?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
advpack: Fix buffer sizes for possibly quoted strings.
[wine.git] / dlls / itss / chm_lib.c
blob3b98d5e781e26c896c3d09d18b3fbd3938419751
1 /***************************************************************************
2 * chm_lib.c - CHM archive manipulation routines *
3 * ------------------- *
4 * *
5 * author: Jed Wing <jedwin@ugcs.caltech.edu> *
6 * version: 0.3 *
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. *
11 * *
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. *
30 * *
31 ***************************************************************************/
32
33 /***************************************************************************
34 *
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.
39 *
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.
44 *
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
48 *
49 ***************************************************************************/
50
51 /***************************************************************************
52 * *
53 * Adapted for Wine by Mike McCormack *
54 * *
55 ***************************************************************************/
56
57 #include "config.h"
58 #include "wine/port.h"
59
60 #include <stdarg.h>
61 #include <stdio.h>
62 #include <stdlib.h>
63 #include <string.h>
64
65 #include "windef.h"
66 #include "winbase.h"
67 #include "wine/unicode.h"
68
69 #include "chm_lib.h"
70 #include "lzx.h"
71
72 #define CHM_ACQUIRE_LOCK(a) do { \
73 EnterCriticalSection(&(a)); \
74 } while(0)
75 #define CHM_RELEASE_LOCK(a) do { \
76 LeaveCriticalSection(&(a)); \
77 } while(0)
78
79 #define CHM_NULL_FD (INVALID_HANDLE_VALUE)
80 #define CHM_CLOSE_FILE(fd) CloseHandle((fd))
81
82 /*
83 * defines related to tuning
84 */
85 #ifndef CHM_MAX_BLOCKS_CACHED
86 #define CHM_MAX_BLOCKS_CACHED 5
87 #endif
88 #define CHM_PARAM_MAX_BLOCKS_CACHED 0
89
90 /*
91 * architecture specific defines
92 *
93 * Note: as soon as C99 is more widespread, the below defines should
94 * probably just use the C99 sized-int types.
95 *
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.
99 */
100
101 /* i386, 32-bit, Windows */
102 typedef BYTE UChar;
103 typedef SHORT Int16;
104 typedef USHORT UInt16;
105 typedef LONG Int32;
106 typedef DWORD UInt32;
107 typedef LONGLONG Int64;
108 typedef ULONGLONG UInt64;
109
110 /* utilities for unmarshalling data */
111 static int _unmarshal_char_array(unsigned char **pData,
112 unsigned int *pLenRemain,
113 char *dest,
114 int count)
115 {
116 if (count <= 0 || (unsigned int)count > *pLenRemain)
117 return 0;
118 memcpy(dest, (*pData), count);
119 *pData += count;
120 *pLenRemain -= count;
121 return 1;
122 }
123
124 static int _unmarshal_uchar_array(unsigned char **pData,
125 unsigned int *pLenRemain,
126 unsigned char *dest,
127 int count)
128 {
129 if (count <= 0 || (unsigned int)count > *pLenRemain)
130 return 0;
131 memcpy(dest, (*pData), count);
132 *pData += count;
133 *pLenRemain -= count;
134 return 1;
135 }
136
137 static int _unmarshal_int32(unsigned char **pData,
138 unsigned int *pLenRemain,
139 Int32 *dest)
140 {
141 if (4 > *pLenRemain)
142 return 0;
143 *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
144 *pData += 4;
145 *pLenRemain -= 4;
146 return 1;
147 }
148
149 static int _unmarshal_uint32(unsigned char **pData,
150 unsigned int *pLenRemain,
151 UInt32 *dest)
152 {
153 if (4 > *pLenRemain)
154 return 0;
155 *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
156 *pData += 4;
157 *pLenRemain -= 4;
158 return 1;
159 }
160
161 static int _unmarshal_int64(unsigned char **pData,
162 unsigned int *pLenRemain,
163 Int64 *dest)
164 {
165 Int64 temp;
166 int i;
167 if (8 > *pLenRemain)
168 return 0;
169 temp=0;
170 for(i=8; i>0; i--)
171 {
172 temp <<= 8;
173 temp |= (*pData)[i-1];
174 }
175 *dest = temp;
176 *pData += 8;
177 *pLenRemain -= 8;
178 return 1;
179 }
180
181 static int _unmarshal_uint64(unsigned char **pData,
182 unsigned int *pLenRemain,
183 UInt64 *dest)
184 {
185 UInt64 temp;
186 int i;
187 if (8 > *pLenRemain)
188 return 0;
189 temp=0;
190 for(i=8; i>0; i--)
191 {
192 temp <<= 8;
193 temp |= (*pData)[i-1];
194 }
195 *dest = temp;
196 *pData += 8;
197 *pLenRemain -= 8;
198 return 1;
199 }
200
201 static int _unmarshal_uuid(unsigned char **pData,
202 unsigned int *pDataLen,
203 unsigned char *dest)
204 {
205 return _unmarshal_uchar_array(pData, pDataLen, dest, 16);
206 }
207
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
219 };
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
225 };
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
231 };
232
233 /*
234 * structures local to this module
235 */
236
237 /* structure of ITSF headers */
238 #define _CHM_ITSF_V2_LEN (0x58)
239 #define _CHM_ITSF_V3_LEN (0x60)
240 struct chmItsfHeader
241 {
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))); */
256
257 static int _unmarshal_itsf_header(unsigned char **pData,
258 unsigned int *pDataLen,
259 struct chmItsfHeader *dest)
260 {
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)
263 return 0;
264
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);
278
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.
282 */
283 if (memcmp(dest->signature, "ITSF", 4) != 0)
284 return 0;
285 if (dest->version == 2)
286 {
287 if (dest->header_len < _CHM_ITSF_V2_LEN)
288 return 0;
289 }
290 else if (dest->version == 3)
291 {
292 if (dest->header_len < _CHM_ITSF_V3_LEN)
293 return 0;
294 }
295 else
296 return 0;
297
298 /* now, if we have a V3 structure, unmarshal the rest.
299 * otherwise, compute it
300 */
301 if (dest->version == 3)
302 {
303 if (*pDataLen != 0)
304 _unmarshal_uint64(pData, pDataLen, &dest->data_offset);
305 else
306 return 0;
307 }
308 else
309 dest->data_offset = dest->dir_offset + dest->dir_len;
310
311 return 1;
312 }
313
314 /* structure of ITSP headers */
315 #define _CHM_ITSP_V1_LEN (0x54)
316 struct chmItspHeader
317 {
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))); */
334
335 static int _unmarshal_itsp_header(unsigned char **pData,
336 unsigned int *pDataLen,
337 struct chmItspHeader *dest)
338 {
339 /* we only know how to deal with a 0x54 byte structures */
340 if (*pDataLen != _CHM_ITSP_V1_LEN)
341 return 0;
342
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);
359
360 /* error check the data */
361 if (memcmp(dest->signature, "ITSP", 4) != 0)
362 return 0;
363 if (dest->version != 1)
364 return 0;
365 if (dest->header_len != _CHM_ITSP_V1_LEN)
366 return 0;
367
368 return 1;
369 }
370
371 /* structure of PMGL headers */
372 static const char _chm_pmgl_marker[4] = "PMGL";
373 #define _CHM_PMGL_LEN (0x14)
374 struct chmPmglHeader
375 {
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))); */
382
383 static int _unmarshal_pmgl_header(unsigned char **pData,
384 unsigned int *pDataLen,
385 struct chmPmglHeader *dest)
386 {
387 /* we only know how to deal with a 0x14 byte structures */
388 if (*pDataLen != _CHM_PMGL_LEN)
389 return 0;
390
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);
397
398 /* check structure */
399 if (memcmp(dest->signature, _chm_pmgl_marker, 4) != 0)
400 return 0;
401
402 return 1;
403 }
404
405 /* structure of PMGI headers */
406 static const char _chm_pmgi_marker[4] = "PMGI";
407 #define _CHM_PMGI_LEN (0x08)
408 struct chmPmgiHeader
409 {
410 char signature[4]; /* 0 (PMGI) */
411 UInt32 free_space; /* 4 */
412 }; /* __attribute__ ((aligned (1))); */
413
414 static int _unmarshal_pmgi_header(unsigned char **pData,
415 unsigned int *pDataLen,
416 struct chmPmgiHeader *dest)
417 {
418 /* we only know how to deal with a 0x8 byte structures */
419 if (*pDataLen != _CHM_PMGI_LEN)
420 return 0;
421
422 /* unmarshal fields */
423 _unmarshal_char_array(pData, pDataLen, dest->signature, 4);
424 _unmarshal_uint32 (pData, pDataLen, &dest->free_space);
425
426 /* check structure */
427 if (memcmp(dest->signature, _chm_pmgi_marker, 4) != 0)
428 return 0;
429
430 return 1;
431 }
432
433 /* structure of LZXC reset table */
434 #define _CHM_LZXC_RESETTABLE_V1_LEN (0x28)
435 struct chmLzxcResetTable
436 {
437 UInt32 version;
438 UInt32 block_count;
439 UInt32 unknown;
440 UInt32 table_offset;
441 UInt64 uncompressed_len;
442 UInt64 compressed_len;
443 UInt64 block_len;
444 }; /* __attribute__ ((aligned (1))); */
445
446 static int _unmarshal_lzxc_reset_table(unsigned char **pData,
447 unsigned int *pDataLen,
448 struct chmLzxcResetTable *dest)
449 {
450 /* we only know how to deal with a 0x28 byte structures */
451 if (*pDataLen != _CHM_LZXC_RESETTABLE_V1_LEN)
452 return 0;
453
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);
462
463 /* check structure */
464 if (dest->version != 2)
465 return 0;
466
467 return 1;
468 }
469
470 /* structure of LZXC control data block */
471 #define _CHM_LZXC_MIN_LEN (0x18)
472 #define _CHM_LZXC_V2_LEN (0x1c)
473 struct chmLzxcControlData
474 {
475 UInt32 size; /* 0 */
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 */
482 };
483
484 static int _unmarshal_lzxc_control_data(unsigned char **pData,
485 unsigned int *pDataLen,
486 struct chmLzxcControlData *dest)
487 {
488 /* we want at least 0x18 bytes */
489 if (*pDataLen < _CHM_LZXC_MIN_LEN)
490 return 0;
491
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);
499
500 if (*pDataLen >= _CHM_LZXC_V2_LEN)
501 _unmarshal_uint32 (pData, pDataLen, &dest->unknown_18);
502 else
503 dest->unknown_18 = 0;
504
505 if (dest->version == 2)
506 {
507 dest->resetInterval *= 0x8000;
508 dest->windowSize *= 0x8000;
509 }
510 if (dest->windowSize == 0 || dest->resetInterval == 0)
511 return 0;
512
513 /* for now, only support resetInterval a multiple of windowSize/2 */
514 if (dest->windowSize == 1)
515 return 0;
516 if ((dest->resetInterval % (dest->windowSize/2)) != 0)
517 return 0;
518
519 /* check structure */
520 if (memcmp(dest->signature, "LZXC", 4) != 0)
521 return 0;
522
523 return 1;
524 }
525
526 /* the structure used for chm file handles */
527 struct chmFile
528 {
529 HANDLE fd;
530
531 CRITICAL_SECTION mutex;
532 CRITICAL_SECTION lzx_mutex;
533 CRITICAL_SECTION cache_mutex;
534
535 UInt64 dir_offset;
536 UInt64 dir_len;
537 UInt64 data_offset;
538 Int32 index_root;
539 Int32 index_head;
540 UInt32 block_len;
541
542 UInt64 span;
543 struct chmUnitInfo rt_unit;
544 struct chmUnitInfo cn_unit;
545 struct chmLzxcResetTable reset_table;
546
547 /* LZX control data */
548 int compression_enabled;
549 UInt32 window_size;
550 UInt32 reset_interval;
551 UInt32 reset_blkcount;
552
553 /* decompressor state */
554 struct LZXstate *lzx_state;
555 int lzx_last_block;
556
557 /* cache for decompressed blocks */
558 UChar **cache_blocks;
559 Int64 *cache_block_indices;
560 Int32 cache_num_blocks;
561 };
562
563 /*
564 * utility functions local to this module
565 */
566
567 /* utility function to handle differences between {pread,read}(64)? */
568 static Int64 _chm_fetch_bytes(struct chmFile *h,
569 UChar *buf,
570 UInt64 os,
571 Int64 len)
572 {
573 Int64 readLen=0;
574 if (h->fd == CHM_NULL_FD)
575 return readLen;
576
577 CHM_ACQUIRE_LOCK(h->mutex);
578 /* NOTE: this might be better done with CreateFileMapping, et cetera... */
579 {
580 LARGE_INTEGER old_pos, new_pos;
581 DWORD actualLen=0;
582
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 );
588
589 /* read the data */
590 if (ReadFile(h->fd,
591 buf,
592 (DWORD)len,
593 &actualLen,
594 NULL))
595 readLen = actualLen;
596 else
597 readLen = 0;
598
599 /* restore original position */
600 SetFilePointerEx( h->fd, old_pos, NULL, FILE_BEGIN );
601 }
602 CHM_RELEASE_LOCK(h->mutex);
603 return readLen;
604 }
605
606 /*
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.
614 */
615 static void chm_set_param(struct chmFile *h,
616 int paramType,
617 int paramVal)
618 {
619 switch (paramType)
620 {
621 case CHM_PARAM_MAX_BLOCKS_CACHED:
622 CHM_ACQUIRE_LOCK(h->cache_mutex);
623 if (paramVal != h->cache_num_blocks)
624 {
625 UChar **newBlocks;
626 Int64 *newIndices;
627 int i;
628
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++)
633 {
634 newBlocks[i] = NULL;
635 newIndices[i] = 0;
636 }
637
638 /* re-distribute old cached blocks */
639 if (h->cache_blocks)
640 {
641 for (i=0; i<h->cache_num_blocks; i++)
642 {
643 int newSlot = (int)(h->cache_block_indices[i] % paramVal);
644
645 if (h->cache_blocks[i])
646 {
647 /* in case of collision, destroy newcomer */
648 if (newBlocks[newSlot])
649 {
650 HeapFree(GetProcessHeap(), 0, h->cache_blocks[i]);
651 h->cache_blocks[i] = NULL;
652 }
653 else
654 {
655 newBlocks[newSlot] = h->cache_blocks[i];
656 newIndices[newSlot] =
657 h->cache_block_indices[i];
658 }
659 }
660 }
661
662 HeapFree(GetProcessHeap(), 0, h->cache_blocks);
663 HeapFree(GetProcessHeap(), 0, h->cache_block_indices);
664 }
665
666 /* now, set new values */
667 h->cache_blocks = newBlocks;
668 h->cache_block_indices = newIndices;
669 h->cache_num_blocks = paramVal;
670 }
671 CHM_RELEASE_LOCK(h->cache_mutex);
672 break;
673
674 default:
675 break;
676 }
677 }
678
679 /* open an ITS archive */
680 struct chmFile *chm_openW(const WCHAR *filename)
681 {
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;
688 #if 0
689 struct chmUnitInfo uiSpan;
690 #endif
691 struct chmUnitInfo uiLzxc;
692 struct chmLzxcControlData ctlData;
693
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;
701
702 /* open file */
703 if ((newHandle->fd=CreateFileW(filename,
704 GENERIC_READ,
705 FILE_SHARE_READ,
706 NULL,
707 OPEN_EXISTING,
708 FILE_ATTRIBUTE_NORMAL,
709 NULL)) == CHM_NULL_FD)
710 {
711 HeapFree(GetProcessHeap(), 0, newHandle);
712 return NULL;
713 }
714
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");
722
723 /* read and verify header */
724 sremain = _CHM_ITSF_V3_LEN;
725 sbufpos = sbuffer;
726 if (_chm_fetch_bytes(newHandle, sbuffer, (UInt64)0, sremain) != sremain ||
727 !_unmarshal_itsf_header(&sbufpos, &sremain, &itsfHeader))
728 {
729 chm_close(newHandle);
730 return NULL;
731 }
732
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;
737
738 /* now, read and verify the directory header chunk */
739 sremain = _CHM_ITSP_V1_LEN;
740 sbufpos = sbuffer;
741 if (_chm_fetch_bytes(newHandle, sbuffer,
742 itsfHeader.dir_offset, sremain) != sremain ||
743 !_unmarshal_itsp_header(&sbufpos, &sremain, &itspHeader))
744 {
745 chm_close(newHandle);
746 return NULL;
747 }
748
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;
755
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
758 */
759 if (newHandle->index_root == -1)
760 newHandle->index_root = newHandle->index_head;
761
762 /* By default, compression is enabled. */
763 newHandle->compression_enabled = 1;
764
765 /* prefetch most commonly needed unit infos */
766 if (CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
767 _CHMU_RESET_TABLE,
768 &newHandle->rt_unit) ||
769 newHandle->rt_unit.space == CHM_COMPRESSED ||
770 CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
771 _CHMU_CONTENT,
772 &newHandle->cn_unit) ||
773 newHandle->cn_unit.space == CHM_COMPRESSED ||
774 CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
775 _CHMU_LZXC_CONTROLDATA,
776 &uiLzxc) ||
777 uiLzxc.space == CHM_COMPRESSED)
778 {
779 newHandle->compression_enabled = 0;
780 }
781
782 /* read reset table info */
783 if (newHandle->compression_enabled)
784 {
785 sremain = _CHM_LZXC_RESETTABLE_V1_LEN;
786 sbufpos = sbuffer;
787 if (chm_retrieve_object(newHandle, &newHandle->rt_unit, sbuffer,
788 0, sremain) != sremain ||
789 !_unmarshal_lzxc_reset_table(&sbufpos, &sremain,
790 &newHandle->reset_table))
791 {
792 newHandle->compression_enabled = 0;
793 }
794 }
795
796 /* read control data */
797 if (newHandle->compression_enabled)
798 {
799 sremain = (unsigned long)uiLzxc.length;
800 sbufpos = sbuffer;
801 if (chm_retrieve_object(newHandle, &uiLzxc, sbuffer,
802 0, sremain) != sremain ||
803 !_unmarshal_lzxc_control_data(&sbufpos, &sremain,
804 &ctlData))
805 {
806 newHandle->compression_enabled = 0;
807 }
808
809 newHandle->window_size = ctlData.windowSize;
810 newHandle->reset_interval = ctlData.resetInterval;
811
812 /* Jed, Mon Jun 28: Experimentally, it appears that the reset block count */
813 /* must be multiplied by this formerly unknown ctrl data field in */
814 /* order to decompress some files. */
815 #if 0
816 newHandle->reset_blkcount = newHandle->reset_interval /
817 (newHandle->window_size / 2);
818 #else
819 newHandle->reset_blkcount = newHandle->reset_interval /
820 (newHandle->window_size / 2) *
821 ctlData.windowsPerReset;
822 #endif
823 }
824
825 /* initialize cache */
826 chm_set_param(newHandle, CHM_PARAM_MAX_BLOCKS_CACHED,
827 CHM_MAX_BLOCKS_CACHED);
828
829 return newHandle;
830 }
831
832 /* close an ITS archive */
833 void chm_close(struct chmFile *h)
834 {
835 if (h != NULL)
836 {
837 if (h->fd != CHM_NULL_FD)
838 CHM_CLOSE_FILE(h->fd);
839 h->fd = CHM_NULL_FD;
840
841 h->mutex.DebugInfo->Spare[0] = 0;
842 DeleteCriticalSection(&h->mutex);
843 h->lzx_mutex.DebugInfo->Spare[0] = 0;
844 DeleteCriticalSection(&h->lzx_mutex);
845 h->cache_mutex.DebugInfo->Spare[0] = 0;
846 DeleteCriticalSection(&h->cache_mutex);
847
848 if (h->lzx_state)
849 LZXteardown(h->lzx_state);
850 h->lzx_state = NULL;
851
852 if (h->cache_blocks)
853 {
854 int i;
855 for (i=0; i<h->cache_num_blocks; i++)
856 {
857 HeapFree(GetProcessHeap(), 0, h->cache_blocks[i]);
858 }
859 HeapFree(GetProcessHeap(), 0, h->cache_blocks);
860 h->cache_blocks = NULL;
861 }
862
863 HeapFree(GetProcessHeap(), 0, h->cache_block_indices);
864 h->cache_block_indices = NULL;
865
866 HeapFree(GetProcessHeap(), 0, h);
867 }
868 }
869
870 /*
871 * helper methods for chm_resolve_object
872 */
873
874 /* skip a compressed dword */
875 static void _chm_skip_cword(UChar **pEntry)
876 {
877 while (*(*pEntry)++ >= 0x80)
878 ;
879 }
880
881 /* skip the data from a PMGL entry */
882 static void _chm_skip_PMGL_entry_data(UChar **pEntry)
883 {
884 _chm_skip_cword(pEntry);
885 _chm_skip_cword(pEntry);
886 _chm_skip_cword(pEntry);
887 }
888
889 /* parse a compressed dword */
890 static UInt64 _chm_parse_cword(UChar **pEntry)
891 {
892 UInt64 accum = 0;
893 UChar temp;
894 while ((temp=*(*pEntry)++) >= 0x80)
895 {
896 accum <<= 7;
897 accum += temp & 0x7f;
898 }
899
900 return (accum << 7) + temp;
901 }
902
903 /* parse a utf-8 string into an ASCII char buffer */
904 static int _chm_parse_UTF8(UChar **pEntry, UInt64 count, WCHAR *path)
905 {
906 /* MJM - Modified to return real Unicode strings */
907 while (count != 0)
908 {
909 *path++ = (*(*pEntry)++);
910 --count;
911 }
912
913 *path = '\0';
914 return 1;
915 }
916
917 /* parse a PMGL entry into a chmUnitInfo struct; return 1 on success. */
918 static int _chm_parse_PMGL_entry(UChar **pEntry, struct chmUnitInfo *ui)
919 {
920 UInt64 strLen;
921
922 /* parse str len */
923 strLen = _chm_parse_cword(pEntry);
924 if (strLen > CHM_MAX_PATHLEN)
925 return 0;
926
927 /* parse path */
928 if (! _chm_parse_UTF8(pEntry, strLen, ui->path))
929 return 0;
930
931 /* parse info */
932 ui->space = (int)_chm_parse_cword(pEntry);
933 ui->start = _chm_parse_cword(pEntry);
934 ui->length = _chm_parse_cword(pEntry);
935 return 1;
936 }
937
938 /* find an exact entry in PMGL; return NULL if we fail */
939 static UChar *_chm_find_in_PMGL(UChar *page_buf,
940 UInt32 block_len,
941 const WCHAR *objPath)
942 {
943 /* XXX: modify this to do a binary search using the nice index structure
944 * that is provided for us.
945 */
946 struct chmPmglHeader header;
947 UInt32 hremain;
948 UChar *end;
949 UChar *cur;
950 UChar *temp;
951 UInt64 strLen;
952 WCHAR buffer[CHM_MAX_PATHLEN+1];
953
954 /* figure out where to start and end */
955 cur = page_buf;
956 hremain = _CHM_PMGL_LEN;
957 if (! _unmarshal_pmgl_header(&cur, &hremain, &header))
958 return NULL;
959 end = page_buf + block_len - (header.free_space);
960
961 /* now, scan progressively */
962 while (cur < end)
963 {
964 /* grab the name */
965 temp = cur;
966 strLen = _chm_parse_cword(&cur);
967 if (! _chm_parse_UTF8(&cur, strLen, buffer))
968 return NULL;
969
970 /* check if it is the right name */
971 if (! strcmpiW(buffer, objPath))
972 return temp;
973
974 _chm_skip_PMGL_entry_data(&cur);
975 }
976
977 return NULL;
978 }
979
980 /* find which block should be searched next for the entry; -1 if no block */
981 static Int32 _chm_find_in_PMGI(UChar *page_buf,
982 UInt32 block_len,
983 const WCHAR *objPath)
984 {
985 /* XXX: modify this to do a binary search using the nice index structure
986 * that is provided for us
987 */
988 struct chmPmgiHeader header;
989 UInt32 hremain;
990 int page=-1;
991 UChar *end;
992 UChar *cur;
993 UInt64 strLen;
994 WCHAR buffer[CHM_MAX_PATHLEN+1];
995
996 /* figure out where to start and end */
997 cur = page_buf;
998 hremain = _CHM_PMGI_LEN;
999 if (! _unmarshal_pmgi_header(&cur, &hremain, &header))
1000 return -1;
1001 end = page_buf + block_len - (header.free_space);
1002
1003 /* now, scan progressively */
1004 while (cur < end)
1005 {
1006 /* grab the name */
1007 strLen = _chm_parse_cword(&cur);
1008 if (! _chm_parse_UTF8(&cur, strLen, buffer))
1009 return -1;
1010
1011 /* check if it is the right name */
1012 if (strcmpiW(buffer, objPath) > 0)
1013 return page;
1014
1015 /* load next value for path */
1016 page = (int)_chm_parse_cword(&cur);
1017 }
1018
1019 return page;
1020 }
1021
1022 /* resolve a particular object from the archive */
1023 int chm_resolve_object(struct chmFile *h,
1024 const WCHAR *objPath,
1025 struct chmUnitInfo *ui)
1026 {
1027 /*
1028 * XXX: implement caching scheme for dir pages
1029 */
1030
1031 Int32 curPage;
1032
1033 /* buffer to hold whatever page we're looking at */
1034 UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, h->block_len);
1035
1036 /* starting page */
1037 curPage = h->index_root;
1038
1039 /* until we have either returned or given up */
1040 while (curPage != -1)
1041 {
1042
1043 /* try to fetch the index page */
1044 if (_chm_fetch_bytes(h, page_buf,
1045 h->dir_offset + (UInt64)curPage*h->block_len,
1046 h->block_len) != h->block_len)
1047 {
1048 HeapFree(GetProcessHeap(), 0, page_buf);
1049 return CHM_RESOLVE_FAILURE;
1050 }
1051
1052 /* now, if it is a leaf node: */
1053 if (memcmp(page_buf, _chm_pmgl_marker, 4) == 0)
1054 {
1055 /* scan block */
1056 UChar *pEntry = _chm_find_in_PMGL(page_buf,
1057 h->block_len,
1058 objPath);
1059 if (pEntry == NULL)
1060 {
1061 HeapFree(GetProcessHeap(), 0, page_buf);
1062 return CHM_RESOLVE_FAILURE;
1063 }
1064
1065 /* parse entry and return */
1066 _chm_parse_PMGL_entry(&pEntry, ui);
1067 HeapFree(GetProcessHeap(), 0, page_buf);
1068 return CHM_RESOLVE_SUCCESS;
1069 }
1070
1071 /* else, if it is a branch node: */
1072 else if (memcmp(page_buf, _chm_pmgi_marker, 4) == 0)
1073 curPage = _chm_find_in_PMGI(page_buf, h->block_len, objPath);
1074
1075 /* else, we are confused. give up. */
1076 else
1077 {
1078 HeapFree(GetProcessHeap(), 0, page_buf);
1079 return CHM_RESOLVE_FAILURE;
1080 }
1081 }
1082
1083 /* didn't find anything. fail. */
1084 HeapFree(GetProcessHeap(), 0, page_buf);
1085 return CHM_RESOLVE_FAILURE;
1086 }
1087
1088 /*
1089 * utility methods for dealing with compressed data
1090 */
1091
1092 /* get the bounds of a compressed block. return 0 on failure */
1093 static int _chm_get_cmpblock_bounds(struct chmFile *h,
1094 UInt64 block,
1095 UInt64 *start,
1096 Int64 *len)
1097 {
1098 UChar buffer[8], *dummy;
1099 UInt32 remain;
1100
1101 /* for all but the last block, use the reset table */
1102 if (block < h->reset_table.block_count-1)
1103 {
1104 /* unpack the start address */
1105 dummy = buffer;
1106 remain = 8;
1107 if (_chm_fetch_bytes(h, buffer,
1108 h->data_offset
1109 + h->rt_unit.start
1110 + h->reset_table.table_offset
1111 + block*8,
1112 remain) != remain ||
1113 !_unmarshal_uint64(&dummy, &remain, start))
1114 return 0;
1115
1116 /* unpack the end address */
1117 dummy = buffer;
1118 remain = 8;
1119 if (_chm_fetch_bytes(h, buffer,
1120 h->data_offset
1121 + h->rt_unit.start
1122 + h->reset_table.table_offset
1123 + block*8 + 8,
1124 remain) != remain ||
1125 !_unmarshal_int64(&dummy, &remain, len))
1126 return 0;
1127 }
1128
1129 /* for the last block, use the span in addition to the reset table */
1130 else
1131 {
1132 /* unpack the start address */
1133 dummy = buffer;
1134 remain = 8;
1135 if (_chm_fetch_bytes(h, buffer,
1136 h->data_offset
1137 + h->rt_unit.start
1138 + h->reset_table.table_offset
1139 + block*8,
1140 remain) != remain ||
1141 !_unmarshal_uint64(&dummy, &remain, start))
1142 return 0;
1143
1144 *len = h->reset_table.compressed_len;
1145 }
1146
1147 /* compute the length and absolute start address */
1148 *len -= *start;
1149 *start += h->data_offset + h->cn_unit.start;
1150
1151 return 1;
1152 }
1153
1154 /* decompress the block. must have lzx_mutex. */
1155 static Int64 _chm_decompress_block(struct chmFile *h,
1156 UInt64 block,
1157 UChar **ubuffer)
1158 {
1159 UChar *cbuffer = HeapAlloc( GetProcessHeap(), 0,
1160 ((unsigned int)h->reset_table.block_len + 6144));
1161 UInt64 cmpStart; /* compressed start */
1162 Int64 cmpLen; /* compressed len */
1163 int indexSlot; /* cache index slot */
1164 UChar *lbuffer; /* local buffer ptr */
1165 UInt32 blockAlign = (UInt32)(block % h->reset_blkcount); /* reset interval align */
1166 UInt32 i; /* local loop index */
1167
1168 /* let the caching system pull its weight! */
1169 if (block - blockAlign <= h->lzx_last_block &&
1170 block >= h->lzx_last_block)
1171 blockAlign = (block - h->lzx_last_block);
1172
1173 /* check if we need previous blocks */
1174 if (blockAlign != 0)
1175 {
1176 /* fetch all required previous blocks since last reset */
1177 for (i = blockAlign; i > 0; i--)
1178 {
1179 UInt32 curBlockIdx = block - i;
1180
1181 /* check if we most recently decompressed the previous block */
1182 if (h->lzx_last_block != curBlockIdx)
1183 {
1184 if ((curBlockIdx % h->reset_blkcount) == 0)
1185 {
1186 #ifdef CHM_DEBUG
1187 fprintf(stderr, "***RESET (1)***\n");
1188 #endif
1189 LZXreset(h->lzx_state);
1190 }
1191
1192 indexSlot = (int)((curBlockIdx) % h->cache_num_blocks);
1193 h->cache_block_indices[indexSlot] = curBlockIdx;
1194 if (! h->cache_blocks[indexSlot])
1195 h->cache_blocks[indexSlot] =
1196 HeapAlloc(GetProcessHeap(), 0,
1197 (unsigned int)(h->reset_table.block_len));
1198 lbuffer = h->cache_blocks[indexSlot];
1199
1200 /* decompress the previous block */
1201 #ifdef CHM_DEBUG
1202 fprintf(stderr, "Decompressing block #%4d (EXTRA)\n", curBlockIdx);
1203 #endif
1204 if (!_chm_get_cmpblock_bounds(h, curBlockIdx, &cmpStart, &cmpLen) ||
1205 _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen ||
1206 LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1207 (int)h->reset_table.block_len) != DECR_OK)
1208 {
1209 #ifdef CHM_DEBUG
1210 fprintf(stderr, " (DECOMPRESS FAILED!)\n");
1211 #endif
1212 HeapFree(GetProcessHeap(), 0, cbuffer);
1213 return (Int64)0;
1214 }
1215
1216 h->lzx_last_block = (int)curBlockIdx;
1217 }
1218 }
1219 }
1220 else
1221 {
1222 if ((block % h->reset_blkcount) == 0)
1223 {
1224 #ifdef CHM_DEBUG
1225 fprintf(stderr, "***RESET (2)***\n");
1226 #endif
1227 LZXreset(h->lzx_state);
1228 }
1229 }
1230
1231 /* allocate slot in cache */
1232 indexSlot = (int)(block % h->cache_num_blocks);
1233 h->cache_block_indices[indexSlot] = block;
1234 if (! h->cache_blocks[indexSlot])
1235 h->cache_blocks[indexSlot] =
1236 HeapAlloc(GetProcessHeap(), 0, ((unsigned int)h->reset_table.block_len));
1237 lbuffer = h->cache_blocks[indexSlot];
1238 *ubuffer = lbuffer;
1239
1240 /* decompress the block we actually want */
1241 #ifdef CHM_DEBUG
1242 fprintf(stderr, "Decompressing block #%4d (REAL )\n", block);
1243 #endif
1244 if (! _chm_get_cmpblock_bounds(h, block, &cmpStart, &cmpLen) ||
1245 _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen ||
1246 LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1247 (int)h->reset_table.block_len) != DECR_OK)
1248 {
1249 #ifdef CHM_DEBUG
1250 fprintf(stderr, " (DECOMPRESS FAILED!)\n");
1251 #endif
1252 HeapFree(GetProcessHeap(), 0, cbuffer);
1253 return (Int64)0;
1254 }
1255 h->lzx_last_block = (int)block;
1256
1257 /* XXX: modify LZX routines to return the length of the data they
1258 * decompressed and return that instead, for an extra sanity check.
1259 */
1260 HeapFree(GetProcessHeap(), 0, cbuffer);
1261 return h->reset_table.block_len;
1262 }
1263
1264 /* grab a region from a compressed block */
1265 static Int64 _chm_decompress_region(struct chmFile *h,
1266 UChar *buf,
1267 UInt64 start,
1268 Int64 len)
1269 {
1270 UInt64 nBlock, nOffset;
1271 UInt64 nLen;
1272 UInt64 gotLen;
1273 UChar *ubuffer = NULL;
1274
1275 if (len <= 0)
1276 return (Int64)0;
1277
1278 /* figure out what we need to read */
1279 nBlock = start / h->reset_table.block_len;
1280 nOffset = start % h->reset_table.block_len;
1281 nLen = len;
1282 if (nLen > (h->reset_table.block_len - nOffset))
1283 nLen = h->reset_table.block_len - nOffset;
1284
1285 /* if block is cached, return data from it. */
1286 CHM_ACQUIRE_LOCK(h->lzx_mutex);
1287 CHM_ACQUIRE_LOCK(h->cache_mutex);
1288 if (h->cache_block_indices[nBlock % h->cache_num_blocks] == nBlock &&
1289 h->cache_blocks[nBlock % h->cache_num_blocks] != NULL)
1290 {
1291 memcpy(buf,
1292 h->cache_blocks[nBlock % h->cache_num_blocks] + nOffset,
1293 (unsigned int)nLen);
1294 CHM_RELEASE_LOCK(h->cache_mutex);
1295 CHM_RELEASE_LOCK(h->lzx_mutex);
1296 return nLen;
1297 }
1298 CHM_RELEASE_LOCK(h->cache_mutex);
1299
1300 /* data request not satisfied, so... start up the decompressor machine */
1301 if (! h->lzx_state)
1302 {
1303 int window_size = ffs(h->window_size) - 1;
1304 h->lzx_last_block = -1;
1305 h->lzx_state = LZXinit(window_size);
1306 }
1307
1308 /* decompress some data */
1309 gotLen = _chm_decompress_block(h, nBlock, &ubuffer);
1310 if (gotLen < nLen)
1311 nLen = gotLen;
1312 memcpy(buf, ubuffer+nOffset, (unsigned int)nLen);
1313 CHM_RELEASE_LOCK(h->lzx_mutex);
1314 return nLen;
1315 }
1316
1317 /* retrieve (part of) an object */
1318 LONGINT64 chm_retrieve_object(struct chmFile *h,
1319 struct chmUnitInfo *ui,
1320 unsigned char *buf,
1321 LONGUINT64 addr,
1322 LONGINT64 len)
1323 {
1324 /* must be valid file handle */
1325 if (h == NULL)
1326 return (Int64)0;
1327
1328 /* starting address must be in correct range */
1329 if (addr >= ui->length)
1330 return (Int64)0;
1331
1332 /* clip length */
1333 if (addr + len > ui->length)
1334 len = ui->length - addr;
1335
1336 /* if the file is uncompressed, it's simple */
1337 if (ui->space == CHM_UNCOMPRESSED)
1338 {
1339 /* read data */
1340 return _chm_fetch_bytes(h,
1341 buf,
1342 h->data_offset + ui->start + addr,
1343 len);
1344 }
1345
1346 /* else if the file is compressed, it's a little trickier */
1347 else /* ui->space == CHM_COMPRESSED */
1348 {
1349 Int64 swath=0, total=0;
1350
1351 /* if compression is not enabled for this file... */
1352 if (! h->compression_enabled)
1353 return total;
1354
1355 do {
1356
1357 /* swill another mouthful */
1358 swath = _chm_decompress_region(h, buf, ui->start + addr, len);
1359
1360 /* if we didn't get any... */
1361 if (swath == 0)
1362 return total;
1363
1364 /* update stats */
1365 total += swath;
1366 len -= swath;
1367 addr += swath;
1368 buf += swath;
1369
1370 } while (len != 0);
1371
1372 return total;
1373 }
1374 }
1375
1376 /* enumerate the objects in the .chm archive */
1377 int chm_enumerate(struct chmFile *h,
1378 int what,
1379 CHM_ENUMERATOR e,
1380 void *context)
1381 {
1382 Int32 curPage;
1383
1384 /* buffer to hold whatever page we're looking at */
1385 UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, h->block_len);
1386 struct chmPmglHeader header;
1387 UChar *end;
1388 UChar *cur;
1389 unsigned int lenRemain;
1390 UInt64 ui_path_len;
1391
1392 /* the current ui */
1393 struct chmUnitInfo ui;
1394 int flag;
1395
1396 /* starting page */
1397 curPage = h->index_head;
1398
1399 /* until we have either returned or given up */
1400 while (curPage != -1)
1401 {
1402
1403 /* try to fetch the index page */
1404 if (_chm_fetch_bytes(h,
1405 page_buf,
1406 h->dir_offset + (UInt64)curPage*h->block_len,
1407 h->block_len) != h->block_len)
1408 {
1409 HeapFree(GetProcessHeap(), 0, page_buf);
1410 return 0;
1411 }
1412
1413 /* figure out start and end for this page */
1414 cur = page_buf;
1415 lenRemain = _CHM_PMGL_LEN;
1416 if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1417 {
1418 HeapFree(GetProcessHeap(), 0, page_buf);
1419 return 0;
1420 }
1421 end = page_buf + h->block_len - (header.free_space);
1422
1423 /* loop over this page */
1424 while (cur < end)
1425 {
1426 if (! _chm_parse_PMGL_entry(&cur, &ui))
1427 {
1428 HeapFree(GetProcessHeap(), 0, page_buf);
1429 return 0;
1430 }
1431
1432 /* get the length of the path */
1433 ui_path_len = strlenW(ui.path)-1;
1434
1435 /* check for DIRS */
1436 if (ui.path[ui_path_len] == '/' && !(what & CHM_ENUMERATE_DIRS))
1437 continue;
1438
1439 /* check for FILES */
1440 if (ui.path[ui_path_len] != '/' && !(what & CHM_ENUMERATE_FILES))
1441 continue;
1442
1443 /* check for NORMAL vs. META */
1444 if (ui.path[0] == '/')
1445 {
1446
1447 /* check for NORMAL vs. SPECIAL */
1448 if (ui.path[1] == '#' || ui.path[1] == '$')
1449 flag = CHM_ENUMERATE_SPECIAL;
1450 else
1451 flag = CHM_ENUMERATE_NORMAL;
1452 }
1453 else
1454 flag = CHM_ENUMERATE_META;
1455 if (! (what & flag))
1456 continue;
1457
1458 /* call the enumerator */
1459 {
1460 int status = (*e)(h, &ui, context);
1461 switch (status)
1462 {
1463 case CHM_ENUMERATOR_FAILURE:
1464 HeapFree(GetProcessHeap(), 0, page_buf);
1465 return 0;
1466 case CHM_ENUMERATOR_CONTINUE:
1467 break;
1468 case CHM_ENUMERATOR_SUCCESS:
1469 HeapFree(GetProcessHeap(), 0, page_buf);
1470 return 1;
1471 default:
1472 break;
1473 }
1474 }
1475 }
1476
1477 /* advance to next page */
1478 curPage = header.block_next;
1479 }
1480
1481 HeapFree(GetProcessHeap(), 0, page_buf);
1482 return 1;
1483 }
1484
1485 int chm_enumerate_dir(struct chmFile *h,
1486 const WCHAR *prefix,
1487 int what,
1488 CHM_ENUMERATOR e,
1489 void *context)
1490 {
1491 /*
1492 * XXX: do this efficiently (i.e. using the tree index)
1493 */
1494
1495 Int32 curPage;
1496
1497 /* buffer to hold whatever page we're looking at */
1498 UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, h->block_len);
1499 struct chmPmglHeader header;
1500 UChar *end;
1501 UChar *cur;
1502 unsigned int lenRemain;
1503
1504 /* set to 1 once we've started */
1505 int it_has_begun=0;
1506
1507 /* the current ui */
1508 struct chmUnitInfo ui;
1509 int flag;
1510 UInt64 ui_path_len;
1511
1512 /* the length of the prefix */
1513 WCHAR prefixRectified[CHM_MAX_PATHLEN+1];
1514 int prefixLen;
1515 WCHAR lastPath[CHM_MAX_PATHLEN];
1516 int lastPathLen;
1517
1518 /* starting page */
1519 curPage = h->index_head;
1520
1521 /* initialize pathname state */
1522 lstrcpynW(prefixRectified, prefix, CHM_MAX_PATHLEN);
1523 prefixLen = strlenW(prefixRectified);
1524 if (prefixLen != 0)
1525 {
1526 if (prefixRectified[prefixLen-1] != '/')
1527 {
1528 prefixRectified[prefixLen] = '/';
1529 prefixRectified[prefixLen+1] = '\0';
1530 ++prefixLen;
1531 }
1532 }
1533 lastPath[0] = '\0';
1534 lastPathLen = -1;
1535
1536 /* until we have either returned or given up */
1537 while (curPage != -1)
1538 {
1539
1540 /* try to fetch the index page */
1541 if (_chm_fetch_bytes(h,
1542 page_buf,
1543 h->dir_offset + (UInt64)curPage*h->block_len,
1544 h->block_len) != h->block_len)
1545 {
1546 HeapFree(GetProcessHeap(), 0, page_buf);
1547 return 0;
1548 }
1549
1550 /* figure out start and end for this page */
1551 cur = page_buf;
1552 lenRemain = _CHM_PMGL_LEN;
1553 if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1554 {
1555 HeapFree(GetProcessHeap(), 0, page_buf);
1556 return 0;
1557 }
1558 end = page_buf + h->block_len - (header.free_space);
1559
1560 /* loop over this page */
1561 while (cur < end)
1562 {
1563 if (! _chm_parse_PMGL_entry(&cur, &ui))
1564 {
1565 HeapFree(GetProcessHeap(), 0, page_buf);
1566 return 0;
1567 }
1568
1569 /* check if we should start */
1570 if (! it_has_begun)
1571 {
1572 if (ui.length == 0 && strncmpiW(ui.path, prefixRectified, prefixLen) == 0)
1573 it_has_begun = 1;
1574 else
1575 continue;
1576
1577 if (ui.path[prefixLen] == '\0')
1578 continue;
1579 }
1580
1581 /* check if we should stop */
1582 else
1583 {
1584 if (strncmpiW(ui.path, prefixRectified, prefixLen) != 0)
1585 {
1586 HeapFree(GetProcessHeap(), 0, page_buf);
1587 return 1;
1588 }
1589 }
1590
1591 /* check if we should include this path */
1592 if (lastPathLen != -1)
1593 {
1594 if (strncmpiW(ui.path, lastPath, lastPathLen) == 0)
1595 continue;
1596 }
1597 strcpyW(lastPath, ui.path);
1598 lastPathLen = strlenW(lastPath);
1599
1600 /* get the length of the path */
1601 ui_path_len = strlenW(ui.path)-1;
1602
1603 /* check for DIRS */
1604 if (ui.path[ui_path_len] == '/' && !(what & CHM_ENUMERATE_DIRS))
1605 continue;
1606
1607 /* check for FILES */
1608 if (ui.path[ui_path_len] != '/' && !(what & CHM_ENUMERATE_FILES))
1609 continue;
1610
1611 /* check for NORMAL vs. META */
1612 if (ui.path[0] == '/')
1613 {
1614
1615 /* check for NORMAL vs. SPECIAL */
1616 if (ui.path[1] == '#' || ui.path[1] == '$')
1617 flag = CHM_ENUMERATE_SPECIAL;
1618 else
1619 flag = CHM_ENUMERATE_NORMAL;
1620 }
1621 else
1622 flag = CHM_ENUMERATE_META;
1623 if (! (what & flag))
1624 continue;
1625
1626 /* call the enumerator */
1627 {
1628 int status = (*e)(h, &ui, context);
1629 switch (status)
1630 {
1631 case CHM_ENUMERATOR_FAILURE:
1632 HeapFree(GetProcessHeap(), 0, page_buf);
1633 return 0;
1634 case CHM_ENUMERATOR_CONTINUE:
1635 break;
1636 case CHM_ENUMERATOR_SUCCESS:
1637 HeapFree(GetProcessHeap(), 0, page_buf);
1638 return 1;
1639 default:
1640 break;
1641 }
1642 }
1643 }
1644
1645 /* advance to next page */
1646 curPage = header.block_next;
1647 }
1648
1649 HeapFree(GetProcessHeap(), 0, page_buf);
1650 return 1;
1651 }
Windows API implementation