?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
oleaut32/tests: Fix the FSF address.
[wine/multimedia.git] / dlls / itss / chm_lib.c
blob074ca3dfa94e4d4225ba326590e12b58d1df84f0
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 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. *
39 * *
40 ***************************************************************************/
41
42 /***************************************************************************
43 * *
44 * Adapted for Wine by Mike McCormack *
45 * *
46 ***************************************************************************/
47
48 #include "config.h"
49 #include "wine/port.h"
50
51 #include <stdarg.h>
52 #include <stdio.h>
53 #include <stdlib.h>
54 #include <string.h>
55
56 #include "windef.h"
57 #include "winbase.h"
58 #include "wine/unicode.h"
59
60 #include "chm_lib.h"
61 #include "lzx.h"
62
63 #define CHM_ACQUIRE_LOCK(a) do { \
64 EnterCriticalSection(&(a)); \
65 } while(0)
66 #define CHM_RELEASE_LOCK(a) do { \
67 LeaveCriticalSection(&(a)); \
68 } while(0)
69
70 #define CHM_NULL_FD (INVALID_HANDLE_VALUE)
71 #define CHM_CLOSE_FILE(fd) CloseHandle((fd))
72
73 /*
74 * defines related to tuning
75 */
76 #ifndef CHM_MAX_BLOCKS_CACHED
77 #define CHM_MAX_BLOCKS_CACHED 5
78 #endif
79 #define CHM_PARAM_MAX_BLOCKS_CACHED 0
80
81 /*
82 * architecture specific defines
83 *
84 * Note: as soon as C99 is more widespread, the below defines should
85 * probably just use the C99 sized-int types.
86 *
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.
90 */
91
92 /* i386, 32-bit, Windows */
93 typedef BYTE UChar;
94 typedef SHORT Int16;
95 typedef USHORT UInt16;
96 typedef LONG Int32;
97 typedef DWORD UInt32;
98 typedef LONGLONG Int64;
99 typedef ULONGLONG UInt64;
100
101 /* utilities for unmarshalling data */
102 static int _unmarshal_char_array(unsigned char **pData,
103 unsigned int *pLenRemain,
104 char *dest,
105 int count)
106 {
107 if (count <= 0 || (unsigned int)count > *pLenRemain)
108 return 0;
109 memcpy(dest, (*pData), count);
110 *pData += count;
111 *pLenRemain -= count;
112 return 1;
113 }
114
115 static int _unmarshal_uchar_array(unsigned char **pData,
116 unsigned int *pLenRemain,
117 unsigned char *dest,
118 int count)
119 {
120 if (count <= 0 || (unsigned int)count > *pLenRemain)
121 return 0;
122 memcpy(dest, (*pData), count);
123 *pData += count;
124 *pLenRemain -= count;
125 return 1;
126 }
127
128 static int _unmarshal_int32(unsigned char **pData,
129 unsigned int *pLenRemain,
130 Int32 *dest)
131 {
132 if (4 > *pLenRemain)
133 return 0;
134 *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
135 *pData += 4;
136 *pLenRemain -= 4;
137 return 1;
138 }
139
140 static int _unmarshal_uint32(unsigned char **pData,
141 unsigned int *pLenRemain,
142 UInt32 *dest)
143 {
144 if (4 > *pLenRemain)
145 return 0;
146 *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
147 *pData += 4;
148 *pLenRemain -= 4;
149 return 1;
150 }
151
152 static int _unmarshal_int64(unsigned char **pData,
153 unsigned int *pLenRemain,
154 Int64 *dest)
155 {
156 Int64 temp;
157 int i;
158 if (8 > *pLenRemain)
159 return 0;
160 temp=0;
161 for(i=8; i>0; i--)
162 {
163 temp <<= 8;
164 temp |= (*pData)[i-1];
165 }
166 *dest = temp;
167 *pData += 8;
168 *pLenRemain -= 8;
169 return 1;
170 }
171
172 static int _unmarshal_uint64(unsigned char **pData,
173 unsigned int *pLenRemain,
174 UInt64 *dest)
175 {
176 UInt64 temp;
177 int i;
178 if (8 > *pLenRemain)
179 return 0;
180 temp=0;
181 for(i=8; i>0; i--)
182 {
183 temp <<= 8;
184 temp |= (*pData)[i-1];
185 }
186 *dest = temp;
187 *pData += 8;
188 *pLenRemain -= 8;
189 return 1;
190 }
191
192 static int _unmarshal_uuid(unsigned char **pData,
193 unsigned int *pDataLen,
194 unsigned char *dest)
195 {
196 return _unmarshal_uchar_array(pData, pDataLen, dest, 16);
197 }
198
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
210 };
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
216 };
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
222 };
223
224 /*
225 * structures local to this module
226 */
227
228 /* structure of ITSF headers */
229 #define _CHM_ITSF_V2_LEN (0x58)
230 #define _CHM_ITSF_V3_LEN (0x60)
231 struct chmItsfHeader
232 {
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))); */
247
248 static int _unmarshal_itsf_header(unsigned char **pData,
249 unsigned int *pDataLen,
250 struct chmItsfHeader *dest)
251 {
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)
254 return 0;
255
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);
269
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.
273 */
274 if (memcmp(dest->signature, "ITSF", 4) != 0)
275 return 0;
276 if (dest->version == 2)
277 {
278 if (dest->header_len < _CHM_ITSF_V2_LEN)
279 return 0;
280 }
281 else if (dest->version == 3)
282 {
283 if (dest->header_len < _CHM_ITSF_V3_LEN)
284 return 0;
285 }
286 else
287 return 0;
288
289 /* now, if we have a V3 structure, unmarshal the rest.
290 * otherwise, compute it
291 */
292 if (dest->version == 3)
293 {
294 if (*pDataLen != 0)
295 _unmarshal_uint64(pData, pDataLen, &dest->data_offset);
296 else
297 return 0;
298 }
299 else
300 dest->data_offset = dest->dir_offset + dest->dir_len;
301
302 return 1;
303 }
304
305 /* structure of ITSP headers */
306 #define _CHM_ITSP_V1_LEN (0x54)
307 struct chmItspHeader
308 {
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))); */
325
326 static int _unmarshal_itsp_header(unsigned char **pData,
327 unsigned int *pDataLen,
328 struct chmItspHeader *dest)
329 {
330 /* we only know how to deal with a 0x54 byte structures */
331 if (*pDataLen != _CHM_ITSP_V1_LEN)
332 return 0;
333
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);
350
351 /* error check the data */
352 if (memcmp(dest->signature, "ITSP", 4) != 0)
353 return 0;
354 if (dest->version != 1)
355 return 0;
356 if (dest->header_len != _CHM_ITSP_V1_LEN)
357 return 0;
358
359 return 1;
360 }
361
362 /* structure of PMGL headers */
363 static const char _chm_pmgl_marker[4] = "PMGL";
364 #define _CHM_PMGL_LEN (0x14)
365 struct chmPmglHeader
366 {
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))); */
373
374 static int _unmarshal_pmgl_header(unsigned char **pData,
375 unsigned int *pDataLen,
376 struct chmPmglHeader *dest)
377 {
378 /* we only know how to deal with a 0x14 byte structures */
379 if (*pDataLen != _CHM_PMGL_LEN)
380 return 0;
381
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);
388
389 /* check structure */
390 if (memcmp(dest->signature, _chm_pmgl_marker, 4) != 0)
391 return 0;
392
393 return 1;
394 }
395
396 /* structure of PMGI headers */
397 static const char _chm_pmgi_marker[4] = "PMGI";
398 #define _CHM_PMGI_LEN (0x08)
399 struct chmPmgiHeader
400 {
401 char signature[4]; /* 0 (PMGI) */
402 UInt32 free_space; /* 4 */
403 }; /* __attribute__ ((aligned (1))); */
404
405 static int _unmarshal_pmgi_header(unsigned char **pData,
406 unsigned int *pDataLen,
407 struct chmPmgiHeader *dest)
408 {
409 /* we only know how to deal with a 0x8 byte structures */
410 if (*pDataLen != _CHM_PMGI_LEN)
411 return 0;
412
413 /* unmarshal fields */
414 _unmarshal_char_array(pData, pDataLen, dest->signature, 4);
415 _unmarshal_uint32 (pData, pDataLen, &dest->free_space);
416
417 /* check structure */
418 if (memcmp(dest->signature, _chm_pmgi_marker, 4) != 0)
419 return 0;
420
421 return 1;
422 }
423
424 /* structure of LZXC reset table */
425 #define _CHM_LZXC_RESETTABLE_V1_LEN (0x28)
426 struct chmLzxcResetTable
427 {
428 UInt32 version;
429 UInt32 block_count;
430 UInt32 unknown;
431 UInt32 table_offset;
432 UInt64 uncompressed_len;
433 UInt64 compressed_len;
434 UInt64 block_len;
435 }; /* __attribute__ ((aligned (1))); */
436
437 static int _unmarshal_lzxc_reset_table(unsigned char **pData,
438 unsigned int *pDataLen,
439 struct chmLzxcResetTable *dest)
440 {
441 /* we only know how to deal with a 0x28 byte structures */
442 if (*pDataLen != _CHM_LZXC_RESETTABLE_V1_LEN)
443 return 0;
444
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);
453
454 /* check structure */
455 if (dest->version != 2)
456 return 0;
457
458 return 1;
459 }
460
461 /* structure of LZXC control data block */
462 #define _CHM_LZXC_MIN_LEN (0x18)
463 #define _CHM_LZXC_V2_LEN (0x1c)
464 struct chmLzxcControlData
465 {
466 UInt32 size; /* 0 */
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 */
473 };
474
475 static int _unmarshal_lzxc_control_data(unsigned char **pData,
476 unsigned int *pDataLen,
477 struct chmLzxcControlData *dest)
478 {
479 /* we want at least 0x18 bytes */
480 if (*pDataLen < _CHM_LZXC_MIN_LEN)
481 return 0;
482
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);
490
491 if (*pDataLen >= _CHM_LZXC_V2_LEN)
492 _unmarshal_uint32 (pData, pDataLen, &dest->unknown_18);
493 else
494 dest->unknown_18 = 0;
495
496 if (dest->version == 2)
497 {
498 dest->resetInterval *= 0x8000;
499 dest->windowSize *= 0x8000;
500 }
501 if (dest->windowSize == 0 || dest->resetInterval == 0)
502 return 0;
503
504 /* for now, only support resetInterval a multiple of windowSize/2 */
505 if (dest->windowSize == 1)
506 return 0;
507 if ((dest->resetInterval % (dest->windowSize/2)) != 0)
508 return 0;
509
510 /* check structure */
511 if (memcmp(dest->signature, "LZXC", 4) != 0)
512 return 0;
513
514 return 1;
515 }
516
517 /* the structure used for chm file handles */
518 struct chmFile
519 {
520 HANDLE fd;
521
522 CRITICAL_SECTION mutex;
523 CRITICAL_SECTION lzx_mutex;
524 CRITICAL_SECTION cache_mutex;
525
526 UInt64 dir_offset;
527 UInt64 dir_len;
528 UInt64 data_offset;
529 Int32 index_root;
530 Int32 index_head;
531 UInt32 block_len;
532
533 UInt64 span;
534 struct chmUnitInfo rt_unit;
535 struct chmUnitInfo cn_unit;
536 struct chmLzxcResetTable reset_table;
537
538 /* LZX control data */
539 int compression_enabled;
540 UInt32 window_size;
541 UInt32 reset_interval;
542 UInt32 reset_blkcount;
543
544 /* decompressor state */
545 struct LZXstate *lzx_state;
546 int lzx_last_block;
547
548 /* cache for decompressed blocks */
549 UChar **cache_blocks;
550 Int64 *cache_block_indices;
551 Int32 cache_num_blocks;
552 };
553
554 /*
555 * utility functions local to this module
556 */
557
558 /* utility function to handle differences between {pread,read}(64)? */
559 static Int64 _chm_fetch_bytes(struct chmFile *h,
560 UChar *buf,
561 UInt64 os,
562 Int64 len)
563 {
564 Int64 readLen=0;
565 if (h->fd == CHM_NULL_FD)
566 return readLen;
567
568 CHM_ACQUIRE_LOCK(h->mutex);
569 /* NOTE: this might be better done with CreateFileMapping, et cetera... */
570 {
571 LARGE_INTEGER old_pos, new_pos;
572 DWORD actualLen=0;
573
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 );
579
580 /* read the data */
581 if (ReadFile(h->fd,
582 buf,
583 (DWORD)len,
584 &actualLen,
585 NULL))
586 readLen = actualLen;
587 else
588 readLen = 0;
589
590 /* restore original position */
591 SetFilePointerEx( h->fd, old_pos, NULL, FILE_BEGIN );
592 }
593 CHM_RELEASE_LOCK(h->mutex);
594 return readLen;
595 }
596
597 /*
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.
605 */
606 static void chm_set_param(struct chmFile *h,
607 int paramType,
608 int paramVal)
609 {
610 switch (paramType)
611 {
612 case CHM_PARAM_MAX_BLOCKS_CACHED:
613 CHM_ACQUIRE_LOCK(h->cache_mutex);
614 if (paramVal != h->cache_num_blocks)
615 {
616 UChar **newBlocks;
617 Int64 *newIndices;
618 int i;
619
620 /* allocate new cached blocks */
621 newBlocks = malloc(paramVal * sizeof (UChar *));
622 newIndices = malloc(paramVal * sizeof (UInt64));
623 for (i=0; i<paramVal; i++)
624 {
625 newBlocks[i] = NULL;
626 newIndices[i] = 0;
627 }
628
629 /* re-distribute old cached blocks */
630 if (h->cache_blocks)
631 {
632 for (i=0; i<h->cache_num_blocks; i++)
633 {
634 int newSlot = (int)(h->cache_block_indices[i] % paramVal);
635
636 if (h->cache_blocks[i])
637 {
638 /* in case of collision, destroy newcomer */
639 if (newBlocks[newSlot])
640 {
641 free(h->cache_blocks[i]);
642 h->cache_blocks[i] = NULL;
643 }
644 else
645 {
646 newBlocks[newSlot] = h->cache_blocks[i];
647 newIndices[newSlot] =
648 h->cache_block_indices[i];
649 }
650 }
651 }
652
653 free(h->cache_blocks);
654 free(h->cache_block_indices);
655 }
656
657 /* now, set new values */
658 h->cache_blocks = newBlocks;
659 h->cache_block_indices = newIndices;
660 h->cache_num_blocks = paramVal;
661 }
662 CHM_RELEASE_LOCK(h->cache_mutex);
663 break;
664
665 default:
666 break;
667 }
668 }
669
670 /* open an ITS archive */
671 struct chmFile *chm_openW(const WCHAR *filename)
672 {
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;
679 #if 0
680 struct chmUnitInfo uiSpan;
681 #endif
682 struct chmUnitInfo uiLzxc;
683 struct chmLzxcControlData ctlData;
684
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;
692
693 /* open file */
694 if ((newHandle->fd=CreateFileW(filename,
695 GENERIC_READ,
696 FILE_SHARE_READ,
697 NULL,
698 OPEN_EXISTING,
699 FILE_ATTRIBUTE_NORMAL,
700 NULL)) == CHM_NULL_FD)
701 {
702 free(newHandle);
703 return NULL;
704 }
705
706 /* initialize mutexes, if needed */
707 InitializeCriticalSection(&newHandle->mutex);
708 InitializeCriticalSection(&newHandle->lzx_mutex);
709 InitializeCriticalSection(&newHandle->cache_mutex);
710
711 /* read and verify header */
712 sremain = _CHM_ITSF_V3_LEN;
713 sbufpos = sbuffer;
714 if (_chm_fetch_bytes(newHandle, sbuffer, (UInt64)0, sremain) != sremain ||
715 !_unmarshal_itsf_header(&sbufpos, &sremain, &itsfHeader))
716 {
717 chm_close(newHandle);
718 return NULL;
719 }
720
721 /* stash important values from header */
722 newHandle->dir_offset = itsfHeader.dir_offset;
723 newHandle->dir_len = itsfHeader.dir_len;
724 newHandle->data_offset = itsfHeader.data_offset;
725
726 /* now, read and verify the directory header chunk */
727 sremain = _CHM_ITSP_V1_LEN;
728 sbufpos = sbuffer;
729 if (_chm_fetch_bytes(newHandle, sbuffer,
730 (UInt64)itsfHeader.dir_offset, sremain) != sremain ||
731 !_unmarshal_itsp_header(&sbufpos, &sremain, &itspHeader))
732 {
733 chm_close(newHandle);
734 return NULL;
735 }
736
737 /* grab essential information from ITSP header */
738 newHandle->dir_offset += itspHeader.header_len;
739 newHandle->dir_len -= itspHeader.header_len;
740 newHandle->index_root = itspHeader.index_root;
741 newHandle->index_head = itspHeader.index_head;
742 newHandle->block_len = itspHeader.block_len;
743
744 /* if the index root is -1, this means we don't have any PMGI blocks.
745 * as a result, we must use the sole PMGL block as the index root
746 */
747 if (newHandle->index_root == -1)
748 newHandle->index_root = newHandle->index_head;
749
750 /* By default, compression is enabled. */
751 newHandle->compression_enabled = 1;
752
753 /* prefetch most commonly needed unit infos */
754 if (CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
755 _CHMU_RESET_TABLE,
756 &newHandle->rt_unit) ||
757 newHandle->rt_unit.space == CHM_COMPRESSED ||
758 CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
759 _CHMU_CONTENT,
760 &newHandle->cn_unit) ||
761 newHandle->cn_unit.space == CHM_COMPRESSED ||
762 CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
763 _CHMU_LZXC_CONTROLDATA,
764 &uiLzxc) ||
765 uiLzxc.space == CHM_COMPRESSED)
766 {
767 newHandle->compression_enabled = 0;
768 }
769
770 /* read reset table info */
771 if (newHandle->compression_enabled)
772 {
773 sremain = _CHM_LZXC_RESETTABLE_V1_LEN;
774 sbufpos = sbuffer;
775 if (chm_retrieve_object(newHandle, &newHandle->rt_unit, sbuffer,
776 0, sremain) != sremain ||
777 !_unmarshal_lzxc_reset_table(&sbufpos, &sremain,
778 &newHandle->reset_table))
779 {
780 newHandle->compression_enabled = 0;
781 }
782 }
783
784 /* read control data */
785 if (newHandle->compression_enabled)
786 {
787 sremain = (unsigned long)uiLzxc.length;
788 sbufpos = sbuffer;
789 if (chm_retrieve_object(newHandle, &uiLzxc, sbuffer,
790 0, sremain) != sremain ||
791 !_unmarshal_lzxc_control_data(&sbufpos, &sremain,
792 &ctlData))
793 {
794 newHandle->compression_enabled = 0;
795 }
796
797 newHandle->window_size = ctlData.windowSize;
798 newHandle->reset_interval = ctlData.resetInterval;
799
800 /* Jed, Mon Jun 28: Experimentally, it appears that the reset block count */
801 /* must be multiplied by this formerly unknown ctrl data field in */
802 /* order to decompress some files. */
803 #if 0
804 newHandle->reset_blkcount = newHandle->reset_interval /
805 (newHandle->window_size / 2);
806 #else
807 newHandle->reset_blkcount = newHandle->reset_interval /
808 (newHandle->window_size / 2) *
809 ctlData.windowsPerReset;
810 #endif
811 }
812
813 /* initialize cache */
814 chm_set_param(newHandle, CHM_PARAM_MAX_BLOCKS_CACHED,
815 CHM_MAX_BLOCKS_CACHED);
816
817 return newHandle;
818 }
819
820 /* close an ITS archive */
821 void chm_close(struct chmFile *h)
822 {
823 if (h != NULL)
824 {
825 if (h->fd != CHM_NULL_FD)
826 CHM_CLOSE_FILE(h->fd);
827 h->fd = CHM_NULL_FD;
828
829 DeleteCriticalSection(&h->mutex);
830 DeleteCriticalSection(&h->lzx_mutex);
831 DeleteCriticalSection(&h->cache_mutex);
832
833 if (h->lzx_state)
834 LZXteardown(h->lzx_state);
835 h->lzx_state = NULL;
836
837 if (h->cache_blocks)
838 {
839 int i;
840 for (i=0; i<h->cache_num_blocks; i++)
841 {
842 if (h->cache_blocks[i])
843 free(h->cache_blocks[i]);
844 }
845 free(h->cache_blocks);
846 h->cache_blocks = NULL;
847 }
848
849 free(h->cache_block_indices);
850 h->cache_block_indices = NULL;
851
852 free(h);
853 }
854 }
855
856 /*
857 * helper methods for chm_resolve_object
858 */
859
860 /* skip a compressed dword */
861 static void _chm_skip_cword(UChar **pEntry)
862 {
863 while (*(*pEntry)++ >= 0x80)
864 ;
865 }
866
867 /* skip the data from a PMGL entry */
868 static void _chm_skip_PMGL_entry_data(UChar **pEntry)
869 {
870 _chm_skip_cword(pEntry);
871 _chm_skip_cword(pEntry);
872 _chm_skip_cword(pEntry);
873 }
874
875 /* parse a compressed dword */
876 static UInt64 _chm_parse_cword(UChar **pEntry)
877 {
878 UInt64 accum = 0;
879 UChar temp;
880 while ((temp=*(*pEntry)++) >= 0x80)
881 {
882 accum <<= 7;
883 accum += temp & 0x7f;
884 }
885
886 return (accum << 7) + temp;
887 }
888
889 /* parse a utf-8 string into an ASCII char buffer */
890 static int _chm_parse_UTF8(UChar **pEntry, UInt64 count, WCHAR *path)
891 {
892 /* MJM - Modified to return real Unicode strings */
893 while (count != 0)
894 {
895 *path++ = (*(*pEntry)++);
896 --count;
897 }
898
899 *path = '\0';
900 return 1;
901 }
902
903 /* parse a PMGL entry into a chmUnitInfo struct; return 1 on success. */
904 static int _chm_parse_PMGL_entry(UChar **pEntry, struct chmUnitInfo *ui)
905 {
906 UInt64 strLen;
907
908 /* parse str len */
909 strLen = _chm_parse_cword(pEntry);
910 if (strLen > CHM_MAX_PATHLEN)
911 return 0;
912
913 /* parse path */
914 if (! _chm_parse_UTF8(pEntry, strLen, ui->path))
915 return 0;
916
917 /* parse info */
918 ui->space = (int)_chm_parse_cword(pEntry);
919 ui->start = _chm_parse_cword(pEntry);
920 ui->length = _chm_parse_cword(pEntry);
921 return 1;
922 }
923
924 /* find an exact entry in PMGL; return NULL if we fail */
925 static UChar *_chm_find_in_PMGL(UChar *page_buf,
926 UInt32 block_len,
927 const WCHAR *objPath)
928 {
929 /* XXX: modify this to do a binary search using the nice index structure
930 * that is provided for us.
931 */
932 struct chmPmglHeader header;
933 UInt32 hremain;
934 UChar *end;
935 UChar *cur;
936 UChar *temp;
937 UInt64 strLen;
938 WCHAR buffer[CHM_MAX_PATHLEN+1];
939
940 /* figure out where to start and end */
941 cur = page_buf;
942 hremain = _CHM_PMGL_LEN;
943 if (! _unmarshal_pmgl_header(&cur, &hremain, &header))
944 return NULL;
945 end = page_buf + block_len - (header.free_space);
946
947 /* now, scan progressively */
948 while (cur < end)
949 {
950 /* grab the name */
951 temp = cur;
952 strLen = _chm_parse_cword(&cur);
953 if (! _chm_parse_UTF8(&cur, strLen, buffer))
954 return NULL;
955
956 /* check if it is the right name */
957 if (! strcmpiW(buffer, objPath))
958 return temp;
959
960 _chm_skip_PMGL_entry_data(&cur);
961 }
962
963 return NULL;
964 }
965
966 /* find which block should be searched next for the entry; -1 if no block */
967 static Int32 _chm_find_in_PMGI(UChar *page_buf,
968 UInt32 block_len,
969 const WCHAR *objPath)
970 {
971 /* XXX: modify this to do a binary search using the nice index structure
972 * that is provided for us
973 */
974 struct chmPmgiHeader header;
975 UInt32 hremain;
976 int page=-1;
977 UChar *end;
978 UChar *cur;
979 UInt64 strLen;
980 WCHAR buffer[CHM_MAX_PATHLEN+1];
981
982 /* figure out where to start and end */
983 cur = page_buf;
984 hremain = _CHM_PMGI_LEN;
985 if (! _unmarshal_pmgi_header(&cur, &hremain, &header))
986 return -1;
987 end = page_buf + block_len - (header.free_space);
988
989 /* now, scan progressively */
990 while (cur < end)
991 {
992 /* grab the name */
993 strLen = _chm_parse_cword(&cur);
994 if (! _chm_parse_UTF8(&cur, strLen, buffer))
995 return -1;
996
997 /* check if it is the right name */
998 if (strcmpiW(buffer, objPath) > 0)
999 return page;
1000
1001 /* load next value for path */
1002 page = (int)_chm_parse_cword(&cur);
1003 }
1004
1005 return page;
1006 }
1007
1008 /* resolve a particular object from the archive */
1009 int chm_resolve_object(struct chmFile *h,
1010 const WCHAR *objPath,
1011 struct chmUnitInfo *ui)
1012 {
1013 /*
1014 * XXX: implement caching scheme for dir pages
1015 */
1016
1017 Int32 curPage;
1018
1019 /* buffer to hold whatever page we're looking at */
1020 UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, h->block_len);
1021
1022 /* starting page */
1023 curPage = h->index_root;
1024
1025 /* until we have either returned or given up */
1026 while (curPage != -1)
1027 {
1028
1029 /* try to fetch the index page */
1030 if (_chm_fetch_bytes(h, page_buf,
1031 (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1032 h->block_len) != h->block_len)
1033 {
1034 HeapFree(GetProcessHeap(), 0, page_buf);
1035 return CHM_RESOLVE_FAILURE;
1036 }
1037
1038 /* now, if it is a leaf node: */
1039 if (memcmp(page_buf, _chm_pmgl_marker, 4) == 0)
1040 {
1041 /* scan block */
1042 UChar *pEntry = _chm_find_in_PMGL(page_buf,
1043 h->block_len,
1044 objPath);
1045 if (pEntry == NULL)
1046 {
1047 HeapFree(GetProcessHeap(), 0, page_buf);
1048 return CHM_RESOLVE_FAILURE;
1049 }
1050
1051 /* parse entry and return */
1052 _chm_parse_PMGL_entry(&pEntry, ui);
1053 HeapFree(GetProcessHeap(), 0, page_buf);
1054 return CHM_RESOLVE_SUCCESS;
1055 }
1056
1057 /* else, if it is a branch node: */
1058 else if (memcmp(page_buf, _chm_pmgi_marker, 4) == 0)
1059 curPage = _chm_find_in_PMGI(page_buf, h->block_len, objPath);
1060
1061 /* else, we are confused. give up. */
1062 else
1063 {
1064 HeapFree(GetProcessHeap(), 0, page_buf);
1065 return CHM_RESOLVE_FAILURE;
1066 }
1067 }
1068
1069 /* didn't find anything. fail. */
1070 HeapFree(GetProcessHeap(), 0, page_buf);
1071 return CHM_RESOLVE_FAILURE;
1072 }
1073
1074 /*
1075 * utility methods for dealing with compressed data
1076 */
1077
1078 /* get the bounds of a compressed block. return 0 on failure */
1079 static int _chm_get_cmpblock_bounds(struct chmFile *h,
1080 UInt64 block,
1081 UInt64 *start,
1082 Int64 *len)
1083 {
1084 UChar buffer[8], *dummy;
1085 UInt32 remain;
1086
1087 /* for all but the last block, use the reset table */
1088 if (block < h->reset_table.block_count-1)
1089 {
1090 /* unpack the start address */
1091 dummy = buffer;
1092 remain = 8;
1093 if (_chm_fetch_bytes(h, buffer,
1094 (UInt64)h->data_offset
1095 + (UInt64)h->rt_unit.start
1096 + (UInt64)h->reset_table.table_offset
1097 + (UInt64)block*8,
1098 remain) != remain ||
1099 !_unmarshal_uint64(&dummy, &remain, start))
1100 return 0;
1101
1102 /* unpack the end address */
1103 dummy = buffer;
1104 remain = 8;
1105 if (_chm_fetch_bytes(h, buffer,
1106 (UInt64)h->data_offset
1107 + (UInt64)h->rt_unit.start
1108 + (UInt64)h->reset_table.table_offset
1109 + (UInt64)block*8 + 8,
1110 remain) != remain ||
1111 !_unmarshal_int64(&dummy, &remain, len))
1112 return 0;
1113 }
1114
1115 /* for the last block, use the span in addition to the reset table */
1116 else
1117 {
1118 /* unpack the start address */
1119 dummy = buffer;
1120 remain = 8;
1121 if (_chm_fetch_bytes(h, buffer,
1122 (UInt64)h->data_offset
1123 + (UInt64)h->rt_unit.start
1124 + (UInt64)h->reset_table.table_offset
1125 + (UInt64)block*8,
1126 remain) != remain ||
1127 !_unmarshal_uint64(&dummy, &remain, start))
1128 return 0;
1129
1130 *len = h->reset_table.compressed_len;
1131 }
1132
1133 /* compute the length and absolute start address */
1134 *len -= *start;
1135 *start += h->data_offset + h->cn_unit.start;
1136
1137 return 1;
1138 }
1139
1140 /* decompress the block. must have lzx_mutex. */
1141 static Int64 _chm_decompress_block(struct chmFile *h,
1142 UInt64 block,
1143 UChar **ubuffer)
1144 {
1145 UChar *cbuffer = HeapAlloc( GetProcessHeap(), 0,
1146 ((unsigned int)h->reset_table.block_len + 6144));
1147 UInt64 cmpStart; /* compressed start */
1148 Int64 cmpLen; /* compressed len */
1149 int indexSlot; /* cache index slot */
1150 UChar *lbuffer; /* local buffer ptr */
1151 UInt32 blockAlign = (UInt32)(block % h->reset_blkcount); /* reset intvl. aln. */
1152 UInt32 i; /* local loop index */
1153
1154 /* let the caching system pull its weight! */
1155 if (block - blockAlign <= h->lzx_last_block &&
1156 block >= h->lzx_last_block)
1157 blockAlign = (block - h->lzx_last_block);
1158
1159 /* check if we need previous blocks */
1160 if (blockAlign != 0)
1161 {
1162 /* fetch all required previous blocks since last reset */
1163 for (i = blockAlign; i > 0; i--)
1164 {
1165 UInt32 curBlockIdx = block - i;
1166
1167 /* check if we most recently decompressed the previous block */
1168 if (h->lzx_last_block != curBlockIdx)
1169 {
1170 if ((curBlockIdx % h->reset_blkcount) == 0)
1171 {
1172 #ifdef CHM_DEBUG
1173 fprintf(stderr, "***RESET (1)***\n");
1174 #endif
1175 LZXreset(h->lzx_state);
1176 }
1177
1178 indexSlot = (int)((curBlockIdx) % h->cache_num_blocks);
1179 h->cache_block_indices[indexSlot] = curBlockIdx;
1180 if (! h->cache_blocks[indexSlot])
1181 h->cache_blocks[indexSlot] = malloc( (unsigned int)(h->reset_table.block_len));
1182 lbuffer = h->cache_blocks[indexSlot];
1183
1184 /* decompress the previous block */
1185 #ifdef CHM_DEBUG
1186 fprintf(stderr, "Decompressing block #%4d (EXTRA)\n", curBlockIdx);
1187 #endif
1188 if (!_chm_get_cmpblock_bounds(h, curBlockIdx, &cmpStart, &cmpLen) ||
1189 _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen ||
1190 LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1191 (int)h->reset_table.block_len) != DECR_OK)
1192 {
1193 #ifdef CHM_DEBUG
1194 fprintf(stderr, " (DECOMPRESS FAILED!)\n");
1195 #endif
1196 HeapFree(GetProcessHeap(), 0, cbuffer);
1197 return (Int64)0;
1198 }
1199
1200 h->lzx_last_block = (int)curBlockIdx;
1201 }
1202 }
1203 }
1204 else
1205 {
1206 if ((block % h->reset_blkcount) == 0)
1207 {
1208 #ifdef CHM_DEBUG
1209 fprintf(stderr, "***RESET (2)***\n");
1210 #endif
1211 LZXreset(h->lzx_state);
1212 }
1213 }
1214
1215 /* allocate slot in cache */
1216 indexSlot = (int)(block % h->cache_num_blocks);
1217 h->cache_block_indices[indexSlot] = block;
1218 if (! h->cache_blocks[indexSlot])
1219 h->cache_blocks[indexSlot] = malloc( ((unsigned int)h->reset_table.block_len));
1220 lbuffer = h->cache_blocks[indexSlot];
1221 *ubuffer = lbuffer;
1222
1223 /* decompress the block we actually want */
1224 #ifdef CHM_DEBUG
1225 fprintf(stderr, "Decompressing block #%4d (REAL )\n", block);
1226 #endif
1227 if (! _chm_get_cmpblock_bounds(h, block, &cmpStart, &cmpLen) ||
1228 _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen ||
1229 LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1230 (int)h->reset_table.block_len) != DECR_OK)
1231 {
1232 #ifdef CHM_DEBUG
1233 fprintf(stderr, " (DECOMPRESS FAILED!)\n");
1234 #endif
1235 HeapFree(GetProcessHeap(), 0, cbuffer);
1236 return (Int64)0;
1237 }
1238 h->lzx_last_block = (int)block;
1239
1240 /* XXX: modify LZX routines to return the length of the data they
1241 * decompressed and return that instead, for an extra sanity check.
1242 */
1243 HeapFree(GetProcessHeap(), 0, cbuffer);
1244 return h->reset_table.block_len;
1245 }
1246
1247 /* grab a region from a compressed block */
1248 static Int64 _chm_decompress_region(struct chmFile *h,
1249 UChar *buf,
1250 UInt64 start,
1251 Int64 len)
1252 {
1253 UInt64 nBlock, nOffset;
1254 UInt64 nLen;
1255 UInt64 gotLen;
1256 UChar *ubuffer = NULL;
1257
1258 if (len <= 0)
1259 return (Int64)0;
1260
1261 /* figure out what we need to read */
1262 nBlock = start / h->reset_table.block_len;
1263 nOffset = start % h->reset_table.block_len;
1264 nLen = len;
1265 if (nLen > (h->reset_table.block_len - nOffset))
1266 nLen = h->reset_table.block_len - nOffset;
1267
1268 /* if block is cached, return data from it. */
1269 CHM_ACQUIRE_LOCK(h->lzx_mutex);
1270 CHM_ACQUIRE_LOCK(h->cache_mutex);
1271 if (h->cache_block_indices[nBlock % h->cache_num_blocks] == nBlock &&
1272 h->cache_blocks[nBlock % h->cache_num_blocks] != NULL)
1273 {
1274 memcpy(buf,
1275 h->cache_blocks[nBlock % h->cache_num_blocks] + nOffset,
1276 (unsigned int)nLen);
1277 CHM_RELEASE_LOCK(h->cache_mutex);
1278 CHM_RELEASE_LOCK(h->lzx_mutex);
1279 return nLen;
1280 }
1281 CHM_RELEASE_LOCK(h->cache_mutex);
1282
1283 /* data request not satisfied, so... start up the decompressor machine */
1284 if (! h->lzx_state)
1285 {
1286 int window_size = ffs(h->window_size) - 1;
1287 h->lzx_last_block = -1;
1288 h->lzx_state = LZXinit(window_size);
1289 }
1290
1291 /* decompress some data */
1292 gotLen = _chm_decompress_block(h, nBlock, &ubuffer);
1293 if (gotLen < nLen)
1294 nLen = gotLen;
1295 memcpy(buf, ubuffer+nOffset, (unsigned int)nLen);
1296 CHM_RELEASE_LOCK(h->lzx_mutex);
1297 return nLen;
1298 }
1299
1300 /* retrieve (part of) an object */
1301 LONGINT64 chm_retrieve_object(struct chmFile *h,
1302 struct chmUnitInfo *ui,
1303 unsigned char *buf,
1304 LONGUINT64 addr,
1305 LONGINT64 len)
1306 {
1307 /* must be valid file handle */
1308 if (h == NULL)
1309 return (Int64)0;
1310
1311 /* starting address must be in correct range */
1312 if (addr < 0 || addr >= ui->length)
1313 return (Int64)0;
1314
1315 /* clip length */
1316 if (addr + len > ui->length)
1317 len = ui->length - addr;
1318
1319 /* if the file is uncompressed, it's simple */
1320 if (ui->space == CHM_UNCOMPRESSED)
1321 {
1322 /* read data */
1323 return _chm_fetch_bytes(h,
1324 buf,
1325 (UInt64)h->data_offset + (UInt64)ui->start + (UInt64)addr,
1326 len);
1327 }
1328
1329 /* else if the file is compressed, it's a little trickier */
1330 else /* ui->space == CHM_COMPRESSED */
1331 {
1332 Int64 swath=0, total=0;
1333
1334 /* if compression is not enabled for this file... */
1335 if (! h->compression_enabled)
1336 return total;
1337
1338 do {
1339
1340 /* swill another mouthful */
1341 swath = _chm_decompress_region(h, buf, ui->start + addr, len);
1342
1343 /* if we didn't get any... */
1344 if (swath == 0)
1345 return total;
1346
1347 /* update stats */
1348 total += swath;
1349 len -= swath;
1350 addr += swath;
1351 buf += swath;
1352
1353 } while (len != 0);
1354
1355 return total;
1356 }
1357 }
1358
1359 /* enumerate the objects in the .chm archive */
1360 int chm_enumerate(struct chmFile *h,
1361 int what,
1362 CHM_ENUMERATOR e,
1363 void *context)
1364 {
1365 Int32 curPage;
1366
1367 /* buffer to hold whatever page we're looking at */
1368 UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, (unsigned int)h->block_len);
1369 struct chmPmglHeader header;
1370 UChar *end;
1371 UChar *cur;
1372 unsigned int lenRemain;
1373 UInt64 ui_path_len;
1374
1375 /* the current ui */
1376 struct chmUnitInfo ui;
1377 int flag;
1378
1379 /* starting page */
1380 curPage = h->index_head;
1381
1382 /* until we have either returned or given up */
1383 while (curPage != -1)
1384 {
1385
1386 /* try to fetch the index page */
1387 if (_chm_fetch_bytes(h,
1388 page_buf,
1389 (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1390 h->block_len) != h->block_len)
1391 {
1392 HeapFree(GetProcessHeap(), 0, page_buf);
1393 return 0;
1394 }
1395
1396 /* figure out start and end for this page */
1397 cur = page_buf;
1398 lenRemain = _CHM_PMGL_LEN;
1399 if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1400 {
1401 HeapFree(GetProcessHeap(), 0, page_buf);
1402 return 0;
1403 }
1404 end = page_buf + h->block_len - (header.free_space);
1405
1406 /* loop over this page */
1407 while (cur < end)
1408 {
1409 if (! _chm_parse_PMGL_entry(&cur, &ui))
1410 {
1411 HeapFree(GetProcessHeap(), 0, page_buf);
1412 return 0;
1413 }
1414
1415 /* get the length of the path */
1416 ui_path_len = strlenW(ui.path)-1;
1417
1418 /* check for DIRS */
1419 if (ui.path[ui_path_len] == '/' && !(what & CHM_ENUMERATE_DIRS))
1420 continue;
1421
1422 /* check for FILES */
1423 if (ui.path[ui_path_len] != '/' && !(what & CHM_ENUMERATE_FILES))
1424 continue;
1425
1426 /* check for NORMAL vs. META */
1427 if (ui.path[0] == '/')
1428 {
1429
1430 /* check for NORMAL vs. SPECIAL */
1431 if (ui.path[1] == '#' || ui.path[1] == '$')
1432 flag = CHM_ENUMERATE_SPECIAL;
1433 else
1434 flag = CHM_ENUMERATE_NORMAL;
1435 }
1436 else
1437 flag = CHM_ENUMERATE_META;
1438 if (! (what & flag))
1439 continue;
1440
1441 /* call the enumerator */
1442 {
1443 int status = (*e)(h, &ui, context);
1444 switch (status)
1445 {
1446 case CHM_ENUMERATOR_FAILURE:
1447 HeapFree(GetProcessHeap(), 0, page_buf);
1448 return 0;
1449 case CHM_ENUMERATOR_CONTINUE:
1450 break;
1451 case CHM_ENUMERATOR_SUCCESS:
1452 HeapFree(GetProcessHeap(), 0, page_buf);
1453 return 1;
1454 default:
1455 break;
1456 }
1457 }
1458 }
1459
1460 /* advance to next page */
1461 curPage = header.block_next;
1462 }
1463
1464 HeapFree(GetProcessHeap(), 0, page_buf);
1465 return 1;
1466 }
1467
1468 int chm_enumerate_dir(struct chmFile *h,
1469 const WCHAR *prefix,
1470 int what,
1471 CHM_ENUMERATOR e,
1472 void *context)
1473 {
1474 /*
1475 * XXX: do this efficiently (i.e. using the tree index)
1476 */
1477
1478 Int32 curPage;
1479
1480 /* buffer to hold whatever page we're looking at */
1481 UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, (unsigned int)h->block_len);
1482 struct chmPmglHeader header;
1483 UChar *end;
1484 UChar *cur;
1485 unsigned int lenRemain;
1486
1487 /* set to 1 once we've started */
1488 int it_has_begun=0;
1489
1490 /* the current ui */
1491 struct chmUnitInfo ui;
1492 int flag;
1493 UInt64 ui_path_len;
1494
1495 /* the length of the prefix */
1496 WCHAR prefixRectified[CHM_MAX_PATHLEN+1];
1497 int prefixLen;
1498 WCHAR lastPath[CHM_MAX_PATHLEN];
1499 int lastPathLen;
1500
1501 /* starting page */
1502 curPage = h->index_head;
1503
1504 /* initialize pathname state */
1505 lstrcpynW(prefixRectified, prefix, CHM_MAX_PATHLEN);
1506 prefixLen = strlenW(prefixRectified);
1507 if (prefixLen != 0)
1508 {
1509 if (prefixRectified[prefixLen-1] != '/')
1510 {
1511 prefixRectified[prefixLen] = '/';
1512 prefixRectified[prefixLen+1] = '\0';
1513 ++prefixLen;
1514 }
1515 }
1516 lastPath[0] = '\0';
1517 lastPathLen = -1;
1518
1519 /* until we have either returned or given up */
1520 while (curPage != -1)
1521 {
1522
1523 /* try to fetch the index page */
1524 if (_chm_fetch_bytes(h,
1525 page_buf,
1526 (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1527 h->block_len) != h->block_len)
1528 {
1529 HeapFree(GetProcessHeap(), 0, page_buf);
1530 return 0;
1531 }
1532
1533 /* figure out start and end for this page */
1534 cur = page_buf;
1535 lenRemain = _CHM_PMGL_LEN;
1536 if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1537 {
1538 HeapFree(GetProcessHeap(), 0, page_buf);
1539 return 0;
1540 }
1541 end = page_buf + h->block_len - (header.free_space);
1542
1543 /* loop over this page */
1544 while (cur < end)
1545 {
1546 if (! _chm_parse_PMGL_entry(&cur, &ui))
1547 {
1548 HeapFree(GetProcessHeap(), 0, page_buf);
1549 return 0;
1550 }
1551
1552 /* check if we should start */
1553 if (! it_has_begun)
1554 {
1555 if (ui.length == 0 && strncmpiW(ui.path, prefixRectified, prefixLen) == 0)
1556 it_has_begun = 1;
1557 else
1558 continue;
1559
1560 if (ui.path[prefixLen] == '\0')
1561 continue;
1562 }
1563
1564 /* check if we should stop */
1565 else
1566 {
1567 if (strncmpiW(ui.path, prefixRectified, prefixLen) != 0)
1568 {
1569 HeapFree(GetProcessHeap(), 0, page_buf);
1570 return 1;
1571 }
1572 }
1573
1574 /* check if we should include this path */
1575 if (lastPathLen != -1)
1576 {
1577 if (strncmpiW(ui.path, lastPath, lastPathLen) == 0)
1578 continue;
1579 }
1580 strcpyW(lastPath, ui.path);
1581 lastPathLen = strlenW(lastPath);
1582
1583 /* get the length of the path */
1584 ui_path_len = strlenW(ui.path)-1;
1585
1586 /* check for DIRS */
1587 if (ui.path[ui_path_len] == '/' && !(what & CHM_ENUMERATE_DIRS))
1588 continue;
1589
1590 /* check for FILES */
1591 if (ui.path[ui_path_len] != '/' && !(what & CHM_ENUMERATE_FILES))
1592 continue;
1593
1594 /* check for NORMAL vs. META */
1595 if (ui.path[0] == '/')
1596 {
1597
1598 /* check for NORMAL vs. SPECIAL */
1599 if (ui.path[1] == '#' || ui.path[1] == '$')
1600 flag = CHM_ENUMERATE_SPECIAL;
1601 else
1602 flag = CHM_ENUMERATE_NORMAL;
1603 }
1604 else
1605 flag = CHM_ENUMERATE_META;
1606 if (! (what & flag))
1607 continue;
1608
1609 /* call the enumerator */
1610 {
1611 int status = (*e)(h, &ui, context);
1612 switch (status)
1613 {
1614 case CHM_ENUMERATOR_FAILURE:
1615 HeapFree(GetProcessHeap(), 0, page_buf);
1616 return 0;
1617 case CHM_ENUMERATOR_CONTINUE:
1618 break;
1619 case CHM_ENUMERATOR_SUCCESS:
1620 HeapFree(GetProcessHeap(), 0, page_buf);
1621 return 1;
1622 default:
1623 break;
1624 }
1625 }
1626 }
1627
1628 /* advance to next page */
1629 curPage = header.block_next;
1630 }
1631
1632 HeapFree(GetProcessHeap(), 0, page_buf);
1633 return 1;
1634 }
Maarten's multimedia branch