
610 * how many decompressed blocks should be cached? A simple
611 * caching scheme is used, wherein the index of the block is
612 * used as a hash value, and hash collision results in the
613 * invalidation of the previously cached block.
615 static void chm_set_param(struct chmFile
*h
,
621 case CHM_PARAM_MAX_BLOCKS_CACHED
:
622 CHM_ACQUIRE_LOCK(h
->cache_mutex
);
623 if (paramVal
!= h
->cache_num_blocks
)
629 /* allocate new cached blocks */
630 newBlocks
= HeapAlloc(GetProcessHeap(), 0, paramVal
* sizeof (UChar
*));
631 newIndices
= HeapAlloc(GetProcessHeap(), 0, paramVal
* sizeof (UInt64
));
632 for (i
=0; i
<paramVal
; i
++)
638 /* re-distribute old cached blocks */
641 for (i
=0; i
<h
->cache_num_blocks
; i
++)
643 int newSlot
= (int)(h
->cache_block_indices
[i
] % paramVal
);
645 if (h
->cache_blocks
[i
])
647 /* in case of collision, destroy newcomer */
648 if (newBlocks
[newSlot
])
650 HeapFree(GetProcessHeap(), 0, h
->cache_blocks
[i
]);
651 h
->cache_blocks
[i
] = NULL
;
655 newBlocks
[newSlot
] = h
->cache_blocks
[i
];
656 newIndices
[newSlot
] =
657 h
->cache_block_indices
[i
];
662 HeapFree(GetProcessHeap(), 0, h
->cache_blocks
);
663 HeapFree(GetProcessHeap(), 0, h
->cache_block_indices
);
666 /* now, set new values */
667 h
->cache_blocks
= newBlocks
;
668 h
->cache_block_indices
= newIndices
;
669 h
->cache_num_blocks
= paramVal
;
671 CHM_RELEASE_LOCK(h
->cache_mutex
);
679 /* open an ITS archive */
680 struct chmFile
*chm_openW(const WCHAR
*filename
)
682 unsigned char sbuffer
[256];
683 unsigned int sremain
;
684 unsigned char *sbufpos
;
685 struct chmFile
*newHandle
=NULL
;
686 struct chmItsfHeader itsfHeader
;
687 struct chmItspHeader itspHeader
;
689 struct chmUnitInfo uiSpan
;
691 struct chmUnitInfo uiLzxc
;
692 struct chmLzxcControlData ctlData
;
694 /* allocate handle */
695 newHandle
= HeapAlloc(GetProcessHeap(), 0, sizeof(struct chmFile
));
696 newHandle
->fd
= CHM_NULL_FD
;
697 newHandle
->lzx_state
= NULL
;
698 newHandle
->cache_blocks
= NULL
;
699 newHandle
->cache_block_indices
= NULL
;
700 newHandle
->cache_num_blocks
= 0;
703 if ((newHandle
->fd
=CreateFileW(filename
,
708 FILE_ATTRIBUTE_NORMAL
,
709 NULL
)) == CHM_NULL_FD
)
711 HeapFree(GetProcessHeap(), 0, newHandle
);
715 /* initialize mutexes, if needed */
716 InitializeCriticalSection(&newHandle
->mutex
);
717 newHandle
->mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.mutex");
718 InitializeCriticalSection(&newHandle
->lzx_mutex
);
719 newHandle
->lzx_mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.lzx_mutex");
720 InitializeCriticalSection(&newHandle
->cache_mutex
);
721 newHandle
->cache_mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.cache_mutex");
723 /* read and verify header */
724 sremain
= _CHM_ITSF_V3_LEN
;
726 if (_chm_fetch_bytes(newHandle
, sbuffer
, 0, sremain
) != sremain
||
727 !_unmarshal_itsf_header(&sbufpos
, &sremain
, &itsfHeader
))
729 chm_close(newHandle
);
733 /* stash important values from header */
734 newHandle
->dir_offset
= itsfHeader
.dir_offset
;
735 newHandle
->dir_len
= itsfHeader
.dir_len
;
736 newHandle
->data_offset
= itsfHeader
.data_offset
;
738 /* now, read and verify the directory header chunk */
739 sremain
= _CHM_ITSP_V1_LEN
;
741 if (_chm_fetch_bytes(newHandle
, sbuffer
,
742 itsfHeader
.dir_offset
, sremain
) != sremain
||
743 !_unmarshal_itsp_header(&sbufpos
, &sremain
, &itspHeader
))
745 chm_close(newHandle
);
749 /* grab essential information from ITSP header */
750 newHandle
->dir_offset
+= itspHeader
.header_len
;
751 newHandle
->dir_len
-= itspHeader
.header_len
;
752 newHandle
->index_root
= itspHeader
.index_root
;
753 newHandle
->index_head
= itspHeader
.index_head
;
754 newHandle
->block_len
= itspHeader
.block_len
;
756 /* if the index root is -1, this means we don't have any PMGI blocks.
757 * as a result, we must use the sole PMGL block as the index root
759 if (newHandle
->index_root
== -1)
760 newHandle
->index_root
= newHandle
->index_head
;
762 /* initialize cache */
763 chm_set_param(newHandle
, CHM_PARAM_MAX_BLOCKS_CACHED
,
764 CHM_MAX_BLOCKS_CACHED
);
766 /* By default, compression is enabled. */
767 newHandle
->compression_enabled
= 1;
769 /* prefetch most commonly needed unit infos */
770 if (CHM_RESOLVE_SUCCESS
!= chm_resolve_object(newHandle
,
772 &newHandle
->rt_unit
) ||
773 newHandle
->rt_unit
.space
== CHM_COMPRESSED
||
774 CHM_RESOLVE_SUCCESS
!= chm_resolve_object(newHandle
,
776 &newHandle
->cn_unit
) ||
777 newHandle
->cn_unit
.space
== CHM_COMPRESSED
||
778 CHM_RESOLVE_SUCCESS
!= chm_resolve_object(newHandle
,
779 _CHMU_LZXC_CONTROLDATA
,
781 uiLzxc
.space
== CHM_COMPRESSED
)
783 newHandle
->compression_enabled
= 0;
786 /* read reset table info */
787 if (newHandle
->compression_enabled
)
789 sremain
= _CHM_LZXC_RESETTABLE_V1_LEN
;
791 if (chm_retrieve_object(newHandle
, &newHandle
->rt_unit
, sbuffer
,
792 0, sremain
) != sremain
||
793 !_unmarshal_lzxc_reset_table(&sbufpos
, &sremain
,
794 &newHandle
->reset_table
))
796 newHandle
->compression_enabled
= 0;
800 /* read control data */
801 if (newHandle
->compression_enabled
)
803 sremain
= (unsigned long)uiLzxc
.length
;
805 if (chm_retrieve_object(newHandle
, &uiLzxc
, sbuffer
,
806 0, sremain
) != sremain
||
807 !_unmarshal_lzxc_control_data(&sbufpos
, &sremain
,
810 newHandle
->compression_enabled
= 0;
813 newHandle
->window_size
= ctlData
.windowSize
;
814 newHandle
->reset_interval
= ctlData
.resetInterval
;
816 /* Jed, Mon Jun 28: Experimentally, it appears that the reset block count */
817 /* must be multiplied by this formerly unknown ctrl data field in */
818 /* order to decompress some files. */
820 newHandle
->reset_blkcount
= newHandle
->reset_interval
/
821 (newHandle
->window_size
/ 2);
823 newHandle
->reset_blkcount
= newHandle
->reset_interval
/
824 (newHandle
->window_size
/ 2) *
825 ctlData
.windowsPerReset
;
832 /* Duplicate an ITS archive handle */
833 struct chmFile
*chm_dup(struct chmFile
*oldHandle
)
835 struct chmFile
*newHandle
=NULL
;
837 newHandle
= HeapAlloc(GetProcessHeap(), 0, sizeof(struct chmFile
));
838 *newHandle
= *oldHandle
;
840 /* duplicate fd handle */
841 DuplicateHandle(GetCurrentProcess(), oldHandle
->fd
,
842 GetCurrentProcess(), &(newHandle
->fd
),
843 0, FALSE
, DUPLICATE_SAME_ACCESS
);
844 newHandle
->lzx_state
= NULL
;
845 newHandle
->cache_blocks
= NULL
;
846 newHandle
->cache_block_indices
= NULL
;
847 newHandle
->cache_num_blocks
= 0;
849 /* initialize mutexes, if needed */
850 InitializeCriticalSection(&newHandle
->mutex
);
851 newHandle
->mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.mutex");
852 InitializeCriticalSection(&newHandle
->lzx_mutex
);
853 newHandle
->lzx_mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.lzx_mutex");
854 InitializeCriticalSection(&newHandle
->cache_mutex
);
855 newHandle
->cache_mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.cache_mutex");
857 /* initialize cache */
858 chm_set_param(newHandle
, CHM_PARAM_MAX_BLOCKS_CACHED
,
859 CHM_MAX_BLOCKS_CACHED
);
864 /* close an ITS archive */
865 void chm_close(struct chmFile
*h
)
869 if (h
->fd
!= CHM_NULL_FD
)
870 CHM_CLOSE_FILE(h
->fd
);
873 h
->mutex
.DebugInfo
->Spare
[0] = 0;
874 DeleteCriticalSection(&h
->mutex
);
875 h
->lzx_mutex
.DebugInfo
->Spare
[0] = 0;
876 DeleteCriticalSection(&h
->lzx_mutex
);
877 h
->cache_mutex
.DebugInfo
->Spare
[0] = 0;
878 DeleteCriticalSection(&h
->cache_mutex
);
881 LZXteardown(h
->lzx_state
);
887 for (i
=0; i
<h
->cache_num_blocks
; i
++)
889 HeapFree(GetProcessHeap(), 0, h
->cache_blocks
[i
]);
891 HeapFree(GetProcessHeap(), 0, h
->cache_blocks
);
892 h
->cache_blocks
= NULL
;
895 HeapFree(GetProcessHeap(), 0, h
->cache_block_indices
);
896 h
->cache_block_indices
= NULL
;
898 HeapFree(GetProcessHeap(), 0, h
);
903 * helper methods for chm_resolve_object
906 /* skip a compressed dword */
907 static void _chm_skip_cword(UChar
**pEntry
)
909 while (*(*pEntry
)++ >= 0x80)
913 /* skip the data from a PMGL entry */
914 static void _chm_skip_PMGL_entry_data(UChar
**pEntry
)
916 _chm_skip_cword(pEntry
);
917 _chm_skip_cword(pEntry
);
918 _chm_skip_cword(pEntry
);
921 /* parse a compressed dword */
922 static UInt64
_chm_parse_cword(UChar
**pEntry
)
926 while ((temp
=*(*pEntry
)++) >= 0x80)
929 accum
+= temp
& 0x7f;
932 return (accum
<< 7) + temp
;
935 /* parse a utf-8 string into an ASCII char buffer */
936 static int _chm_parse_UTF8(UChar
**pEntry
, UInt64 count
, WCHAR
*path
)
938 /* MJM - Modified to return real Unicode strings */
941 *path
++ = (*(*pEntry
)++);
949 /* parse a PMGL entry into a chmUnitInfo struct; return 1 on success. */
950 static int _chm_parse_PMGL_entry(UChar
**pEntry
, struct chmUnitInfo
*ui
)
955 strLen
= _chm_parse_cword(pEntry
);
956 if (strLen
> CHM_MAX_PATHLEN
)
960 if (! _chm_parse_UTF8(pEntry
, strLen
, ui
->path
))
964 ui
->space
= (int)_chm_parse_cword(pEntry
);
965 ui
->start
= _chm_parse_cword(pEntry
);
966 ui
->length
= _chm_parse_cword(pEntry
);
970 /* find an exact entry in PMGL; return NULL if we fail */
971 static UChar
*_chm_find_in_PMGL(UChar
*page_buf
,
973 const WCHAR
*objPath
)
975 /* XXX: modify this to do a binary search using the nice index structure
976 * that is provided for us.
978 struct chmPmglHeader header
;
984 WCHAR buffer
[CHM_MAX_PATHLEN
+1];
986 /* figure out where to start and end */
988 hremain
= _CHM_PMGL_LEN
;
989 if (! _unmarshal_pmgl_header(&cur
, &hremain
, &header
))
991 end
= page_buf
+ block_len
- (header
.free_space
);
993 /* now, scan progressively */
998 strLen
= _chm_parse_cword(&cur
);
999 if (! _chm_parse_UTF8(&cur
, strLen
, buffer
))
1002 /* check if it is the right name */
1003 if (! strcmpiW(buffer
, objPath
))
1006 _chm_skip_PMGL_entry_data(&cur
);
1012 /* find which block should be searched next for the entry; -1 if no block */
1013 static Int32
_chm_find_in_PMGI(UChar
*page_buf
,
1015 const WCHAR
*objPath
)
1017 /* XXX: modify this to do a binary search using the nice index structure
1018 * that is provided for us
1020 struct chmPmgiHeader header
;
1026 WCHAR buffer
[CHM_MAX_PATHLEN
+1];
1028 /* figure out where to start and end */
1030 hremain
= _CHM_PMGI_LEN
;
1031 if (! _unmarshal_pmgi_header(&cur
, &hremain
, &header
))
1033 end
= page_buf
+ block_len
- (header
.free_space
);
1035 /* now, scan progressively */
1039 strLen
= _chm_parse_cword(&cur
);
1040 if (! _chm_parse_UTF8(&cur
, strLen
, buffer
))
1043 /* check if it is the right name */
1044 if (strcmpiW(buffer
, objPath
) > 0)
1047 /* load next value for path */
1048 page
= (int)_chm_parse_cword(&cur
);
1054 /* resolve a particular object from the archive */
1055 int chm_resolve_object(struct chmFile
*h
,
1056 const WCHAR
*objPath
,
1057 struct chmUnitInfo
*ui
)
1060 * XXX: implement caching scheme for dir pages
1065 /* buffer to hold whatever page we're looking at */
1066 UChar
*page_buf
= HeapAlloc(GetProcessHeap(), 0, h
->block_len
);
1069 curPage
= h
->index_root
;
1071 /* until we have either returned or given up */
1072 while (curPage
!= -1)
1075 /* try to fetch the index page */
1076 if (_chm_fetch_bytes(h
, page_buf
,
1077 h
->dir_offset
+ (UInt64
)curPage
*h
->block_len
,
1078 h
->block_len
) != h
->block_len
)
1080 HeapFree(GetProcessHeap(), 0, page_buf
);
1081 return CHM_RESOLVE_FAILURE
;
1084 /* now, if it is a leaf node: */
1085 if (memcmp(page_buf
, _chm_pmgl_marker
, 4) == 0)
1088 UChar
*pEntry
= _chm_find_in_PMGL(page_buf
,
1093 HeapFree(GetProcessHeap(), 0, page_buf
);
1094 return CHM_RESOLVE_FAILURE
;
1097 /* parse entry and return */
1098 _chm_parse_PMGL_entry(&pEntry
, ui
);
1099 HeapFree(GetProcessHeap(), 0, page_buf
);
1100 return CHM_RESOLVE_SUCCESS
;
1103 /* else, if it is a branch node: */
1104 else if (memcmp(page_buf
, _chm_pmgi_marker
, 4) == 0)
1105 curPage
= _chm_find_in_PMGI(page_buf
, h
->block_len
, objPath
);
1107 /* else, we are confused. give up. */
1110 HeapFree(GetProcessHeap(), 0, page_buf
);
1111 return CHM_RESOLVE_FAILURE
;
1115 /* didn't find anything. fail. */
1116 HeapFree(GetProcessHeap(), 0, page_buf
);
1117 return CHM_RESOLVE_FAILURE
;
1121 * utility methods for dealing with compressed data
1124 /* get the bounds of a compressed block. return 0 on failure */
1125 static int _chm_get_cmpblock_bounds(struct chmFile
*h
,
1130 UChar buffer
[8], *dummy
;
1133 /* for all but the last block, use the reset table */
1134 if (block
< h
->reset_table
.block_count
-1)
1136 /* unpack the start address */
1139 if (_chm_fetch_bytes(h
, buffer
,
1142 + h
->reset_table
.table_offset
1144 remain
) != remain
||
1145 !_unmarshal_uint64(&dummy
, &remain
, start
))
1148 /* unpack the end address */
1151 if (_chm_fetch_bytes(h
, buffer
,
1154 + h
->reset_table
.table_offset
1156 remain
) != remain
||
1157 !_unmarshal_int64(&dummy
, &remain
, len
))
1161 /* for the last block, use the span in addition to the reset table */
1164 /* unpack the start address */
1167 if (_chm_fetch_bytes(h
, buffer
,
1170 + h
->reset_table
.table_offset
1172 remain
) != remain
||
1173 !_unmarshal_uint64(&dummy
, &remain
, start
))
1176 *len
= h
->reset_table
.compressed_len
;
1179 /* compute the length and absolute start address */
1181 *start
+= h
->data_offset
+ h
->cn_unit
.start
;
1186 /* decompress the block. must have lzx_mutex. */
1187 static Int64
_chm_decompress_block(struct chmFile
*h
,
1191 UChar
*cbuffer
= HeapAlloc( GetProcessHeap(), 0,
1192 ((unsigned int)h
->reset_table
.block_len
+ 6144));
1193 UInt64 cmpStart
; /* compressed start */
1194 Int64 cmpLen
; /* compressed len */
1195 int indexSlot
; /* cache index slot */
1196 UChar
*lbuffer
; /* local buffer ptr */
1197 UInt32 blockAlign
= (UInt32
)(block
% h
->reset_blkcount
); /* reset interval align */
1198 UInt32 i
; /* local loop index */
1200 /* let the caching system pull its weight! */
1201 if (block
- blockAlign
<= h
->lzx_last_block
&&
1202 block
>= h
->lzx_last_block
)
1203 blockAlign
= (block
- h
->lzx_last_block
);
1205 /* check if we need previous blocks */
1206 if (blockAlign
!= 0)
1208 /* fetch all required previous blocks since last reset */
1209 for (i
= blockAlign
; i
> 0; i
--)
1211 UInt32 curBlockIdx
= block
- i
;
1213 /* check if we most recently decompressed the previous block */
1214 if (h
->lzx_last_block
!= curBlockIdx
)
1216 if ((curBlockIdx
% h
->reset_blkcount
) == 0)
1219 fprintf(stderr
, "***RESET (1)***\n");
1221 LZXreset(h
->lzx_state
);
1224 indexSlot
= (int)((curBlockIdx
) % h
->cache_num_blocks
);
1225 h
->cache_block_indices
[indexSlot
] = curBlockIdx
;
1226 if (! h
->cache_blocks
[indexSlot
])
1227 h
->cache_blocks
[indexSlot
] =
1228 HeapAlloc(GetProcessHeap(), 0,
1229 (unsigned int)(h
->reset_table
.block_len
));
1230 lbuffer
= h
->cache_blocks
[indexSlot
];
1232 /* decompress the previous block */
1234 fprintf(stderr
, "Decompressing block #%4d (EXTRA)\n", curBlockIdx
);
1236 if (!_chm_get_cmpblock_bounds(h
, curBlockIdx
, &cmpStart
, &cmpLen
) ||
1237 _chm_fetch_bytes(h
, cbuffer
, cmpStart
, cmpLen
) != cmpLen
||
1238 LZXdecompress(h
->lzx_state
, cbuffer
, lbuffer
, (int)cmpLen
,
1239 (int)h
->reset_table
.block_len
) != DECR_OK
)
1242 fprintf(stderr
, " (DECOMPRESS FAILED!)\n");
1244 HeapFree(GetProcessHeap(), 0, cbuffer
);
1248 h
->lzx_last_block
= (int)curBlockIdx
;
1254 if ((block
% h
->reset_blkcount
) == 0)
1257 fprintf(stderr
, "***RESET (2)***\n");
1259 LZXreset(h
->lzx_state
);
1263 /* allocate slot in cache */
1264 indexSlot
= (int)(block
% h
->cache_num_blocks
);
1265 h
->cache_block_indices
[indexSlot
] = block
;
1266 if (! h
->cache_blocks
[indexSlot
])
1267 h
->cache_blocks
[indexSlot
] =
1268 HeapAlloc(GetProcessHeap(), 0, ((unsigned int)h
->reset_table
.block_len
));
1269 lbuffer
= h
->cache_blocks
[indexSlot
];
1272 /* decompress the block we actually want */
1274 fprintf(stderr
, "Decompressing block #%4d (REAL )\n", block
);
1276 if (! _chm_get_cmpblock_bounds(h
, block
, &cmpStart
, &cmpLen
) ||
1277 _chm_fetch_bytes(h
, cbuffer
, cmpStart
, cmpLen
) != cmpLen
||
1278 LZXdecompress(h
->lzx_state
, cbuffer
, lbuffer
, (int)cmpLen
,
1279 (int)h
->reset_table
.block_len
) != DECR_OK
)
1282 fprintf(stderr
, " (DECOMPRESS FAILED!)\n");
1284 HeapFree(GetProcessHeap(), 0, cbuffer
);
1287 h
->lzx_last_block
= (int)block
;
1289 /* XXX: modify LZX routines to return the length of the data they
1290 * decompressed and return that instead, for an extra sanity check.
1292 HeapFree(GetProcessHeap(), 0, cbuffer
);
1293 return h
->reset_table
.block_len
;
1296 /* grab a region from a compressed block */
1297 static Int64
_chm_decompress_region(struct chmFile
*h
,
1302 UInt64 nBlock
, nOffset
;
1305 UChar
*ubuffer
= NULL
;
1310 /* figure out what we need to read */
1311 nBlock
= start
/ h
->reset_table
.block_len
;
1312 nOffset
= start
% h
->reset_table
.block_len
;
1314 if (nLen
> (h
->reset_table
.block_len
- nOffset
))
1315 nLen
= h
->reset_table
.block_len
- nOffset
;
1317 /* if block is cached, return data from it. */
1318 CHM_ACQUIRE_LOCK(h
->lzx_mutex
);
1319 CHM_ACQUIRE_LOCK(h
->cache_mutex
);
1320 if (h
->cache_block_indices
[nBlock
% h
->cache_num_blocks
] == nBlock
&&
1321 h
->cache_blocks
[nBlock
% h
->cache_num_blocks
] != NULL
)
1324 h
->cache_blocks
[nBlock
% h
->cache_num_blocks
] + nOffset
,
1325 (unsigned int)nLen
);
1326 CHM_RELEASE_LOCK(h
->cache_mutex
);
1327 CHM_RELEASE_LOCK(h
->lzx_mutex
);
1330 CHM_RELEASE_LOCK(h
->cache_mutex
);
1332 /* data request not satisfied, so... start up the decompressor machine */
1335 int window_size
= ffs(h
->window_size
) - 1;
1336 h
->lzx_last_block
= -1;
1337 h
->lzx_state
= LZXinit(window_size
);
1340 /* decompress some data */
1341 gotLen
= _chm_decompress_block(h
, nBlock
, &ubuffer
);
1344 memcpy(buf
, ubuffer
+nOffset
, (unsigned int)nLen
);
1345 CHM_RELEASE_LOCK(h
->lzx_mutex
);
1349 /* retrieve (part of) an object */
1350 LONGINT64
chm_retrieve_object(struct chmFile
*h
,
1351 struct chmUnitInfo
*ui
,
1356 /* must be valid file handle */
1360 /* starting address must be in correct range */
1361 if (addr
>= ui
->length
)
1365 if (addr
+ len
> ui
->length
)
1366 len
= ui
->length
- addr
;
1368 /* if the file is uncompressed, it's simple */
1369 if (ui
->space
== CHM_UNCOMPRESSED
)
1372 return _chm_fetch_bytes(h
,
1374 h
->data_offset
+ ui
->start
+ addr
,
1378 /* else if the file is compressed, it's a little trickier */
1379 else /* ui->space == CHM_COMPRESSED */
1381 Int64 swath
=0, total
=0;
1383 /* if compression is not enabled for this file... */
1384 if (! h
->compression_enabled
)
1389 /* swill another mouthful */
1390 swath
= _chm_decompress_region(h
, buf
, ui
->start
+ addr
, len
);
1392 /* if we didn't get any... */
1408 int chm_enumerate_dir(struct chmFile
*h
,
1409 const WCHAR
*prefix
,
1415 * XXX: do this efficiently (i.e. using the tree index)
1420 /* buffer to hold whatever page we're looking at */
1421 UChar
*page_buf
= HeapAlloc(GetProcessHeap(), 0, h
->block_len
);
1422 struct chmPmglHeader header
;
1425 unsigned int lenRemain
;
1427 /* set to 1 once we've started */
1430 /* the current ui */
1431 struct chmUnitInfo ui
;
1435 /* the length of the prefix */
1436 WCHAR prefixRectified
[CHM_MAX_PATHLEN
+1];
1438 WCHAR lastPath
[CHM_MAX_PATHLEN
];
1442 curPage
= h
->index_head
;
1444 /* initialize pathname state */
1445 lstrcpynW(prefixRectified
, prefix
, CHM_MAX_PATHLEN
);
1446 prefixLen
= strlenW(prefixRectified
);
1449 if (prefixRectified
[prefixLen
-1] != '/')
1451 prefixRectified
[prefixLen
] = '/';
1452 prefixRectified
[prefixLen
+1] = '\0';
1459 /* until we have either returned or given up */
1460 while (curPage
!= -1)
1463 /* try to fetch the index page */
1464 if (_chm_fetch_bytes(h
,
1466 h
->dir_offset
+ (UInt64
)curPage
*h
->block_len
,
1467 h
->block_len
) != h
->block_len
)
1469 HeapFree(GetProcessHeap(), 0, page_buf
);
1473 /* figure out start and end for this page */
1475 lenRemain
= _CHM_PMGL_LEN
;
1476 if (! _unmarshal_pmgl_header(&cur
, &lenRemain
, &header
))
1478 HeapFree(GetProcessHeap(), 0, page_buf
);
1481 end
= page_buf
+ h
->block_len
- (header
.free_space
);
1483 /* loop over this page */
1486 if (! _chm_parse_PMGL_entry(&cur
, &ui
))
1488 HeapFree(GetProcessHeap(), 0, page_buf
);
1492 /* check if we should start */
1495 if (ui
.length
== 0 && strncmpiW(ui
.path
, prefixRectified
, prefixLen
) == 0)
1500 if (ui
.path
[prefixLen
] == '\0')
1504 /* check if we should stop */
1507 if (strncmpiW(ui
.path
, prefixRectified
, prefixLen
) != 0)
1509 HeapFree(GetProcessHeap(), 0, page_buf
);
1514 /* check if we should include this path */
1515 if (lastPathLen
!= -1)
1517 if (strncmpiW(ui
.path
, lastPath
, lastPathLen
) == 0)
1520 strcpyW(lastPath
, ui
.path
);
1521 lastPathLen
= strlenW(lastPath
);
1523 /* get the length of the path */
1524 ui_path_len
= strlenW(ui
.path
)-1;
1526 /* check for DIRS */
1527 if (ui
.path
[ui_path_len
] == '/' && !(what
& CHM_ENUMERATE_DIRS
))
1530 /* check for FILES */
1531 if (ui
.path
[ui_path_len
] != '/' && !(what
& CHM_ENUMERATE_FILES
))
1534 /* check for NORMAL vs. META */
1535 if (ui
.path
[0] == '/')
1538 /* check for NORMAL vs. SPECIAL */
1539 if (ui
.path
[1] == '#' || ui
.path
[1] == '$')
1540 flag
= CHM_ENUMERATE_SPECIAL
;
1542 flag
= CHM_ENUMERATE_NORMAL
;
1545 flag
= CHM_ENUMERATE_META
;
1546 if (! (what
& flag
))
1549 /* call the enumerator */
1551 int status
= (*e
)(h
, &ui
, context
);
1554 case CHM_ENUMERATOR_FAILURE
:
1555 HeapFree(GetProcessHeap(), 0, page_buf
);
1557 case CHM_ENUMERATOR_CONTINUE
:
1559 case CHM_ENUMERATOR_SUCCESS
:
1560 HeapFree(GetProcessHeap(), 0, page_buf
);
1568 /* advance to next page */
1569 curPage
= header
.block_next
;
1572 HeapFree(GetProcessHeap(), 0, page_buf
);