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 modify *
36 * it under the terms of the GNU Lesser General Public License as *
37 * published by the Free Software Foundation; either version 2.1 of the *
38 * License, or (at your option) any later version. *
40 ***************************************************************************/
42 /***************************************************************************
44 * Adapted for Wine by Mike McCormack *
46 ***************************************************************************/
49 #include "wine/port.h"
58 #include "wine/unicode.h"
63 #define CHM_ACQUIRE_LOCK(a) do { \
64 EnterCriticalSection(&(a)); \
66 #define CHM_RELEASE_LOCK(a) do { \
67 LeaveCriticalSection(&(a)); \
70 #define CHM_NULL_FD (INVALID_HANDLE_VALUE)
71 #define CHM_CLOSE_FILE(fd) CloseHandle((fd))
74 * defines related to tuning
76 #ifndef CHM_MAX_BLOCKS_CACHED
77 #define CHM_MAX_BLOCKS_CACHED 5
79 #define CHM_PARAM_MAX_BLOCKS_CACHED 0
82 * architecture specific defines
84 * Note: as soon as C99 is more widespread, the below defines should
85 * probably just use the C99 sized-int types.
87 * The following settings will probably work for many platforms. The sizes
88 * don't have to be exactly correct, but the types must accommodate at least as
89 * many bits as they specify.
92 /* i386, 32-bit, Windows */
95 typedef USHORT UInt16
;
98 typedef LONGLONG Int64
;
99 typedef ULONGLONG UInt64
;
101 /* utilities for unmarshalling data */
102 static int _unmarshal_char_array(unsigned char **pData
,
103 unsigned int *pLenRemain
,
107 if (count
<= 0 || (unsigned int)count
> *pLenRemain
)
109 memcpy(dest
, (*pData
), count
);
111 *pLenRemain
-= count
;
115 static int _unmarshal_uchar_array(unsigned char **pData
,
116 unsigned int *pLenRemain
,
120 if (count
<= 0 || (unsigned int)count
> *pLenRemain
)
122 memcpy(dest
, (*pData
), count
);
124 *pLenRemain
-= count
;
128 static int _unmarshal_int32(unsigned char **pData
,
129 unsigned int *pLenRemain
,
134 *dest
= (*pData
)[0] | (*pData
)[1]<<8 | (*pData
)[2]<<16 | (*pData
)[3]<<24;
140 static int _unmarshal_uint32(unsigned char **pData
,
141 unsigned int *pLenRemain
,
146 *dest
= (*pData
)[0] | (*pData
)[1]<<8 | (*pData
)[2]<<16 | (*pData
)[3]<<24;
152 static int _unmarshal_int64(unsigned char **pData
,
153 unsigned int *pLenRemain
,
164 temp
|= (*pData
)[i
-1];
172 static int _unmarshal_uint64(unsigned char **pData
,
173 unsigned int *pLenRemain
,
184 temp
|= (*pData
)[i
-1];
192 static int _unmarshal_uuid(unsigned char **pData
,
193 unsigned int *pDataLen
,
196 return _unmarshal_uchar_array(pData
, pDataLen
, dest
, 16);
199 /* names of sections essential to decompression */
200 static const WCHAR _CHMU_RESET_TABLE
[] = {
201 ':',':','D','a','t','a','S','p','a','c','e','/',
202 'S','t','o','r','a','g','e','/',
203 'M','S','C','o','m','p','r','e','s','s','e','d','/',
204 'T','r','a','n','s','f','o','r','m','/',
205 '{','7','F','C','2','8','9','4','0','-','9','D','3','1',
206 '-','1','1','D','0','-','9','B','2','7','-',
207 '0','0','A','0','C','9','1','E','9','C','7','C','}','/',
208 'I','n','s','t','a','n','c','e','D','a','t','a','/',
209 'R','e','s','e','t','T','a','b','l','e',0
211 static const WCHAR _CHMU_LZXC_CONTROLDATA
[] = {
212 ':',':','D','a','t','a','S','p','a','c','e','/',
213 'S','t','o','r','a','g','e','/',
214 'M','S','C','o','m','p','r','e','s','s','e','d','/',
215 'C','o','n','t','r','o','l','D','a','t','a',0
217 static const WCHAR _CHMU_CONTENT
[] = {
218 ':',':','D','a','t','a','S','p','a','c','e','/',
219 'S','t','o','r','a','g','e','/',
220 'M','S','C','o','m','p','r','e','s','s','e','d','/',
221 'C','o','n','t','e','n','t',0
225 * structures local to this module
228 /* structure of ITSF headers */
229 #define _CHM_ITSF_V2_LEN (0x58)
230 #define _CHM_ITSF_V3_LEN (0x60)
233 char signature
[4]; /* 0 (ITSF) */
234 Int32 version
; /* 4 */
235 Int32 header_len
; /* 8 */
236 Int32 unknown_000c
; /* c */
237 UInt32 last_modified
; /* 10 */
238 UInt32 lang_id
; /* 14 */
239 UChar dir_uuid
[16]; /* 18 */
240 UChar stream_uuid
[16]; /* 28 */
241 UInt64 unknown_offset
; /* 38 */
242 UInt64 unknown_len
; /* 40 */
243 UInt64 dir_offset
; /* 48 */
244 UInt64 dir_len
; /* 50 */
245 UInt64 data_offset
; /* 58 (Not present before V3) */
246 }; /* __attribute__ ((aligned (1))); */
248 static int _unmarshal_itsf_header(unsigned char **pData
,
249 unsigned int *pDataLen
,
250 struct chmItsfHeader
*dest
)
252 /* we only know how to deal with the 0x58 and 0x60 byte structures */
253 if (*pDataLen
!= _CHM_ITSF_V2_LEN
&& *pDataLen
!= _CHM_ITSF_V3_LEN
)
256 /* unmarshal common fields */
257 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
258 _unmarshal_int32 (pData
, pDataLen
, &dest
->version
);
259 _unmarshal_int32 (pData
, pDataLen
, &dest
->header_len
);
260 _unmarshal_int32 (pData
, pDataLen
, &dest
->unknown_000c
);
261 _unmarshal_uint32 (pData
, pDataLen
, &dest
->last_modified
);
262 _unmarshal_uint32 (pData
, pDataLen
, &dest
->lang_id
);
263 _unmarshal_uuid (pData
, pDataLen
, dest
->dir_uuid
);
264 _unmarshal_uuid (pData
, pDataLen
, dest
->stream_uuid
);
265 _unmarshal_uint64 (pData
, pDataLen
, &dest
->unknown_offset
);
266 _unmarshal_uint64 (pData
, pDataLen
, &dest
->unknown_len
);
267 _unmarshal_uint64 (pData
, pDataLen
, &dest
->dir_offset
);
268 _unmarshal_uint64 (pData
, pDataLen
, &dest
->dir_len
);
270 /* error check the data */
271 /* XXX: should also check UUIDs, probably, though with a version 3 file,
272 * current MS tools do not seem to use them.
274 if (memcmp(dest
->signature
, "ITSF", 4) != 0)
276 if (dest
->version
== 2)
278 if (dest
->header_len
< _CHM_ITSF_V2_LEN
)
281 else if (dest
->version
== 3)
283 if (dest
->header_len
< _CHM_ITSF_V3_LEN
)
289 /* now, if we have a V3 structure, unmarshal the rest.
290 * otherwise, compute it
292 if (dest
->version
== 3)
295 _unmarshal_uint64(pData
, pDataLen
, &dest
->data_offset
);
300 dest
->data_offset
= dest
->dir_offset
+ dest
->dir_len
;
305 /* structure of ITSP headers */
306 #define _CHM_ITSP_V1_LEN (0x54)
309 char signature
[4]; /* 0 (ITSP) */
310 Int32 version
; /* 4 */
311 Int32 header_len
; /* 8 */
312 Int32 unknown_000c
; /* c */
313 UInt32 block_len
; /* 10 */
314 Int32 blockidx_intvl
; /* 14 */
315 Int32 index_depth
; /* 18 */
316 Int32 index_root
; /* 1c */
317 Int32 index_head
; /* 20 */
318 Int32 unknown_0024
; /* 24 */
319 UInt32 num_blocks
; /* 28 */
320 Int32 unknown_002c
; /* 2c */
321 UInt32 lang_id
; /* 30 */
322 UChar system_uuid
[16]; /* 34 */
323 UChar unknown_0044
[16]; /* 44 */
324 }; /* __attribute__ ((aligned (1))); */
326 static int _unmarshal_itsp_header(unsigned char **pData
,
327 unsigned int *pDataLen
,
328 struct chmItspHeader
*dest
)
330 /* we only know how to deal with a 0x54 byte structures */
331 if (*pDataLen
!= _CHM_ITSP_V1_LEN
)
334 /* unmarshal fields */
335 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
336 _unmarshal_int32 (pData
, pDataLen
, &dest
->version
);
337 _unmarshal_int32 (pData
, pDataLen
, &dest
->header_len
);
338 _unmarshal_int32 (pData
, pDataLen
, &dest
->unknown_000c
);
339 _unmarshal_uint32 (pData
, pDataLen
, &dest
->block_len
);
340 _unmarshal_int32 (pData
, pDataLen
, &dest
->blockidx_intvl
);
341 _unmarshal_int32 (pData
, pDataLen
, &dest
->index_depth
);
342 _unmarshal_int32 (pData
, pDataLen
, &dest
->index_root
);
343 _unmarshal_int32 (pData
, pDataLen
, &dest
->index_head
);
344 _unmarshal_int32 (pData
, pDataLen
, &dest
->unknown_0024
);
345 _unmarshal_uint32 (pData
, pDataLen
, &dest
->num_blocks
);
346 _unmarshal_int32 (pData
, pDataLen
, &dest
->unknown_002c
);
347 _unmarshal_uint32 (pData
, pDataLen
, &dest
->lang_id
);
348 _unmarshal_uuid (pData
, pDataLen
, dest
->system_uuid
);
349 _unmarshal_uchar_array(pData
, pDataLen
, dest
->unknown_0044
, 16);
351 /* error check the data */
352 if (memcmp(dest
->signature
, "ITSP", 4) != 0)
354 if (dest
->version
!= 1)
356 if (dest
->header_len
!= _CHM_ITSP_V1_LEN
)
362 /* structure of PMGL headers */
363 static const char _chm_pmgl_marker
[4] = "PMGL";
364 #define _CHM_PMGL_LEN (0x14)
367 char signature
[4]; /* 0 (PMGL) */
368 UInt32 free_space
; /* 4 */
369 UInt32 unknown_0008
; /* 8 */
370 Int32 block_prev
; /* c */
371 Int32 block_next
; /* 10 */
372 }; /* __attribute__ ((aligned (1))); */
374 static int _unmarshal_pmgl_header(unsigned char **pData
,
375 unsigned int *pDataLen
,
376 struct chmPmglHeader
*dest
)
378 /* we only know how to deal with a 0x14 byte structures */
379 if (*pDataLen
!= _CHM_PMGL_LEN
)
382 /* unmarshal fields */
383 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
384 _unmarshal_uint32 (pData
, pDataLen
, &dest
->free_space
);
385 _unmarshal_uint32 (pData
, pDataLen
, &dest
->unknown_0008
);
386 _unmarshal_int32 (pData
, pDataLen
, &dest
->block_prev
);
387 _unmarshal_int32 (pData
, pDataLen
, &dest
->block_next
);
389 /* check structure */
390 if (memcmp(dest
->signature
, _chm_pmgl_marker
, 4) != 0)
396 /* structure of PMGI headers */
397 static const char _chm_pmgi_marker
[4] = "PMGI";
398 #define _CHM_PMGI_LEN (0x08)
401 char signature
[4]; /* 0 (PMGI) */
402 UInt32 free_space
; /* 4 */
403 }; /* __attribute__ ((aligned (1))); */
405 static int _unmarshal_pmgi_header(unsigned char **pData
,
406 unsigned int *pDataLen
,
407 struct chmPmgiHeader
*dest
)
409 /* we only know how to deal with a 0x8 byte structures */
410 if (*pDataLen
!= _CHM_PMGI_LEN
)
413 /* unmarshal fields */
414 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
415 _unmarshal_uint32 (pData
, pDataLen
, &dest
->free_space
);
417 /* check structure */
418 if (memcmp(dest
->signature
, _chm_pmgi_marker
, 4) != 0)
424 /* structure of LZXC reset table */
425 #define _CHM_LZXC_RESETTABLE_V1_LEN (0x28)
426 struct chmLzxcResetTable
432 UInt64 uncompressed_len
;
433 UInt64 compressed_len
;
435 }; /* __attribute__ ((aligned (1))); */
437 static int _unmarshal_lzxc_reset_table(unsigned char **pData
,
438 unsigned int *pDataLen
,
439 struct chmLzxcResetTable
*dest
)
441 /* we only know how to deal with a 0x28 byte structures */
442 if (*pDataLen
!= _CHM_LZXC_RESETTABLE_V1_LEN
)
445 /* unmarshal fields */
446 _unmarshal_uint32 (pData
, pDataLen
, &dest
->version
);
447 _unmarshal_uint32 (pData
, pDataLen
, &dest
->block_count
);
448 _unmarshal_uint32 (pData
, pDataLen
, &dest
->unknown
);
449 _unmarshal_uint32 (pData
, pDataLen
, &dest
->table_offset
);
450 _unmarshal_uint64 (pData
, pDataLen
, &dest
->uncompressed_len
);
451 _unmarshal_uint64 (pData
, pDataLen
, &dest
->compressed_len
);
452 _unmarshal_uint64 (pData
, pDataLen
, &dest
->block_len
);
454 /* check structure */
455 if (dest
->version
!= 2)
461 /* structure of LZXC control data block */
462 #define _CHM_LZXC_MIN_LEN (0x18)
463 #define _CHM_LZXC_V2_LEN (0x1c)
464 struct chmLzxcControlData
467 char signature
[4]; /* 4 (LZXC) */
468 UInt32 version
; /* 8 */
469 UInt32 resetInterval
; /* c */
470 UInt32 windowSize
; /* 10 */
471 UInt32 windowsPerReset
; /* 14 */
472 UInt32 unknown_18
; /* 18 */
475 static int _unmarshal_lzxc_control_data(unsigned char **pData
,
476 unsigned int *pDataLen
,
477 struct chmLzxcControlData
*dest
)
479 /* we want at least 0x18 bytes */
480 if (*pDataLen
< _CHM_LZXC_MIN_LEN
)
483 /* unmarshal fields */
484 _unmarshal_uint32 (pData
, pDataLen
, &dest
->size
);
485 _unmarshal_char_array(pData
, pDataLen
, dest
->signature
, 4);
486 _unmarshal_uint32 (pData
, pDataLen
, &dest
->version
);
487 _unmarshal_uint32 (pData
, pDataLen
, &dest
->resetInterval
);
488 _unmarshal_uint32 (pData
, pDataLen
, &dest
->windowSize
);
489 _unmarshal_uint32 (pData
, pDataLen
, &dest
->windowsPerReset
);
491 if (*pDataLen
>= _CHM_LZXC_V2_LEN
)
492 _unmarshal_uint32 (pData
, pDataLen
, &dest
->unknown_18
);
494 dest
->unknown_18
= 0;
496 if (dest
->version
== 2)
498 dest
->resetInterval
*= 0x8000;
499 dest
->windowSize
*= 0x8000;
501 if (dest
->windowSize
== 0 || dest
->resetInterval
== 0)
504 /* for now, only support resetInterval a multiple of windowSize/2 */
505 if (dest
->windowSize
== 1)
507 if ((dest
->resetInterval
% (dest
->windowSize
/2)) != 0)
510 /* check structure */
511 if (memcmp(dest
->signature
, "LZXC", 4) != 0)
517 /* the structure used for chm file handles */
522 CRITICAL_SECTION mutex
;
523 CRITICAL_SECTION lzx_mutex
;
524 CRITICAL_SECTION cache_mutex
;
534 struct chmUnitInfo rt_unit
;
535 struct chmUnitInfo cn_unit
;
536 struct chmLzxcResetTable reset_table
;
538 /* LZX control data */
539 int compression_enabled
;
541 UInt32 reset_interval
;
542 UInt32 reset_blkcount
;
544 /* decompressor state */
545 struct LZXstate
*lzx_state
;
548 /* cache for decompressed blocks */
549 UChar
**cache_blocks
;
550 Int64
*cache_block_indices
;
551 Int32 cache_num_blocks
;
555 * utility functions local to this module
558 /* utility function to handle differences between {pread,read}(64)? */
559 static Int64
_chm_fetch_bytes(struct chmFile
*h
,
565 if (h
->fd
== CHM_NULL_FD
)
568 CHM_ACQUIRE_LOCK(h
->mutex
);
569 /* NOTE: this might be better done with CreateFileMapping, et cetera... */
571 LARGE_INTEGER old_pos
, new_pos
;
574 /* awkward Win32 Seek/Tell */
575 new_pos
.QuadPart
= 0;
576 SetFilePointerEx( h
->fd
, new_pos
, &old_pos
, FILE_CURRENT
);
577 new_pos
.QuadPart
= os
;
578 SetFilePointerEx( h
->fd
, new_pos
, NULL
, FILE_BEGIN
);
590 /* restore original position */
591 SetFilePointerEx( h
->fd
, old_pos
, NULL
, FILE_BEGIN
);
593 CHM_RELEASE_LOCK(h
->mutex
);
598 * set a parameter on the file handle.
599 * valid parameter types:
600 * CHM_PARAM_MAX_BLOCKS_CACHED:
601 * how many decompressed blocks should be cached? A simple
602 * caching scheme is used, wherein the index of the block is
603 * used as a hash value, and hash collision results in the
604 * invalidation of the previously cached block.
606 static void chm_set_param(struct chmFile
*h
,
612 case CHM_PARAM_MAX_BLOCKS_CACHED
:
613 CHM_ACQUIRE_LOCK(h
->cache_mutex
);
614 if (paramVal
!= h
->cache_num_blocks
)
620 /* allocate new cached blocks */
621 newBlocks
= malloc(paramVal
* sizeof (UChar
*));
622 newIndices
= malloc(paramVal
* sizeof (UInt64
));
623 for (i
=0; i
<paramVal
; i
++)
629 /* re-distribute old cached blocks */
632 for (i
=0; i
<h
->cache_num_blocks
; i
++)
634 int newSlot
= (int)(h
->cache_block_indices
[i
] % paramVal
);
636 if (h
->cache_blocks
[i
])
638 /* in case of collision, destroy newcomer */
639 if (newBlocks
[newSlot
])
641 free(h
->cache_blocks
[i
]);
642 h
->cache_blocks
[i
] = NULL
;
646 newBlocks
[newSlot
] = h
->cache_blocks
[i
];
647 newIndices
[newSlot
] =
648 h
->cache_block_indices
[i
];
653 free(h
->cache_blocks
);
654 free(h
->cache_block_indices
);
657 /* now, set new values */
658 h
->cache_blocks
= newBlocks
;
659 h
->cache_block_indices
= newIndices
;
660 h
->cache_num_blocks
= paramVal
;
662 CHM_RELEASE_LOCK(h
->cache_mutex
);
670 /* open an ITS archive */
671 struct chmFile
*chm_openW(const WCHAR
*filename
)
673 unsigned char sbuffer
[256];
674 unsigned int sremain
;
675 unsigned char *sbufpos
;
676 struct chmFile
*newHandle
=NULL
;
677 struct chmItsfHeader itsfHeader
;
678 struct chmItspHeader itspHeader
;
680 struct chmUnitInfo uiSpan
;
682 struct chmUnitInfo uiLzxc
;
683 struct chmLzxcControlData ctlData
;
685 /* allocate handle */
686 newHandle
= malloc(sizeof(struct chmFile
));
687 newHandle
->fd
= CHM_NULL_FD
;
688 newHandle
->lzx_state
= NULL
;
689 newHandle
->cache_blocks
= NULL
;
690 newHandle
->cache_block_indices
= NULL
;
691 newHandle
->cache_num_blocks
= 0;
694 if ((newHandle
->fd
=CreateFileW(filename
,
699 FILE_ATTRIBUTE_NORMAL
,
700 NULL
)) == CHM_NULL_FD
)
706 /* initialize mutexes, if needed */
707 InitializeCriticalSection(&newHandle
->mutex
);
708 newHandle
->mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.mutex");
709 InitializeCriticalSection(&newHandle
->lzx_mutex
);
710 newHandle
->lzx_mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.lzx_mutex");
711 InitializeCriticalSection(&newHandle
->cache_mutex
);
712 newHandle
->cache_mutex
.DebugInfo
->Spare
[0] = (DWORD_PTR
)(__FILE__
": chmFile.cache_mutex");
714 /* read and verify header */
715 sremain
= _CHM_ITSF_V3_LEN
;
717 if (_chm_fetch_bytes(newHandle
, sbuffer
, (UInt64
)0, sremain
) != sremain
||
718 !_unmarshal_itsf_header(&sbufpos
, &sremain
, &itsfHeader
))
720 chm_close(newHandle
);
724 /* stash important values from header */
725 newHandle
->dir_offset
= itsfHeader
.dir_offset
;
726 newHandle
->dir_len
= itsfHeader
.dir_len
;
727 newHandle
->data_offset
= itsfHeader
.data_offset
;
729 /* now, read and verify the directory header chunk */
730 sremain
= _CHM_ITSP_V1_LEN
;
732 if (_chm_fetch_bytes(newHandle
, sbuffer
,
733 (UInt64
)itsfHeader
.dir_offset
, sremain
) != sremain
||
734 !_unmarshal_itsp_header(&sbufpos
, &sremain
, &itspHeader
))
736 chm_close(newHandle
);
740 /* grab essential information from ITSP header */
741 newHandle
->dir_offset
+= itspHeader
.header_len
;
742 newHandle
->dir_len
-= itspHeader
.header_len
;
743 newHandle
->index_root
= itspHeader
.index_root
;
744 newHandle
->index_head
= itspHeader
.index_head
;
745 newHandle
->block_len
= itspHeader
.block_len
;
747 /* if the index root is -1, this means we don't have any PMGI blocks.
748 * as a result, we must use the sole PMGL block as the index root
750 if (newHandle
->index_root
== -1)
751 newHandle
->index_root
= newHandle
->index_head
;
753 /* By default, compression is enabled. */
754 newHandle
->compression_enabled
= 1;
756 /* prefetch most commonly needed unit infos */
757 if (CHM_RESOLVE_SUCCESS
!= chm_resolve_object(newHandle
,
759 &newHandle
->rt_unit
) ||
760 newHandle
->rt_unit
.space
== CHM_COMPRESSED
||
761 CHM_RESOLVE_SUCCESS
!= chm_resolve_object(newHandle
,
763 &newHandle
->cn_unit
) ||
764 newHandle
->cn_unit
.space
== CHM_COMPRESSED
||
765 CHM_RESOLVE_SUCCESS
!= chm_resolve_object(newHandle
,
766 _CHMU_LZXC_CONTROLDATA
,
768 uiLzxc
.space
== CHM_COMPRESSED
)
770 newHandle
->compression_enabled
= 0;
773 /* read reset table info */
774 if (newHandle
->compression_enabled
)
776 sremain
= _CHM_LZXC_RESETTABLE_V1_LEN
;
778 if (chm_retrieve_object(newHandle
, &newHandle
->rt_unit
, sbuffer
,
779 0, sremain
) != sremain
||
780 !_unmarshal_lzxc_reset_table(&sbufpos
, &sremain
,
781 &newHandle
->reset_table
))
783 newHandle
->compression_enabled
= 0;
787 /* read control data */
788 if (newHandle
->compression_enabled
)
790 sremain
= (unsigned long)uiLzxc
.length
;
792 if (chm_retrieve_object(newHandle
, &uiLzxc
, sbuffer
,
793 0, sremain
) != sremain
||
794 !_unmarshal_lzxc_control_data(&sbufpos
, &sremain
,
797 newHandle
->compression_enabled
= 0;
800 newHandle
->window_size
= ctlData
.windowSize
;
801 newHandle
->reset_interval
= ctlData
.resetInterval
;
803 /* Jed, Mon Jun 28: Experimentally, it appears that the reset block count */
804 /* must be multiplied by this formerly unknown ctrl data field in */
805 /* order to decompress some files. */
807 newHandle
->reset_blkcount
= newHandle
->reset_interval
/
808 (newHandle
->window_size
/ 2);
810 newHandle
->reset_blkcount
= newHandle
->reset_interval
/
811 (newHandle
->window_size
/ 2) *
812 ctlData
.windowsPerReset
;
816 /* initialize cache */
817 chm_set_param(newHandle
, CHM_PARAM_MAX_BLOCKS_CACHED
,
818 CHM_MAX_BLOCKS_CACHED
);
823 /* close an ITS archive */
824 void chm_close(struct chmFile
*h
)
828 if (h
->fd
!= CHM_NULL_FD
)
829 CHM_CLOSE_FILE(h
->fd
);
832 h
->mutex
.DebugInfo
->Spare
[0] = 0;
833 DeleteCriticalSection(&h
->mutex
);
834 h
->lzx_mutex
.DebugInfo
->Spare
[0] = 0;
835 DeleteCriticalSection(&h
->lzx_mutex
);
836 h
->cache_mutex
.DebugInfo
->Spare
[0] = 0;
837 DeleteCriticalSection(&h
->cache_mutex
);
840 LZXteardown(h
->lzx_state
);
846 for (i
=0; i
<h
->cache_num_blocks
; i
++)
848 if (h
->cache_blocks
[i
])
849 free(h
->cache_blocks
[i
]);
851 free(h
->cache_blocks
);
852 h
->cache_blocks
= NULL
;
855 free(h
->cache_block_indices
);
856 h
->cache_block_indices
= NULL
;
863 * helper methods for chm_resolve_object
866 /* skip a compressed dword */
867 static void _chm_skip_cword(UChar
**pEntry
)
869 while (*(*pEntry
)++ >= 0x80)
873 /* skip the data from a PMGL entry */
874 static void _chm_skip_PMGL_entry_data(UChar
**pEntry
)
876 _chm_skip_cword(pEntry
);
877 _chm_skip_cword(pEntry
);
878 _chm_skip_cword(pEntry
);
881 /* parse a compressed dword */
882 static UInt64
_chm_parse_cword(UChar
**pEntry
)
886 while ((temp
=*(*pEntry
)++) >= 0x80)
889 accum
+= temp
& 0x7f;
892 return (accum
<< 7) + temp
;
895 /* parse a utf-8 string into an ASCII char buffer */
896 static int _chm_parse_UTF8(UChar
**pEntry
, UInt64 count
, WCHAR
*path
)
898 /* MJM - Modified to return real Unicode strings */
901 *path
++ = (*(*pEntry
)++);
909 /* parse a PMGL entry into a chmUnitInfo struct; return 1 on success. */
910 static int _chm_parse_PMGL_entry(UChar
**pEntry
, struct chmUnitInfo
*ui
)
915 strLen
= _chm_parse_cword(pEntry
);
916 if (strLen
> CHM_MAX_PATHLEN
)
920 if (! _chm_parse_UTF8(pEntry
, strLen
, ui
->path
))
924 ui
->space
= (int)_chm_parse_cword(pEntry
);
925 ui
->start
= _chm_parse_cword(pEntry
);
926 ui
->length
= _chm_parse_cword(pEntry
);
930 /* find an exact entry in PMGL; return NULL if we fail */
931 static UChar
*_chm_find_in_PMGL(UChar
*page_buf
,
933 const WCHAR
*objPath
)
935 /* XXX: modify this to do a binary search using the nice index structure
936 * that is provided for us.
938 struct chmPmglHeader header
;
944 WCHAR buffer
[CHM_MAX_PATHLEN
+1];
946 /* figure out where to start and end */
948 hremain
= _CHM_PMGL_LEN
;
949 if (! _unmarshal_pmgl_header(&cur
, &hremain
, &header
))
951 end
= page_buf
+ block_len
- (header
.free_space
);
953 /* now, scan progressively */
958 strLen
= _chm_parse_cword(&cur
);
959 if (! _chm_parse_UTF8(&cur
, strLen
, buffer
))
962 /* check if it is the right name */
963 if (! strcmpiW(buffer
, objPath
))
966 _chm_skip_PMGL_entry_data(&cur
);
972 /* find which block should be searched next for the entry; -1 if no block */
973 static Int32
_chm_find_in_PMGI(UChar
*page_buf
,
975 const WCHAR
*objPath
)
977 /* XXX: modify this to do a binary search using the nice index structure
978 * that is provided for us
980 struct chmPmgiHeader header
;
986 WCHAR buffer
[CHM_MAX_PATHLEN
+1];
988 /* figure out where to start and end */
990 hremain
= _CHM_PMGI_LEN
;
991 if (! _unmarshal_pmgi_header(&cur
, &hremain
, &header
))
993 end
= page_buf
+ block_len
- (header
.free_space
);
995 /* now, scan progressively */
999 strLen
= _chm_parse_cword(&cur
);
1000 if (! _chm_parse_UTF8(&cur
, strLen
, buffer
))
1003 /* check if it is the right name */
1004 if (strcmpiW(buffer
, objPath
) > 0)
1007 /* load next value for path */
1008 page
= (int)_chm_parse_cword(&cur
);
1014 /* resolve a particular object from the archive */
1015 int chm_resolve_object(struct chmFile
*h
,
1016 const WCHAR
*objPath
,
1017 struct chmUnitInfo
*ui
)
1020 * XXX: implement caching scheme for dir pages
1025 /* buffer to hold whatever page we're looking at */
1026 UChar
*page_buf
= HeapAlloc(GetProcessHeap(), 0, h
->block_len
);
1029 curPage
= h
->index_root
;
1031 /* until we have either returned or given up */
1032 while (curPage
!= -1)
1035 /* try to fetch the index page */
1036 if (_chm_fetch_bytes(h
, page_buf
,
1037 (UInt64
)h
->dir_offset
+ (UInt64
)curPage
*h
->block_len
,
1038 h
->block_len
) != h
->block_len
)
1040 HeapFree(GetProcessHeap(), 0, page_buf
);
1041 return CHM_RESOLVE_FAILURE
;
1044 /* now, if it is a leaf node: */
1045 if (memcmp(page_buf
, _chm_pmgl_marker
, 4) == 0)
1048 UChar
*pEntry
= _chm_find_in_PMGL(page_buf
,
1053 HeapFree(GetProcessHeap(), 0, page_buf
);
1054 return CHM_RESOLVE_FAILURE
;
1057 /* parse entry and return */
1058 _chm_parse_PMGL_entry(&pEntry
, ui
);
1059 HeapFree(GetProcessHeap(), 0, page_buf
);
1060 return CHM_RESOLVE_SUCCESS
;
1063 /* else, if it is a branch node: */
1064 else if (memcmp(page_buf
, _chm_pmgi_marker
, 4) == 0)
1065 curPage
= _chm_find_in_PMGI(page_buf
, h
->block_len
, objPath
);
1067 /* else, we are confused. give up. */
1070 HeapFree(GetProcessHeap(), 0, page_buf
);
1071 return CHM_RESOLVE_FAILURE
;
1075 /* didn't find anything. fail. */
1076 HeapFree(GetProcessHeap(), 0, page_buf
);
1077 return CHM_RESOLVE_FAILURE
;
1081 * utility methods for dealing with compressed data
1084 /* get the bounds of a compressed block. return 0 on failure */
1085 static int _chm_get_cmpblock_bounds(struct chmFile
*h
,
1090 UChar buffer
[8], *dummy
;
1093 /* for all but the last block, use the reset table */
1094 if (block
< h
->reset_table
.block_count
-1)
1096 /* unpack the start address */
1099 if (_chm_fetch_bytes(h
, buffer
,
1100 (UInt64
)h
->data_offset
1101 + (UInt64
)h
->rt_unit
.start
1102 + (UInt64
)h
->reset_table
.table_offset
1104 remain
) != remain
||
1105 !_unmarshal_uint64(&dummy
, &remain
, start
))
1108 /* unpack the end address */
1111 if (_chm_fetch_bytes(h
, buffer
,
1112 (UInt64
)h
->data_offset
1113 + (UInt64
)h
->rt_unit
.start
1114 + (UInt64
)h
->reset_table
.table_offset
1115 + (UInt64
)block
*8 + 8,
1116 remain
) != remain
||
1117 !_unmarshal_int64(&dummy
, &remain
, len
))
1121 /* for the last block, use the span in addition to the reset table */
1124 /* unpack the start address */
1127 if (_chm_fetch_bytes(h
, buffer
,
1128 (UInt64
)h
->data_offset
1129 + (UInt64
)h
->rt_unit
.start
1130 + (UInt64
)h
->reset_table
.table_offset
1132 remain
) != remain
||
1133 !_unmarshal_uint64(&dummy
, &remain
, start
))
1136 *len
= h
->reset_table
.compressed_len
;
1139 /* compute the length and absolute start address */
1141 *start
+= h
->data_offset
+ h
->cn_unit
.start
;
1146 /* decompress the block. must have lzx_mutex. */
1147 static Int64
_chm_decompress_block(struct chmFile
*h
,
1151 UChar
*cbuffer
= HeapAlloc( GetProcessHeap(), 0,
1152 ((unsigned int)h
->reset_table
.block_len
+ 6144));
1153 UInt64 cmpStart
; /* compressed start */
1154 Int64 cmpLen
; /* compressed len */
1155 int indexSlot
; /* cache index slot */
1156 UChar
*lbuffer
; /* local buffer ptr */
1157 UInt32 blockAlign
= (UInt32
)(block
% h
->reset_blkcount
); /* reset intvl. aln. */
1158 UInt32 i
; /* local loop index */
1160 /* let the caching system pull its weight! */
1161 if (block
- blockAlign
<= h
->lzx_last_block
&&
1162 block
>= h
->lzx_last_block
)
1163 blockAlign
= (block
- h
->lzx_last_block
);
1165 /* check if we need previous blocks */
1166 if (blockAlign
!= 0)
1168 /* fetch all required previous blocks since last reset */
1169 for (i
= blockAlign
; i
> 0; i
--)
1171 UInt32 curBlockIdx
= block
- i
;
1173 /* check if we most recently decompressed the previous block */
1174 if (h
->lzx_last_block
!= curBlockIdx
)
1176 if ((curBlockIdx
% h
->reset_blkcount
) == 0)
1179 fprintf(stderr
, "***RESET (1)***\n");
1181 LZXreset(h
->lzx_state
);
1184 indexSlot
= (int)((curBlockIdx
) % h
->cache_num_blocks
);
1185 h
->cache_block_indices
[indexSlot
] = curBlockIdx
;
1186 if (! h
->cache_blocks
[indexSlot
])
1187 h
->cache_blocks
[indexSlot
] = malloc( (unsigned int)(h
->reset_table
.block_len
));
1188 lbuffer
= h
->cache_blocks
[indexSlot
];
1190 /* decompress the previous block */
1192 fprintf(stderr
, "Decompressing block #%4d (EXTRA)\n", curBlockIdx
);
1194 if (!_chm_get_cmpblock_bounds(h
, curBlockIdx
, &cmpStart
, &cmpLen
) ||
1195 _chm_fetch_bytes(h
, cbuffer
, cmpStart
, cmpLen
) != cmpLen
||
1196 LZXdecompress(h
->lzx_state
, cbuffer
, lbuffer
, (int)cmpLen
,
1197 (int)h
->reset_table
.block_len
) != DECR_OK
)
1200 fprintf(stderr
, " (DECOMPRESS FAILED!)\n");
1202 HeapFree(GetProcessHeap(), 0, cbuffer
);
1206 h
->lzx_last_block
= (int)curBlockIdx
;
1212 if ((block
% h
->reset_blkcount
) == 0)
1215 fprintf(stderr
, "***RESET (2)***\n");
1217 LZXreset(h
->lzx_state
);
1221 /* allocate slot in cache */
1222 indexSlot
= (int)(block
% h
->cache_num_blocks
);
1223 h
->cache_block_indices
[indexSlot
] = block
;
1224 if (! h
->cache_blocks
[indexSlot
])
1225 h
->cache_blocks
[indexSlot
] = malloc( ((unsigned int)h
->reset_table
.block_len
));
1226 lbuffer
= h
->cache_blocks
[indexSlot
];
1229 /* decompress the block we actually want */
1231 fprintf(stderr
, "Decompressing block #%4d (REAL )\n", block
);
1233 if (! _chm_get_cmpblock_bounds(h
, block
, &cmpStart
, &cmpLen
) ||
1234 _chm_fetch_bytes(h
, cbuffer
, cmpStart
, cmpLen
) != cmpLen
||
1235 LZXdecompress(h
->lzx_state
, cbuffer
, lbuffer
, (int)cmpLen
,
1236 (int)h
->reset_table
.block_len
) != DECR_OK
)
1239 fprintf(stderr
, " (DECOMPRESS FAILED!)\n");
1241 HeapFree(GetProcessHeap(), 0, cbuffer
);
1244 h
->lzx_last_block
= (int)block
;
1246 /* XXX: modify LZX routines to return the length of the data they
1247 * decompressed and return that instead, for an extra sanity check.
1249 HeapFree(GetProcessHeap(), 0, cbuffer
);
1250 return h
->reset_table
.block_len
;
1253 /* grab a region from a compressed block */
1254 static Int64
_chm_decompress_region(struct chmFile
*h
,
1259 UInt64 nBlock
, nOffset
;
1262 UChar
*ubuffer
= NULL
;
1267 /* figure out what we need to read */
1268 nBlock
= start
/ h
->reset_table
.block_len
;
1269 nOffset
= start
% h
->reset_table
.block_len
;
1271 if (nLen
> (h
->reset_table
.block_len
- nOffset
))
1272 nLen
= h
->reset_table
.block_len
- nOffset
;
1274 /* if block is cached, return data from it. */
1275 CHM_ACQUIRE_LOCK(h
->lzx_mutex
);
1276 CHM_ACQUIRE_LOCK(h
->cache_mutex
);
1277 if (h
->cache_block_indices
[nBlock
% h
->cache_num_blocks
] == nBlock
&&
1278 h
->cache_blocks
[nBlock
% h
->cache_num_blocks
] != NULL
)
1281 h
->cache_blocks
[nBlock
% h
->cache_num_blocks
] + nOffset
,
1282 (unsigned int)nLen
);
1283 CHM_RELEASE_LOCK(h
->cache_mutex
);
1284 CHM_RELEASE_LOCK(h
->lzx_mutex
);
1287 CHM_RELEASE_LOCK(h
->cache_mutex
);
1289 /* data request not satisfied, so... start up the decompressor machine */
1292 int window_size
= ffs(h
->window_size
) - 1;
1293 h
->lzx_last_block
= -1;
1294 h
->lzx_state
= LZXinit(window_size
);
1297 /* decompress some data */
1298 gotLen
= _chm_decompress_block(h
, nBlock
, &ubuffer
);
1301 memcpy(buf
, ubuffer
+nOffset
, (unsigned int)nLen
);
1302 CHM_RELEASE_LOCK(h
->lzx_mutex
);
1306 /* retrieve (part of) an object */
1307 LONGINT64
chm_retrieve_object(struct chmFile
*h
,
1308 struct chmUnitInfo
*ui
,
1313 /* must be valid file handle */
1317 /* starting address must be in correct range */
1318 if (addr
< 0 || addr
>= ui
->length
)
1322 if (addr
+ len
> ui
->length
)
1323 len
= ui
->length
- addr
;
1325 /* if the file is uncompressed, it's simple */
1326 if (ui
->space
== CHM_UNCOMPRESSED
)
1329 return _chm_fetch_bytes(h
,
1331 (UInt64
)h
->data_offset
+ (UInt64
)ui
->start
+ (UInt64
)addr
,
1335 /* else if the file is compressed, it's a little trickier */
1336 else /* ui->space == CHM_COMPRESSED */
1338 Int64 swath
=0, total
=0;
1340 /* if compression is not enabled for this file... */
1341 if (! h
->compression_enabled
)
1346 /* swill another mouthful */
1347 swath
= _chm_decompress_region(h
, buf
, ui
->start
+ addr
, len
);
1349 /* if we didn't get any... */
1365 /* enumerate the objects in the .chm archive */
1366 int chm_enumerate(struct chmFile
*h
,
1373 /* buffer to hold whatever page we're looking at */
1374 UChar
*page_buf
= HeapAlloc(GetProcessHeap(), 0, (unsigned int)h
->block_len
);
1375 struct chmPmglHeader header
;
1378 unsigned int lenRemain
;
1381 /* the current ui */
1382 struct chmUnitInfo ui
;
1386 curPage
= h
->index_head
;
1388 /* until we have either returned or given up */
1389 while (curPage
!= -1)
1392 /* try to fetch the index page */
1393 if (_chm_fetch_bytes(h
,
1395 (UInt64
)h
->dir_offset
+ (UInt64
)curPage
*h
->block_len
,
1396 h
->block_len
) != h
->block_len
)
1398 HeapFree(GetProcessHeap(), 0, page_buf
);
1402 /* figure out start and end for this page */
1404 lenRemain
= _CHM_PMGL_LEN
;
1405 if (! _unmarshal_pmgl_header(&cur
, &lenRemain
, &header
))
1407 HeapFree(GetProcessHeap(), 0, page_buf
);
1410 end
= page_buf
+ h
->block_len
- (header
.free_space
);
1412 /* loop over this page */
1415 if (! _chm_parse_PMGL_entry(&cur
, &ui
))
1417 HeapFree(GetProcessHeap(), 0, page_buf
);
1421 /* get the length of the path */
1422 ui_path_len
= strlenW(ui
.path
)-1;
1424 /* check for DIRS */
1425 if (ui
.path
[ui_path_len
] == '/' && !(what
& CHM_ENUMERATE_DIRS
))
1428 /* check for FILES */
1429 if (ui
.path
[ui_path_len
] != '/' && !(what
& CHM_ENUMERATE_FILES
))
1432 /* check for NORMAL vs. META */
1433 if (ui
.path
[0] == '/')
1436 /* check for NORMAL vs. SPECIAL */
1437 if (ui
.path
[1] == '#' || ui
.path
[1] == '$')
1438 flag
= CHM_ENUMERATE_SPECIAL
;
1440 flag
= CHM_ENUMERATE_NORMAL
;
1443 flag
= CHM_ENUMERATE_META
;
1444 if (! (what
& flag
))
1447 /* call the enumerator */
1449 int status
= (*e
)(h
, &ui
, context
);
1452 case CHM_ENUMERATOR_FAILURE
:
1453 HeapFree(GetProcessHeap(), 0, page_buf
);
1455 case CHM_ENUMERATOR_CONTINUE
:
1457 case CHM_ENUMERATOR_SUCCESS
:
1458 HeapFree(GetProcessHeap(), 0, page_buf
);
1466 /* advance to next page */
1467 curPage
= header
.block_next
;
1470 HeapFree(GetProcessHeap(), 0, page_buf
);
1474 int chm_enumerate_dir(struct chmFile
*h
,
1475 const WCHAR
*prefix
,
1481 * XXX: do this efficiently (i.e. using the tree index)
1486 /* buffer to hold whatever page we're looking at */
1487 UChar
*page_buf
= HeapAlloc(GetProcessHeap(), 0, (unsigned int)h
->block_len
);
1488 struct chmPmglHeader header
;
1491 unsigned int lenRemain
;
1493 /* set to 1 once we've started */
1496 /* the current ui */
1497 struct chmUnitInfo ui
;
1501 /* the length of the prefix */
1502 WCHAR prefixRectified
[CHM_MAX_PATHLEN
+1];
1504 WCHAR lastPath
[CHM_MAX_PATHLEN
];
1508 curPage
= h
->index_head
;
1510 /* initialize pathname state */
1511 lstrcpynW(prefixRectified
, prefix
, CHM_MAX_PATHLEN
);
1512 prefixLen
= strlenW(prefixRectified
);
1515 if (prefixRectified
[prefixLen
-1] != '/')
1517 prefixRectified
[prefixLen
] = '/';
1518 prefixRectified
[prefixLen
+1] = '\0';
1525 /* until we have either returned or given up */
1526 while (curPage
!= -1)
1529 /* try to fetch the index page */
1530 if (_chm_fetch_bytes(h
,
1532 (UInt64
)h
->dir_offset
+ (UInt64
)curPage
*h
->block_len
,
1533 h
->block_len
) != h
->block_len
)
1535 HeapFree(GetProcessHeap(), 0, page_buf
);
1539 /* figure out start and end for this page */
1541 lenRemain
= _CHM_PMGL_LEN
;
1542 if (! _unmarshal_pmgl_header(&cur
, &lenRemain
, &header
))
1544 HeapFree(GetProcessHeap(), 0, page_buf
);
1547 end
= page_buf
+ h
->block_len
- (header
.free_space
);
1549 /* loop over this page */
1552 if (! _chm_parse_PMGL_entry(&cur
, &ui
))
1554 HeapFree(GetProcessHeap(), 0, page_buf
);
1558 /* check if we should start */
1561 if (ui
.length
== 0 && strncmpiW(ui
.path
, prefixRectified
, prefixLen
) == 0)
1566 if (ui
.path
[prefixLen
] == '\0')
1570 /* check if we should stop */
1573 if (strncmpiW(ui
.path
, prefixRectified
, prefixLen
) != 0)
1575 HeapFree(GetProcessHeap(), 0, page_buf
);
1580 /* check if we should include this path */
1581 if (lastPathLen
!= -1)
1583 if (strncmpiW(ui
.path
, lastPath
, lastPathLen
) == 0)
1586 strcpyW(lastPath
, ui
.path
);
1587 lastPathLen
= strlenW(lastPath
);
1589 /* get the length of the path */
1590 ui_path_len
= strlenW(ui
.path
)-1;
1592 /* check for DIRS */
1593 if (ui
.path
[ui_path_len
] == '/' && !(what
& CHM_ENUMERATE_DIRS
))
1596 /* check for FILES */
1597 if (ui
.path
[ui_path_len
] != '/' && !(what
& CHM_ENUMERATE_FILES
))
1600 /* check for NORMAL vs. META */
1601 if (ui
.path
[0] == '/')
1604 /* check for NORMAL vs. SPECIAL */
1605 if (ui
.path
[1] == '#' || ui
.path
[1] == '$')
1606 flag
= CHM_ENUMERATE_SPECIAL
;
1608 flag
= CHM_ENUMERATE_NORMAL
;
1611 flag
= CHM_ENUMERATE_META
;
1612 if (! (what
& flag
))
1615 /* call the enumerator */
1617 int status
= (*e
)(h
, &ui
, context
);
1620 case CHM_ENUMERATOR_FAILURE
:
1621 HeapFree(GetProcessHeap(), 0, page_buf
);
1623 case CHM_ENUMERATOR_CONTINUE
:
1625 case CHM_ENUMERATOR_SUCCESS
:
1626 HeapFree(GetProcessHeap(), 0, page_buf
);
1634 /* advance to next page */
1635 curPage
= header
.block_next
;
1638 HeapFree(GetProcessHeap(), 0, page_buf
);