dlls/itss/chm_lib.c

   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
  58 #include <stdarg.h>
  59 #include <stdio.h>
  60 #include <stdlib.h>
  61 #include <string.h>
  62
  63 #include "windef.h"
  64 #include "winbase.h"
  65 #include "winnls.h"
  66
  67 #include "chm_lib.h"
  68 #include "lzx.h"
  69
  70 #define CHM_ACQUIRE_LOCK(a) do {                        \
  71         EnterCriticalSection(&(a));                     \
  72     } while(0)
  73 #define CHM_RELEASE_LOCK(a) do {                        \
  74         LeaveCriticalSection(&(a));                     \
  75     } while(0)
  76
  77 #define CHM_NULL_FD (INVALID_HANDLE_VALUE)
  78 #define CHM_CLOSE_FILE(fd) CloseHandle((fd))
  79
  80 /*
  81  * defines related to tuning
  82  */
  83 #ifndef CHM_MAX_BLOCKS_CACHED
  84 #define CHM_MAX_BLOCKS_CACHED 5
  85 #endif
  86 #define CHM_PARAM_MAX_BLOCKS_CACHED 0
  87
  88 /*
  89  * architecture specific defines
  90  *
  91  * Note: as soon as C99 is more widespread, the below defines should
  92  * probably just use the C99 sized-int types.
  93  *
  94  * The following settings will probably work for many platforms.  The sizes
  95  * don't have to be exactly correct, but the types must accommodate at least as
  96  * many bits as they specify.
  97  */
  98
  99 /* i386, 32-bit, Windows */
 100 typedef BYTE   UChar;
 101 typedef SHORT  Int16;
 102 typedef USHORT UInt16;
 103 typedef LONG   Int32;
 104 typedef DWORD      UInt32;
 105 typedef LONGLONG   Int64;
 106 typedef ULONGLONG  UInt64;
 107
 108 /* utilities for unmarshalling data */
 109 static BOOL _unmarshal_char_array(unsigned char **pData,
 110                                   unsigned int *pLenRemain,
 111                                   char *dest,
 112                                   int count)
 113 {
 114     if (count <= 0  ||  (unsigned int)count > *pLenRemain)
 115         return FALSE;
 116     memcpy(dest, (*pData), count);
 117     *pData += count;
 118     *pLenRemain -= count;
 119     return TRUE;
 120 }
 121
 122 static BOOL _unmarshal_uchar_array(unsigned char **pData,
 123                                    unsigned int *pLenRemain,
 124                                    unsigned char *dest,
 125                                    int count)
 126 {
 127     if (count <= 0  || (unsigned int)count > *pLenRemain)
 128         return FALSE;
 129     memcpy(dest, (*pData), count);
 130     *pData += count;
 131     *pLenRemain -= count;
 132     return TRUE;
 133 }
 134
 135 static BOOL _unmarshal_int32(unsigned char **pData,
 136                              unsigned int *pLenRemain,
 137                              Int32 *dest)
 138 {
 139     if (4 > *pLenRemain)
 140         return FALSE;
 141     *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
 142     *pData += 4;
 143     *pLenRemain -= 4;
 144     return TRUE;
 145 }
 146
 147 static BOOL _unmarshal_uint32(unsigned char **pData,
 148                               unsigned int *pLenRemain,
 149                               UInt32 *dest)
 150 {
 151     if (4 > *pLenRemain)
 152         return FALSE;
 153     *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
 154     *pData += 4;
 155     *pLenRemain -= 4;
 156     return TRUE;
 157 }
 158
 159 static BOOL _unmarshal_int64(unsigned char **pData,
 160                              unsigned int *pLenRemain,
 161                              Int64 *dest)
 162 {
 163     Int64 temp;
 164     int i;
 165     if (8 > *pLenRemain)
 166         return FALSE;
 167     temp=0;
 168     for(i=8; i>0; i--)
 169     {
 170         temp <<= 8;
 171         temp |= (*pData)[i-1];
 172     }
 173     *dest = temp;
 174     *pData += 8;
 175     *pLenRemain -= 8;
 176     return TRUE;
 177 }
 178
 179 static BOOL _unmarshal_uint64(unsigned char **pData,
 180                               unsigned int *pLenRemain,
 181                               UInt64 *dest)
 182 {
 183     UInt64 temp;
 184     int i;
 185     if (8 > *pLenRemain)
 186         return FALSE;
 187     temp=0;
 188     for(i=8; i>0; i--)
 189     {
 190         temp <<= 8;
 191         temp |= (*pData)[i-1];
 192     }
 193     *dest = temp;
 194     *pData += 8;
 195     *pLenRemain -= 8;
 196     return TRUE;
 197 }
 198
 199 static BOOL _unmarshal_uuid(unsigned char **pData,
 200                             unsigned int *pDataLen,
 201                             unsigned char *dest)
 202 {
 203     return _unmarshal_uchar_array(pData, pDataLen, dest, 16);
 204 }
 205
 206 /*
 207  * structures local to this module
 208  */
 209
 210 /* structure of ITSF headers */
 211 #define _CHM_ITSF_V2_LEN (0x58)
 212 #define _CHM_ITSF_V3_LEN (0x60)
 213 struct chmItsfHeader
 214 {
 215     char        signature[4];           /*  0 (ITSF) */
 216     Int32       version;                /*  4 */
 217     Int32       header_len;             /*  8 */
 218     Int32       unknown_000c;           /*  c */
 219     UInt32      last_modified;          /* 10 */
 220     UInt32      lang_id;                /* 14 */
 221     UChar       dir_uuid[16];           /* 18 */
 222     UChar       stream_uuid[16];        /* 28 */
 223     UInt64      unknown_offset;         /* 38 */
 224     UInt64      unknown_len;            /* 40 */
 225     UInt64      dir_offset;             /* 48 */
 226     UInt64      dir_len;                /* 50 */
 227     UInt64      data_offset;            /* 58 (Not present before V3) */
 228 }; /* __attribute__ ((aligned (1))); */
 229
 230 static BOOL _unmarshal_itsf_header(unsigned char **pData,
 231                                    unsigned int *pDataLen,
 232                                    struct chmItsfHeader *dest)
 233 {
 234     /* we only know how to deal with the 0x58 and 0x60 byte structures */
 235     if (*pDataLen != _CHM_ITSF_V2_LEN  &&  *pDataLen != _CHM_ITSF_V3_LEN)
 236         return FALSE;
 237
 238     /* unmarshal common fields */
 239     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 240     _unmarshal_int32     (pData, pDataLen, &dest->version);
 241     _unmarshal_int32     (pData, pDataLen, &dest->header_len);
 242     _unmarshal_int32     (pData, pDataLen, &dest->unknown_000c);
 243     _unmarshal_uint32    (pData, pDataLen, &dest->last_modified);
 244     _unmarshal_uint32    (pData, pDataLen, &dest->lang_id);
 245     _unmarshal_uuid      (pData, pDataLen,  dest->dir_uuid);
 246     _unmarshal_uuid      (pData, pDataLen,  dest->stream_uuid);
 247     _unmarshal_uint64    (pData, pDataLen, &dest->unknown_offset);
 248     _unmarshal_uint64    (pData, pDataLen, &dest->unknown_len);
 249     _unmarshal_uint64    (pData, pDataLen, &dest->dir_offset);
 250     _unmarshal_uint64    (pData, pDataLen, &dest->dir_len);
 251
 252     /* error check the data */
 253     /* XXX: should also check UUIDs, probably, though with a version 3 file,
 254      * current MS tools do not seem to use them.
 255      */
 256     if (memcmp(dest->signature, "ITSF", 4) != 0)
 257         return FALSE;
 258     if (dest->version == 2)
 259     {
 260         if (dest->header_len < _CHM_ITSF_V2_LEN)
 261             return FALSE;
 262     }
 263     else if (dest->version == 3)
 264     {
 265         if (dest->header_len < _CHM_ITSF_V3_LEN)
 266             return FALSE;
 267     }
 268     else
 269         return FALSE;
 270
 271     /* now, if we have a V3 structure, unmarshal the rest.
 272      * otherwise, compute it
 273      */
 274     if (dest->version == 3)
 275     {
 276         if (*pDataLen != 0)
 277             _unmarshal_uint64(pData, pDataLen, &dest->data_offset);
 278         else
 279             return FALSE;
 280     }
 281     else
 282         dest->data_offset = dest->dir_offset + dest->dir_len;
 283
 284     return TRUE;
 285 }
 286
 287 /* structure of ITSP headers */
 288 #define _CHM_ITSP_V1_LEN (0x54)
 289 struct chmItspHeader
 290 {
 291     char        signature[4];           /*  0 (ITSP) */
 292     Int32       version;                /*  4 */
 293     Int32       header_len;             /*  8 */
 294     Int32       unknown_000c;           /*  c */
 295     UInt32      block_len;              /* 10 */
 296     Int32       blockidx_intvl;         /* 14 */
 297     Int32       index_depth;            /* 18 */
 298     Int32       index_root;             /* 1c */
 299     Int32       index_head;             /* 20 */
 300     Int32       unknown_0024;           /* 24 */
 301     UInt32      num_blocks;             /* 28 */
 302     Int32       unknown_002c;           /* 2c */
 303     UInt32      lang_id;                /* 30 */
 304     UChar       system_uuid[16];        /* 34 */
 305     UChar       unknown_0044[16];       /* 44 */
 306 }; /* __attribute__ ((aligned (1))); */
 307
 308 static BOOL _unmarshal_itsp_header(unsigned char **pData,
 309                                    unsigned int *pDataLen,
 310                                    struct chmItspHeader *dest)
 311 {
 312     /* we only know how to deal with a 0x54 byte structures */
 313     if (*pDataLen != _CHM_ITSP_V1_LEN)
 314         return FALSE;
 315
 316     /* unmarshal fields */
 317     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 318     _unmarshal_int32     (pData, pDataLen, &dest->version);
 319     _unmarshal_int32     (pData, pDataLen, &dest->header_len);
 320     _unmarshal_int32     (pData, pDataLen, &dest->unknown_000c);
 321     _unmarshal_uint32    (pData, pDataLen, &dest->block_len);
 322     _unmarshal_int32     (pData, pDataLen, &dest->blockidx_intvl);
 323     _unmarshal_int32     (pData, pDataLen, &dest->index_depth);
 324     _unmarshal_int32     (pData, pDataLen, &dest->index_root);
 325     _unmarshal_int32     (pData, pDataLen, &dest->index_head);
 326     _unmarshal_int32     (pData, pDataLen, &dest->unknown_0024);
 327     _unmarshal_uint32    (pData, pDataLen, &dest->num_blocks);
 328     _unmarshal_int32     (pData, pDataLen, &dest->unknown_002c);
 329     _unmarshal_uint32    (pData, pDataLen, &dest->lang_id);
 330     _unmarshal_uuid      (pData, pDataLen,  dest->system_uuid);
 331     _unmarshal_uchar_array(pData, pDataLen, dest->unknown_0044, 16);
 332
 333     /* error check the data */
 334     if (memcmp(dest->signature, "ITSP", 4) != 0)
 335         return FALSE;
 336     if (dest->version != 1)
 337         return FALSE;
 338     if (dest->header_len != _CHM_ITSP_V1_LEN)
 339         return FALSE;
 340
 341     return TRUE;
 342 }
 343
 344 /* structure of PMGL headers */
 345 static const char _chm_pmgl_marker[4] = "PMGL";
 346 #define _CHM_PMGL_LEN (0x14)
 347 struct chmPmglHeader
 348 {
 349     char        signature[4];           /*  0 (PMGL) */
 350     UInt32      free_space;             /*  4 */
 351     UInt32      unknown_0008;           /*  8 */
 352     Int32       block_prev;             /*  c */
 353     Int32       block_next;             /* 10 */
 354 }; /* __attribute__ ((aligned (1))); */
 355
 356 static BOOL _unmarshal_pmgl_header(unsigned char **pData,
 357                                    unsigned int *pDataLen,
 358                                    struct chmPmglHeader *dest)
 359 {
 360     /* we only know how to deal with a 0x14 byte structures */
 361     if (*pDataLen != _CHM_PMGL_LEN)
 362         return FALSE;
 363
 364     /* unmarshal fields */
 365     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 366     _unmarshal_uint32    (pData, pDataLen, &dest->free_space);
 367     _unmarshal_uint32    (pData, pDataLen, &dest->unknown_0008);
 368     _unmarshal_int32     (pData, pDataLen, &dest->block_prev);
 369     _unmarshal_int32     (pData, pDataLen, &dest->block_next);
 370
 371     /* check structure */
 372     if (memcmp(dest->signature, _chm_pmgl_marker, 4) != 0)
 373         return FALSE;
 374
 375     return TRUE;
 376 }
 377
 378 /* structure of PMGI headers */
 379 static const char _chm_pmgi_marker[4] = "PMGI";
 380 #define _CHM_PMGI_LEN (0x08)
 381 struct chmPmgiHeader
 382 {
 383     char        signature[4];           /*  0 (PMGI) */
 384     UInt32      free_space;             /*  4 */
 385 }; /* __attribute__ ((aligned (1))); */
 386
 387 static BOOL _unmarshal_pmgi_header(unsigned char **pData,
 388                                    unsigned int *pDataLen,
 389                                    struct chmPmgiHeader *dest)
 390 {
 391     /* we only know how to deal with a 0x8 byte structures */
 392     if (*pDataLen != _CHM_PMGI_LEN)
 393         return FALSE;
 394
 395     /* unmarshal fields */
 396     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 397     _unmarshal_uint32    (pData, pDataLen, &dest->free_space);
 398
 399     /* check structure */
 400     if (memcmp(dest->signature, _chm_pmgi_marker, 4) != 0)
 401         return FALSE;
 402
 403     return TRUE;
 404 }
 405
 406 /* structure of LZXC reset table */
 407 #define _CHM_LZXC_RESETTABLE_V1_LEN (0x28)
 408 struct chmLzxcResetTable
 409 {
 410     UInt32      version;
 411     UInt32      block_count;
 412     UInt32      unknown;
 413     UInt32      table_offset;
 414     UInt64      uncompressed_len;
 415     UInt64      compressed_len;
 416     UInt64      block_len;
 417 }; /* __attribute__ ((aligned (1))); */
 418
 419 static BOOL _unmarshal_lzxc_reset_table(unsigned char **pData,
 420                                         unsigned int *pDataLen,
 421                                         struct chmLzxcResetTable *dest)
 422 {
 423     /* we only know how to deal with a 0x28 byte structures */
 424     if (*pDataLen != _CHM_LZXC_RESETTABLE_V1_LEN)
 425         return FALSE;
 426
 427     /* unmarshal fields */
 428     _unmarshal_uint32    (pData, pDataLen, &dest->version);
 429     _unmarshal_uint32    (pData, pDataLen, &dest->block_count);
 430     _unmarshal_uint32    (pData, pDataLen, &dest->unknown);
 431     _unmarshal_uint32    (pData, pDataLen, &dest->table_offset);
 432     _unmarshal_uint64    (pData, pDataLen, &dest->uncompressed_len);
 433     _unmarshal_uint64    (pData, pDataLen, &dest->compressed_len);
 434     _unmarshal_uint64    (pData, pDataLen, &dest->block_len);
 435
 436     /* check structure */
 437     if (dest->version != 2)
 438         return FALSE;
 439
 440     return TRUE;
 441 }
 442
 443 /* structure of LZXC control data block */
 444 #define _CHM_LZXC_MIN_LEN (0x18)
 445 #define _CHM_LZXC_V2_LEN (0x1c)
 446 struct chmLzxcControlData
 447 {
 448     UInt32      size;                   /*  0        */
 449     char        signature[4];           /*  4 (LZXC) */
 450     UInt32      version;                /*  8        */
 451     UInt32      resetInterval;          /*  c        */
 452     UInt32      windowSize;             /* 10        */
 453     UInt32      windowsPerReset;        /* 14        */
 454     UInt32      unknown_18;             /* 18        */
 455 };
 456
 457 static BOOL _unmarshal_lzxc_control_data(unsigned char **pData,
 458                                          unsigned int *pDataLen,
 459                                          struct chmLzxcControlData *dest)
 460 {
 461     /* we want at least 0x18 bytes */
 462     if (*pDataLen < _CHM_LZXC_MIN_LEN)
 463         return FALSE;
 464
 465     /* unmarshal fields */
 466     _unmarshal_uint32    (pData, pDataLen, &dest->size);
 467     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 468     _unmarshal_uint32    (pData, pDataLen, &dest->version);
 469     _unmarshal_uint32    (pData, pDataLen, &dest->resetInterval);
 470     _unmarshal_uint32    (pData, pDataLen, &dest->windowSize);
 471     _unmarshal_uint32    (pData, pDataLen, &dest->windowsPerReset);
 472
 473     if (*pDataLen >= _CHM_LZXC_V2_LEN)
 474         _unmarshal_uint32    (pData, pDataLen, &dest->unknown_18);
 475     else
 476         dest->unknown_18 = 0;
 477
 478     if (dest->version == 2)
 479     {
 480         dest->resetInterval *= 0x8000;
 481         dest->windowSize *= 0x8000;
 482     }
 483     if (dest->windowSize == 0  ||  dest->resetInterval == 0)
 484         return FALSE;
 485
 486     /* for now, only support resetInterval a multiple of windowSize/2 */
 487     if (dest->windowSize == 1)
 488         return FALSE;
 489     if ((dest->resetInterval % (dest->windowSize/2)) != 0)
 490         return FALSE;
 491
 492     /* check structure */
 493     if (memcmp(dest->signature, "LZXC", 4) != 0)
 494         return FALSE;
 495
 496     return TRUE;
 497 }
 498
 499 /* the structure used for chm file handles */
 500 struct chmFile
 501 {
 502     HANDLE              fd;
 503
 504     CRITICAL_SECTION    mutex;
 505     CRITICAL_SECTION    lzx_mutex;
 506     CRITICAL_SECTION    cache_mutex;
 507
 508     UInt64              dir_offset;
 509     UInt64              dir_len;
 510     UInt64              data_offset;
 511     Int32               index_root;
 512     Int32               index_head;
 513     UInt32              block_len;
 514
 515     UInt64              span;
 516     struct chmUnitInfo  rt_unit;
 517     struct chmUnitInfo  cn_unit;
 518     struct chmLzxcResetTable reset_table;
 519
 520     /* LZX control data */
 521     int                 compression_enabled;
 522     UInt32              window_size;
 523     UInt32              reset_interval;
 524     UInt32              reset_blkcount;
 525
 526     /* decompressor state */
 527     struct LZXstate    *lzx_state;
 528     int                 lzx_last_block;
 529
 530     /* cache for decompressed blocks */
 531     UChar             **cache_blocks;
 532     Int64              *cache_block_indices;
 533     Int32               cache_num_blocks;
 534 };
 535
 536 /*
 537  * utility functions local to this module
 538  */
 539
 540 /* utility function to handle differences between {pread,read}(64)? */
 541 static Int64 _chm_fetch_bytes(struct chmFile *h,
 542                               UChar *buf,
 543                               UInt64 os,
 544                               Int64 len)
 545 {
 546     Int64 readLen=0;
 547     if (h->fd  ==  CHM_NULL_FD)
 548         return readLen;
 549
 550     CHM_ACQUIRE_LOCK(h->mutex);
 551     /* NOTE: this might be better done with CreateFileMapping, et cetera... */
 552     {
 553         LARGE_INTEGER old_pos, new_pos;
 554         DWORD actualLen=0;
 555
 556         /* awkward Win32 Seek/Tell */
 557         new_pos.QuadPart = 0;
 558         SetFilePointerEx( h->fd, new_pos, &old_pos, FILE_CURRENT );
 559         new_pos.QuadPart = os;
 560         SetFilePointerEx( h->fd, new_pos, NULL, FILE_BEGIN );
 561
 562         /* read the data */
 563         if (ReadFile(h->fd,
 564                      buf,
 565                      (DWORD)len,
 566                      &actualLen,
 567                      NULL))
 568             readLen = actualLen;
 569         else
 570             readLen = 0;
 571
 572         /* restore original position */
 573         SetFilePointerEx( h->fd, old_pos, NULL, FILE_BEGIN );
 574     }
 575     CHM_RELEASE_LOCK(h->mutex);
 576     return readLen;
 577 }
 578
 579 /*
 580  * set a parameter on the file handle.
 581  * valid parameter types:
 582  *          CHM_PARAM_MAX_BLOCKS_CACHED:
 583  *                 how many decompressed blocks should be cached?  A simple
 584  *                 caching scheme is used, wherein the index of the block is
 585  *                 used as a hash value, and hash collision results in the
 586  *                 invalidation of the previously cached block.
 587  */
 588 static void chm_set_param(struct chmFile *h,
 589                           int paramType,
 590                           int paramVal)
 591 {
 592     switch (paramType)
 593     {
 594         case CHM_PARAM_MAX_BLOCKS_CACHED:
 595             CHM_ACQUIRE_LOCK(h->cache_mutex);
 596             if (paramVal != h->cache_num_blocks)
 597             {
 598                 UChar **newBlocks;
 599                 Int64 *newIndices;
 600                 int     i;
 601
 602                 /* allocate new cached blocks */
 603                 newBlocks = HeapAlloc(GetProcessHeap(), 0, paramVal * sizeof (UChar *));
 604                 newIndices = HeapAlloc(GetProcessHeap(), 0, paramVal * sizeof (UInt64));
 605                 for (i=0; i<paramVal; i++)
 606                 {
 607                     newBlocks[i] = NULL;
 608                     newIndices[i] = 0;
 609                 }
 610
 611                 /* re-distribute old cached blocks */
 612                 if (h->cache_blocks)
 613                 {
 614                     for (i=0; i<h->cache_num_blocks; i++)
 615                     {
 616                         int newSlot = (int)(h->cache_block_indices[i] % paramVal);
 617
 618                         if (h->cache_blocks[i])
 619                         {
 620                             /* in case of collision, destroy newcomer */
 621                             if (newBlocks[newSlot])
 622                             {
 623                                 HeapFree(GetProcessHeap(), 0, h->cache_blocks[i]);
 624                                 h->cache_blocks[i] = NULL;
 625                             }
 626                             else
 627                             {
 628                                 newBlocks[newSlot] = h->cache_blocks[i];
 629                                 newIndices[newSlot] =
 630                                             h->cache_block_indices[i];
 631                             }
 632                         }
 633                     }
 634
 635                     HeapFree(GetProcessHeap(), 0, h->cache_blocks);
 636                     HeapFree(GetProcessHeap(), 0, h->cache_block_indices);
 637                 }
 638
 639                 /* now, set new values */
 640                 h->cache_blocks = newBlocks;
 641                 h->cache_block_indices = newIndices;
 642                 h->cache_num_blocks = paramVal;
 643             }
 644             CHM_RELEASE_LOCK(h->cache_mutex);
 645             break;
 646
 647         default:
 648             break;
 649     }
 650 }
 651
 652 /* open an ITS archive */
 653 struct chmFile *chm_openW(const WCHAR *filename)
 654 {
 655     unsigned char               sbuffer[256];
 656     unsigned int                sremain;
 657     unsigned char              *sbufpos;
 658     struct chmFile             *newHandle=NULL;
 659     struct chmItsfHeader        itsfHeader;
 660     struct chmItspHeader        itspHeader;
 661 #if 0
 662     struct chmUnitInfo          uiSpan;
 663 #endif
 664     struct chmUnitInfo          uiLzxc;
 665     struct chmLzxcControlData   ctlData;
 666
 667     /* allocate handle */
 668     newHandle = HeapAlloc(GetProcessHeap(), 0, sizeof(struct chmFile));
 669     newHandle->fd = CHM_NULL_FD;
 670     newHandle->lzx_state = NULL;
 671     newHandle->cache_blocks = NULL;
 672     newHandle->cache_block_indices = NULL;
 673     newHandle->cache_num_blocks = 0;
 674
 675     /* open file */
 676     if ((newHandle->fd=CreateFileW(filename,
 677                                    GENERIC_READ,
 678                                    FILE_SHARE_READ,
 679                                    NULL,
 680                                    OPEN_EXISTING,
 681                                    FILE_ATTRIBUTE_NORMAL,
 682                                    NULL)) == CHM_NULL_FD)
 683     {
 684         HeapFree(GetProcessHeap(), 0, newHandle);
 685         return NULL;
 686     }
 687
 688     /* initialize mutexes, if needed */
 689     InitializeCriticalSection(&newHandle->mutex);
 690     newHandle->mutex.DebugInfo->Spare[0] = (DWORD_PTR)(__FILE__ ": chmFile.mutex");
 691     InitializeCriticalSection(&newHandle->lzx_mutex);
 692     newHandle->lzx_mutex.DebugInfo->Spare[0] = (DWORD_PTR)(__FILE__ ": chmFile.lzx_mutex");
 693     InitializeCriticalSection(&newHandle->cache_mutex);
 694     newHandle->cache_mutex.DebugInfo->Spare[0] = (DWORD_PTR)(__FILE__ ": chmFile.cache_mutex");
 695
 696     /* read and verify header */
 697     sremain = _CHM_ITSF_V3_LEN;
 698     sbufpos = sbuffer;
 699     if (_chm_fetch_bytes(newHandle, sbuffer, 0, sremain) != sremain    ||
 700         !_unmarshal_itsf_header(&sbufpos, &sremain, &itsfHeader))
 701     {
 702         chm_close(newHandle);
 703         return NULL;
 704     }
 705
 706     /* stash important values from header */
 707     newHandle->dir_offset  = itsfHeader.dir_offset;
 708     newHandle->dir_len     = itsfHeader.dir_len;
 709     newHandle->data_offset = itsfHeader.data_offset;
 710
 711     /* now, read and verify the directory header chunk */
 712     sremain = _CHM_ITSP_V1_LEN;
 713     sbufpos = sbuffer;
 714     if (_chm_fetch_bytes(newHandle, sbuffer,
 715                          itsfHeader.dir_offset, sremain) != sremain    ||
 716         !_unmarshal_itsp_header(&sbufpos, &sremain, &itspHeader))
 717     {
 718         chm_close(newHandle);
 719         return NULL;
 720     }
 721
 722     /* grab essential information from ITSP header */
 723     newHandle->dir_offset += itspHeader.header_len;
 724     newHandle->dir_len    -= itspHeader.header_len;
 725     newHandle->index_root  = itspHeader.index_root;
 726     newHandle->index_head  = itspHeader.index_head;
 727     newHandle->block_len   = itspHeader.block_len;
 728
 729     /* if the index root is -1, this means we don't have any PMGI blocks.
 730      * as a result, we must use the sole PMGL block as the index root
 731      */
 732     if (newHandle->index_root == -1)
 733         newHandle->index_root = newHandle->index_head;
 734
 735     /* initialize cache */
 736     chm_set_param(newHandle, CHM_PARAM_MAX_BLOCKS_CACHED,
 737                   CHM_MAX_BLOCKS_CACHED);
 738
 739     /* By default, compression is enabled. */
 740     newHandle->compression_enabled = 1;
 741
 742     /* prefetch most commonly needed unit infos */
 743     if (CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
 744                                                   L"::DataSpace/Storage/MSCompressed/Transform/"
 745                                                   "{7FC28940-9D31-11D0-9B27-00A0C91E9C7C}/"
 746                                                   "InstanceData/ResetTable",
 747                                                   &newHandle->rt_unit)    ||
 748         newHandle->rt_unit.space == CHM_COMPRESSED                        ||
 749         CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
 750                                                   L"::DataSpace/Storage/MSCompressed/Content",
 751                                                   &newHandle->cn_unit)    ||
 752         newHandle->cn_unit.space == CHM_COMPRESSED                        ||
 753         CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
 754                                                   L"::DataSpace/Storage/MSCompressed/ControlData",
 755                                                   &uiLzxc)                ||
 756         uiLzxc.space == CHM_COMPRESSED)
 757     {
 758         newHandle->compression_enabled = 0;
 759     }
 760
 761     /* read reset table info */
 762     if (newHandle->compression_enabled)
 763     {
 764         sremain = _CHM_LZXC_RESETTABLE_V1_LEN;
 765         sbufpos = sbuffer;
 766         if (chm_retrieve_object(newHandle, &newHandle->rt_unit, sbuffer,
 767                                 0, sremain) != sremain                        ||
 768             !_unmarshal_lzxc_reset_table(&sbufpos, &sremain,
 769                                          &newHandle->reset_table))
 770         {
 771             newHandle->compression_enabled = 0;
 772         }
 773     }
 774
 775     /* read control data */
 776     if (newHandle->compression_enabled)
 777     {
 778         sremain = (unsigned long)uiLzxc.length;
 779         sbufpos = sbuffer;
 780         if (chm_retrieve_object(newHandle, &uiLzxc, sbuffer,
 781                                 0, sremain) != sremain                       ||
 782             !_unmarshal_lzxc_control_data(&sbufpos, &sremain,
 783                                           &ctlData))
 784         {
 785             newHandle->compression_enabled = 0;
 786         }
 787         else
 788         {
 789             newHandle->window_size = ctlData.windowSize;
 790             newHandle->reset_interval = ctlData.resetInterval;
 791 /* Jed, Mon Jun 28: Experimentally, it appears that the reset block count */
 792 /*       must be multiplied by this formerly unknown ctrl data field in   */
 793 /*       order to decompress some files.                                  */
 794             newHandle->reset_blkcount = newHandle->reset_interval / (newHandle->window_size / 2) *
 795                                         ctlData.windowsPerReset;
 796         }
 797     }
 798
 799     return newHandle;
 800 }
 801
 802 /* Duplicate an ITS archive handle */
 803 struct chmFile *chm_dup(struct chmFile *oldHandle)
 804 {
 805     struct chmFile *newHandle=NULL;
 806
 807     newHandle = HeapAlloc(GetProcessHeap(), 0, sizeof(struct chmFile));
 808     *newHandle = *oldHandle;
 809
 810     /* duplicate fd handle */
 811     DuplicateHandle(GetCurrentProcess(), oldHandle->fd,
 812                     GetCurrentProcess(), &(newHandle->fd),
 813                     0, FALSE, DUPLICATE_SAME_ACCESS);
 814     newHandle->lzx_state = NULL;
 815     newHandle->cache_blocks = NULL;
 816     newHandle->cache_block_indices = NULL;
 817     newHandle->cache_num_blocks = 0;
 818
 819     /* initialize mutexes, if needed */
 820     InitializeCriticalSection(&newHandle->mutex);
 821     newHandle->mutex.DebugInfo->Spare[0] = (DWORD_PTR)(__FILE__ ": chmFile.mutex");
 822     InitializeCriticalSection(&newHandle->lzx_mutex);
 823     newHandle->lzx_mutex.DebugInfo->Spare[0] = (DWORD_PTR)(__FILE__ ": chmFile.lzx_mutex");
 824     InitializeCriticalSection(&newHandle->cache_mutex);
 825     newHandle->cache_mutex.DebugInfo->Spare[0] = (DWORD_PTR)(__FILE__ ": chmFile.cache_mutex");
 826
 827     /* initialize cache */
 828     chm_set_param(newHandle, CHM_PARAM_MAX_BLOCKS_CACHED,
 829                   CHM_MAX_BLOCKS_CACHED);
 830
 831     return newHandle;
 832 }
 833
 834 /* close an ITS archive */
 835 void chm_close(struct chmFile *h)
 836 {
 837     if (h != NULL)
 838     {
 839         if (h->fd != CHM_NULL_FD)
 840             CHM_CLOSE_FILE(h->fd);
 841         h->fd = CHM_NULL_FD;
 842
 843         h->mutex.DebugInfo->Spare[0] = 0;
 844         DeleteCriticalSection(&h->mutex);
 845         h->lzx_mutex.DebugInfo->Spare[0] = 0;
 846         DeleteCriticalSection(&h->lzx_mutex);
 847         h->cache_mutex.DebugInfo->Spare[0] = 0;
 848         DeleteCriticalSection(&h->cache_mutex);
 849
 850         if (h->lzx_state)
 851             LZXteardown(h->lzx_state);
 852         h->lzx_state = NULL;
 853
 854         if (h->cache_blocks)
 855         {
 856             int i;
 857             for (i=0; i<h->cache_num_blocks; i++)
 858             {
 859                 HeapFree(GetProcessHeap(), 0, h->cache_blocks[i]);
 860             }
 861             HeapFree(GetProcessHeap(), 0, h->cache_blocks);
 862             h->cache_blocks = NULL;
 863         }
 864
 865         HeapFree(GetProcessHeap(), 0, h->cache_block_indices);
 866         h->cache_block_indices = NULL;
 867
 868         HeapFree(GetProcessHeap(), 0, h);
 869     }
 870 }
 871
 872 /*
 873  * helper methods for chm_resolve_object
 874  */
 875
 876 /* skip a compressed dword */
 877 static void _chm_skip_cword(UChar **pEntry)
 878 {
 879     while (*(*pEntry)++ >= 0x80)
 880         ;
 881 }
 882
 883 /* skip the data from a PMGL entry */
 884 static void _chm_skip_PMGL_entry_data(UChar **pEntry)
 885 {
 886     _chm_skip_cword(pEntry);
 887     _chm_skip_cword(pEntry);
 888     _chm_skip_cword(pEntry);
 889 }
 890
 891 /* parse a compressed dword */
 892 static UInt64 _chm_parse_cword(UChar **pEntry)
 893 {
 894     UInt64 accum = 0;
 895     UChar temp;
 896     while ((temp=*(*pEntry)++) >= 0x80)
 897     {
 898         accum <<= 7;
 899         accum += temp & 0x7f;
 900     }
 901
 902     return (accum << 7) + temp;
 903 }
 904
 905 /* parse a utf-8 string into an ASCII char buffer */
 906 static BOOL _chm_parse_UTF8(UChar **pEntry, UInt64 count, WCHAR *path)
 907 {
 908     DWORD length = MultiByteToWideChar(CP_UTF8, 0, (char *)*pEntry, count, path, CHM_MAX_PATHLEN);
 909     path[length] = '\0';
 910     *pEntry += count;
 911     return !!length;
 912 }
 913
 914 /* parse a PMGL entry into a chmUnitInfo struct; return 1 on success. */
 915 static BOOL _chm_parse_PMGL_entry(UChar **pEntry, struct chmUnitInfo *ui)
 916 {
 917     UInt64 strLen;
 918
 919     /* parse str len */
 920     strLen = _chm_parse_cword(pEntry);
 921     if (strLen > CHM_MAX_PATHLEN)
 922         return FALSE;
 923
 924     /* parse path */
 925     if (! _chm_parse_UTF8(pEntry, strLen, ui->path))
 926         return FALSE;
 927
 928     /* parse info */
 929     ui->space  = (int)_chm_parse_cword(pEntry);
 930     ui->start  = _chm_parse_cword(pEntry);
 931     ui->length = _chm_parse_cword(pEntry);
 932     return TRUE;
 933 }
 934
 935 /* find an exact entry in PMGL; return NULL if we fail */
 936 static UChar *_chm_find_in_PMGL(UChar *page_buf,
 937                          UInt32 block_len,
 938                          const WCHAR *objPath)
 939 {
 940     /* XXX: modify this to do a binary search using the nice index structure
 941      *      that is provided for us.
 942      */
 943     struct chmPmglHeader header;
 944     unsigned int hremain;
 945     UChar *end;
 946     UChar *cur;
 947     UChar *temp;
 948     UInt64 strLen;
 949     WCHAR buffer[CHM_MAX_PATHLEN+1];
 950
 951     /* figure out where to start and end */
 952     cur = page_buf;
 953     hremain = _CHM_PMGL_LEN;
 954     if (! _unmarshal_pmgl_header(&cur, &hremain, &header))
 955         return NULL;
 956     end = page_buf + block_len - (header.free_space);
 957
 958     /* now, scan progressively */
 959     while (cur < end)
 960     {
 961         /* grab the name */
 962         temp = cur;
 963         strLen = _chm_parse_cword(&cur);
 964         if (! _chm_parse_UTF8(&cur, strLen, buffer))
 965             return NULL;
 966
 967         /* check if it is the right name */
 968         if (! wcsicmp(buffer, objPath))
 969             return temp;
 970
 971         _chm_skip_PMGL_entry_data(&cur);
 972     }
 973
 974     return NULL;
 975 }
 976
 977 /* find which block should be searched next for the entry; -1 if no block */
 978 static Int32 _chm_find_in_PMGI(UChar *page_buf,
 979                         UInt32 block_len,
 980                         const WCHAR *objPath)
 981 {
 982     /* XXX: modify this to do a binary search using the nice index structure
 983      *      that is provided for us
 984      */
 985     struct chmPmgiHeader header;
 986     unsigned int hremain;
 987     int page=-1;
 988     UChar *end;
 989     UChar *cur;
 990     UInt64 strLen;
 991     WCHAR buffer[CHM_MAX_PATHLEN+1];
 992
 993     /* figure out where to start and end */
 994     cur = page_buf;
 995     hremain = _CHM_PMGI_LEN;
 996     if (! _unmarshal_pmgi_header(&cur, &hremain, &header))
 997         return -1;
 998     end = page_buf + block_len - (header.free_space);
 999
1000     /* now, scan progressively */
1001     while (cur < end)
1002     {
1003         /* grab the name */
1004         strLen = _chm_parse_cword(&cur);
1005         if (! _chm_parse_UTF8(&cur, strLen, buffer))
1006             return -1;
1007
1008         /* check if it is the right name */
1009         if (wcsicmp(buffer, objPath) > 0)
1010             return page;
1011
1012         /* load next value for path */
1013         page = (int)_chm_parse_cword(&cur);
1014     }
1015
1016     return page;
1017 }
1018
1019 /* resolve a particular object from the archive */
1020 int chm_resolve_object(struct chmFile *h,
1021                        const WCHAR *objPath,
1022                        struct chmUnitInfo *ui)
1023 {
1024     /*
1025      * XXX: implement caching scheme for dir pages
1026      */
1027
1028     Int32 curPage;
1029
1030     /* buffer to hold whatever page we're looking at */
1031     UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, h->block_len);
1032
1033     /* starting page */
1034     curPage = h->index_root;
1035
1036     /* until we have either returned or given up */
1037     while (curPage != -1)
1038     {
1039
1040         /* try to fetch the index page */
1041         if (_chm_fetch_bytes(h, page_buf,
1042                              h->dir_offset + (UInt64)curPage*h->block_len,
1043                              h->block_len) != h->block_len)
1044         {
1045             HeapFree(GetProcessHeap(), 0, page_buf);
1046             return CHM_RESOLVE_FAILURE;
1047         }
1048
1049         /* now, if it is a leaf node: */
1050         if (memcmp(page_buf, _chm_pmgl_marker, 4) == 0)
1051         {
1052             /* scan block */
1053             UChar *pEntry = _chm_find_in_PMGL(page_buf,
1054                                               h->block_len,
1055                                               objPath);
1056             if (pEntry == NULL)
1057             {
1058                 HeapFree(GetProcessHeap(), 0, page_buf);
1059                 return CHM_RESOLVE_FAILURE;
1060             }
1061
1062             /* parse entry and return */
1063             _chm_parse_PMGL_entry(&pEntry, ui);
1064             HeapFree(GetProcessHeap(), 0, page_buf);
1065             return CHM_RESOLVE_SUCCESS;
1066         }
1067
1068         /* else, if it is a branch node: */
1069         else if (memcmp(page_buf, _chm_pmgi_marker, 4) == 0)
1070             curPage = _chm_find_in_PMGI(page_buf, h->block_len, objPath);
1071
1072         /* else, we are confused.  give up. */
1073         else
1074         {
1075             HeapFree(GetProcessHeap(), 0, page_buf);
1076             return CHM_RESOLVE_FAILURE;
1077         }
1078     }
1079
1080     /* didn't find anything.  fail. */
1081     HeapFree(GetProcessHeap(), 0, page_buf);
1082     return CHM_RESOLVE_FAILURE;
1083 }
1084
1085 /*
1086  * utility methods for dealing with compressed data
1087  */
1088
1089 /* get the bounds of a compressed block. Returns FALSE on failure */
1090 static BOOL _chm_get_cmpblock_bounds(struct chmFile *h,
1091                                      UInt64 block,
1092                                      UInt64 *start,
1093                                      Int64 *len)
1094 {
1095     UChar buffer[8], *dummy;
1096     unsigned int remain;
1097
1098     /* for all but the last block, use the reset table */
1099     if (block < h->reset_table.block_count-1)
1100     {
1101         /* unpack the start address */
1102         dummy = buffer;
1103         remain = 8;
1104         if (_chm_fetch_bytes(h, buffer,
1105                              h->data_offset
1106                                 + h->rt_unit.start
1107                                 + h->reset_table.table_offset
1108                                 + block*8,
1109                              remain) != remain                            ||
1110             !_unmarshal_uint64(&dummy, &remain, start))
1111             return FALSE;
1112
1113         /* unpack the end address */
1114         dummy = buffer;
1115         remain = 8;
1116         if (_chm_fetch_bytes(h, buffer,
1117                              h->data_offset
1118                                 + h->rt_unit.start
1119                                 + h->reset_table.table_offset
1120                                 + block*8 + 8,
1121                          remain) != remain                                ||
1122             !_unmarshal_int64(&dummy, &remain, len))
1123             return FALSE;
1124     }
1125
1126     /* for the last block, use the span in addition to the reset table */
1127     else
1128     {
1129         /* unpack the start address */
1130         dummy = buffer;
1131         remain = 8;
1132         if (_chm_fetch_bytes(h, buffer,
1133                              h->data_offset
1134                                 + h->rt_unit.start
1135                                 + h->reset_table.table_offset
1136                                 + block*8,
1137                              remain) != remain                            ||
1138             !_unmarshal_uint64(&dummy, &remain, start))
1139             return FALSE;
1140
1141         *len = h->reset_table.compressed_len;
1142     }
1143
1144     /* compute the length and absolute start address */
1145     *len -= *start;
1146     *start += h->data_offset + h->cn_unit.start;
1147
1148     return TRUE;
1149 }
1150
1151 /* decompress the block.  must have lzx_mutex. */
1152 static Int64 _chm_decompress_block(struct chmFile *h,
1153                                    UInt64 block,
1154                                    UChar **ubuffer)
1155 {
1156     UChar *cbuffer = HeapAlloc( GetProcessHeap(), 0,
1157                               ((unsigned int)h->reset_table.block_len + 6144));
1158     UInt64 cmpStart;                                    /* compressed start  */
1159     Int64 cmpLen;                                       /* compressed len    */
1160     int indexSlot;                                      /* cache index slot  */
1161     UChar *lbuffer;                                     /* local buffer ptr  */
1162     UInt32 blockAlign = (UInt32)(block % h->reset_blkcount); /* reset interval align */
1163     UInt32 i;                                           /* local loop index  */
1164
1165     /* let the caching system pull its weight! */
1166     if (block - blockAlign <= h->lzx_last_block  &&
1167         block              >= h->lzx_last_block)
1168         blockAlign = (block - h->lzx_last_block);
1169
1170     /* check if we need previous blocks */
1171     if (blockAlign != 0)
1172     {
1173         /* fetch all required previous blocks since last reset */
1174         for (i = blockAlign; i > 0; i--)
1175         {
1176             UInt32 curBlockIdx = block - i;
1177
1178             /* check if we most recently decompressed the previous block */
1179             if (h->lzx_last_block != curBlockIdx)
1180             {
1181                 if ((curBlockIdx % h->reset_blkcount) == 0)
1182                 {
1183 #ifdef CHM_DEBUG
1184                     fprintf(stderr, "***RESET (1)***\n");
1185 #endif
1186                     LZXreset(h->lzx_state);
1187                 }
1188
1189                 indexSlot = (int)((curBlockIdx) % h->cache_num_blocks);
1190                 h->cache_block_indices[indexSlot] = curBlockIdx;
1191                 if (! h->cache_blocks[indexSlot])
1192                     h->cache_blocks[indexSlot] =
1193                       HeapAlloc(GetProcessHeap(), 0,
1194                                 (unsigned int)(h->reset_table.block_len));
1195                 lbuffer = h->cache_blocks[indexSlot];
1196
1197                 /* decompress the previous block */
1198 #ifdef CHM_DEBUG
1199                 fprintf(stderr, "Decompressing block #%4d (EXTRA)\n", curBlockIdx);
1200 #endif
1201                 if (!_chm_get_cmpblock_bounds(h, curBlockIdx, &cmpStart, &cmpLen) ||
1202                     _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen      ||
1203                     LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1204                                   (int)h->reset_table.block_len) != DECR_OK)
1205                 {
1206 #ifdef CHM_DEBUG
1207                     fprintf(stderr, "   (DECOMPRESS FAILED!)\n");
1208 #endif
1209                     HeapFree(GetProcessHeap(), 0, cbuffer);
1210                     return 0;
1211                 }
1212
1213                 h->lzx_last_block = (int)curBlockIdx;
1214             }
1215         }
1216     }
1217     else
1218     {
1219         if ((block % h->reset_blkcount) == 0)
1220         {
1221 #ifdef CHM_DEBUG
1222             fprintf(stderr, "***RESET (2)***\n");
1223 #endif
1224             LZXreset(h->lzx_state);
1225         }
1226     }
1227
1228     /* allocate slot in cache */
1229     indexSlot = (int)(block % h->cache_num_blocks);
1230     h->cache_block_indices[indexSlot] = block;
1231     if (! h->cache_blocks[indexSlot])
1232         h->cache_blocks[indexSlot] =
1233           HeapAlloc(GetProcessHeap(), 0, ((unsigned int)h->reset_table.block_len));
1234     lbuffer = h->cache_blocks[indexSlot];
1235     *ubuffer = lbuffer;
1236
1237     /* decompress the block we actually want */
1238 #ifdef CHM_DEBUG
1239     fprintf(stderr, "Decompressing block #%4d (REAL )\n", block);
1240 #endif
1241     if (! _chm_get_cmpblock_bounds(h, block, &cmpStart, &cmpLen)          ||
1242         _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen          ||
1243         LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1244                       (int)h->reset_table.block_len) != DECR_OK)
1245     {
1246 #ifdef CHM_DEBUG
1247         fprintf(stderr, "   (DECOMPRESS FAILED!)\n");
1248 #endif
1249         HeapFree(GetProcessHeap(), 0, cbuffer);
1250         return 0;
1251     }
1252     h->lzx_last_block = (int)block;
1253
1254     /* XXX: modify LZX routines to return the length of the data they
1255      * decompressed and return that instead, for an extra sanity check.
1256      */
1257     HeapFree(GetProcessHeap(), 0, cbuffer);
1258     return h->reset_table.block_len;
1259 }
1260
1261 /* grab a region from a compressed block */
1262 static Int64 _chm_decompress_region(struct chmFile *h,
1263                                     UChar *buf,
1264                                     UInt64 start,
1265                                     Int64 len)
1266 {
1267     UInt64 nBlock, nOffset;
1268     UInt64 nLen;
1269     UInt64 gotLen;
1270     UChar *ubuffer = NULL;
1271
1272         if (len <= 0)
1273                 return 0;
1274
1275     /* figure out what we need to read */
1276     nBlock = start / h->reset_table.block_len;
1277     nOffset = start % h->reset_table.block_len;
1278     nLen = len;
1279     if (nLen > (h->reset_table.block_len - nOffset))
1280         nLen = h->reset_table.block_len - nOffset;
1281
1282     /* if block is cached, return data from it. */
1283     CHM_ACQUIRE_LOCK(h->lzx_mutex);
1284     CHM_ACQUIRE_LOCK(h->cache_mutex);
1285     if (h->cache_block_indices[nBlock % h->cache_num_blocks] == nBlock    &&
1286         h->cache_blocks[nBlock % h->cache_num_blocks] != NULL)
1287     {
1288         memcpy(buf,
1289                h->cache_blocks[nBlock % h->cache_num_blocks] + nOffset,
1290                (unsigned int)nLen);
1291         CHM_RELEASE_LOCK(h->cache_mutex);
1292         CHM_RELEASE_LOCK(h->lzx_mutex);
1293         return nLen;
1294     }
1295     CHM_RELEASE_LOCK(h->cache_mutex);
1296
1297     /* data request not satisfied, so... start up the decompressor machine */
1298     if (! h->lzx_state)
1299     {
1300         h->lzx_last_block = -1;
1301         h->lzx_state = LZXinit(h->window_size);
1302     }
1303
1304     /* decompress some data */
1305     gotLen = _chm_decompress_block(h, nBlock, &ubuffer);
1306     if (gotLen < nLen)
1307         nLen = gotLen;
1308     memcpy(buf, ubuffer+nOffset, (unsigned int)nLen);
1309     CHM_RELEASE_LOCK(h->lzx_mutex);
1310     return nLen;
1311 }
1312
1313 /* retrieve (part of) an object */
1314 LONGINT64 chm_retrieve_object(struct chmFile *h,
1315                                struct chmUnitInfo *ui,
1316                                unsigned char *buf,
1317                                LONGUINT64 addr,
1318                                LONGINT64 len)
1319 {
1320     /* must be valid file handle */
1321     if (h == NULL)
1322         return 0;
1323
1324     /* starting address must be in correct range */
1325     if (addr >= ui->length)
1326         return 0;
1327
1328     /* clip length */
1329     if (addr + len > ui->length)
1330         len = ui->length - addr;
1331
1332     /* if the file is uncompressed, it's simple */
1333     if (ui->space == CHM_UNCOMPRESSED)
1334     {
1335         /* read data */
1336         return _chm_fetch_bytes(h,
1337                                 buf,
1338                                 h->data_offset + ui->start + addr,
1339                                 len);
1340     }
1341
1342     /* else if the file is compressed, it's a little trickier */
1343     else /* ui->space == CHM_COMPRESSED */
1344     {
1345         Int64 swath=0, total=0;
1346
1347         /* if compression is not enabled for this file... */
1348         if (! h->compression_enabled)
1349             return total;
1350
1351         do {
1352
1353             /* swill another mouthful */
1354             swath = _chm_decompress_region(h, buf, ui->start + addr, len);
1355
1356             /* if we didn't get any... */
1357             if (swath == 0)
1358                 return total;
1359
1360             /* update stats */
1361             total += swath;
1362             len -= swath;
1363             addr += swath;
1364             buf += swath;
1365
1366         } while (len != 0);
1367
1368         return total;
1369     }
1370 }
1371
1372 BOOL chm_enumerate_dir(struct chmFile *h,
1373                        const WCHAR *prefix,
1374                        int what,
1375                        CHM_ENUMERATOR e,
1376                        void *context)
1377 {
1378     /*
1379      * XXX: do this efficiently (i.e. using the tree index)
1380      */
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
1391     /* set to TRUE once we've started */
1392     BOOL it_has_begun = FALSE;
1393
1394     /* the current ui */
1395     struct chmUnitInfo ui;
1396     int flag;
1397     UInt64 ui_path_len;
1398
1399     /* the length of the prefix */
1400     WCHAR prefixRectified[CHM_MAX_PATHLEN+1];
1401     int prefixLen;
1402     WCHAR lastPath[CHM_MAX_PATHLEN];
1403     int lastPathLen;
1404
1405     /* starting page */
1406     curPage = h->index_head;
1407
1408     /* initialize pathname state */
1409     lstrcpynW(prefixRectified, prefix, CHM_MAX_PATHLEN);
1410     prefixLen = lstrlenW(prefixRectified);
1411     if (prefixLen != 0)
1412     {
1413         if (prefixRectified[prefixLen-1] != '/')
1414         {
1415             prefixRectified[prefixLen] = '/';
1416             prefixRectified[prefixLen+1] = '\0';
1417             ++prefixLen;
1418         }
1419     }
1420     lastPath[0] = '\0';
1421     lastPathLen = -1;
1422
1423     /* until we have either returned or given up */
1424     while (curPage != -1)
1425     {
1426
1427         /* try to fetch the index page */
1428         if (_chm_fetch_bytes(h,
1429                              page_buf,
1430                              h->dir_offset + (UInt64)curPage*h->block_len,
1431                              h->block_len) != h->block_len)
1432         {
1433             HeapFree(GetProcessHeap(), 0, page_buf);
1434             return FALSE;
1435         }
1436
1437         /* figure out start and end for this page */
1438         cur = page_buf;
1439         lenRemain = _CHM_PMGL_LEN;
1440         if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1441         {
1442             HeapFree(GetProcessHeap(), 0, page_buf);
1443             return FALSE;
1444         }
1445         end = page_buf + h->block_len - (header.free_space);
1446
1447         /* loop over this page */
1448         while (cur < end)
1449         {
1450             if (! _chm_parse_PMGL_entry(&cur, &ui))
1451             {
1452                 HeapFree(GetProcessHeap(), 0, page_buf);
1453                 return FALSE;
1454             }
1455
1456             /* check if we should start */
1457             if (! it_has_begun)
1458             {
1459                 if (ui.length == 0  &&  wcsnicmp(ui.path, prefixRectified, prefixLen) == 0)
1460                     it_has_begun = TRUE;
1461                 else
1462                     continue;
1463
1464                 if (ui.path[prefixLen] == '\0')
1465                     continue;
1466             }
1467
1468             /* check if we should stop */
1469             else
1470             {
1471                 if (wcsnicmp(ui.path, prefixRectified, prefixLen) != 0)
1472                 {
1473                     HeapFree(GetProcessHeap(), 0, page_buf);
1474                     return TRUE;
1475                 }
1476             }
1477
1478             /* check if we should include this path */
1479             if (lastPathLen != -1)
1480             {
1481                 if (wcsnicmp(ui.path, lastPath, lastPathLen) == 0)
1482                     continue;
1483             }
1484             lstrcpyW(lastPath, ui.path);
1485             lastPathLen = lstrlenW(lastPath);
1486
1487             /* get the length of the path */
1488             ui_path_len = lstrlenW(ui.path)-1;
1489
1490             /* check for DIRS */
1491             if (ui.path[ui_path_len] == '/'  &&  !(what & CHM_ENUMERATE_DIRS))
1492                 continue;
1493
1494             /* check for FILES */
1495             if (ui.path[ui_path_len] != '/'  &&  !(what & CHM_ENUMERATE_FILES))
1496                 continue;
1497
1498             /* check for NORMAL vs. META */
1499             if (ui.path[0] == '/')
1500             {
1501
1502                 /* check for NORMAL vs. SPECIAL */
1503                 if (ui.path[1] == '#'  ||  ui.path[1] == '$')
1504                     flag = CHM_ENUMERATE_SPECIAL;
1505                 else
1506                     flag = CHM_ENUMERATE_NORMAL;
1507             }
1508             else
1509                 flag = CHM_ENUMERATE_META;
1510             if (! (what & flag))
1511                 continue;
1512
1513             /* call the enumerator */
1514             {
1515                 int status = (*e)(h, &ui, context);
1516                 switch (status)
1517                 {
1518                     case CHM_ENUMERATOR_FAILURE:
1519                         HeapFree(GetProcessHeap(), 0, page_buf);
1520                         return FALSE;
1521                     case CHM_ENUMERATOR_CONTINUE:
1522                         break;
1523                     case CHM_ENUMERATOR_SUCCESS:
1524                         HeapFree(GetProcessHeap(), 0, page_buf);
1525                         return TRUE;
1526                     default:
1527                         break;
1528                 }
1529             }
1530         }
1531
1532         /* advance to next page */
1533         curPage = header.block_next;
1534     }
1535
1536     HeapFree(GetProcessHeap(), 0, page_buf);
1537     return TRUE;
1538 }