git.oblomov.eu Git - wine/blob - 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 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
  80 /*
  81  * architecture specific defines
  82  *
  83  * Note: as soon as C99 is more widespread, the below defines should
  84  * probably just use the C99 sized-int types.
  85  *
  86  * The following settings will probably work for many platforms.  The sizes
  87  * don't have to be exactly correct, but the types must accommodate at least as
  88  * many bits as they specify.
  89  */
  90
  91 /* i386, 32-bit, Windows */
  92 typedef BYTE   UChar;
  93 typedef SHORT  Int16;
  94 typedef USHORT UInt16;
  95 typedef LONG   Int32;
  96 typedef DWORD      UInt32;
  97 typedef LONGLONG   Int64;
  98 typedef ULONGLONG  UInt64;
  99
 100 /* utilities for unmarshalling data */
 101 static int _unmarshal_char_array(unsigned char **pData,
 102                                  unsigned int *pLenRemain,
 103                                  char *dest,
 104                                  int count)
 105 {
 106     if (count <= 0  ||  (unsigned int)count > *pLenRemain)
 107         return 0;
 108     memcpy(dest, (*pData), count);
 109     *pData += count;
 110     *pLenRemain -= count;
 111     return 1;
 112 }
 113
 114 static int _unmarshal_uchar_array(unsigned char **pData,
 115                                   unsigned int *pLenRemain,
 116                                   unsigned char *dest,
 117                                   int count)
 118 {
 119         if (count <= 0  ||  (unsigned int)count > *pLenRemain)
 120         return 0;
 121     memcpy(dest, (*pData), count);
 122     *pData += count;
 123     *pLenRemain -= count;
 124     return 1;
 125 }
 126
 127 static int _unmarshal_int32(unsigned char **pData,
 128                             unsigned int *pLenRemain,
 129                             Int32 *dest)
 130 {
 131     if (4 > *pLenRemain)
 132         return 0;
 133     *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
 134     *pData += 4;
 135     *pLenRemain -= 4;
 136     return 1;
 137 }
 138
 139 static int _unmarshal_uint32(unsigned char **pData,
 140                              unsigned int *pLenRemain,
 141                              UInt32 *dest)
 142 {
 143     if (4 > *pLenRemain)
 144         return 0;
 145     *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
 146     *pData += 4;
 147     *pLenRemain -= 4;
 148     return 1;
 149 }
 150
 151 static int _unmarshal_int64(unsigned char **pData,
 152                             unsigned int *pLenRemain,
 153                             Int64 *dest)
 154 {
 155     Int64 temp;
 156     int i;
 157     if (8 > *pLenRemain)
 158         return 0;
 159     temp=0;
 160     for(i=8; i>0; i--)
 161     {
 162         temp <<= 8;
 163         temp |= (*pData)[i-1];
 164     }
 165     *dest = temp;
 166     *pData += 8;
 167     *pLenRemain -= 8;
 168     return 1;
 169 }
 170
 171 static int _unmarshal_uint64(unsigned char **pData,
 172                              unsigned int *pLenRemain,
 173                              UInt64 *dest)
 174 {
 175     UInt64 temp;
 176     int i;
 177     if (8 > *pLenRemain)
 178         return 0;
 179     temp=0;
 180     for(i=8; i>0; i--)
 181     {
 182         temp <<= 8;
 183         temp |= (*pData)[i-1];
 184     }
 185     *dest = temp;
 186     *pData += 8;
 187     *pLenRemain -= 8;
 188     return 1;
 189 }
 190
 191 static int _unmarshal_uuid(unsigned char **pData,
 192                            unsigned int *pDataLen,
 193                            unsigned char *dest)
 194 {
 195     return _unmarshal_uchar_array(pData, pDataLen, dest, 16);
 196 }
 197
 198 /* names of sections essential to decompression */
 199 static const WCHAR _CHMU_RESET_TABLE[] = {
 200 ':',':','D','a','t','a','S','p','a','c','e','/',
 201         'S','t','o','r','a','g','e','/',
 202         'M','S','C','o','m','p','r','e','s','s','e','d','/',
 203         'T','r','a','n','s','f','o','r','m','/',
 204         '{','7','F','C','2','8','9','4','0','-','9','D','3','1',
 205           '-','1','1','D','0','-','9','B','2','7','-',
 206           '0','0','A','0','C','9','1','E','9','C','7','C','}','/',
 207         'I','n','s','t','a','n','c','e','D','a','t','a','/',
 208         'R','e','s','e','t','T','a','b','l','e',0
 209 };
 210 static const WCHAR _CHMU_LZXC_CONTROLDATA[] = {
 211 ':',':','D','a','t','a','S','p','a','c','e','/',
 212         'S','t','o','r','a','g','e','/',
 213         'M','S','C','o','m','p','r','e','s','s','e','d','/',
 214         'C','o','n','t','r','o','l','D','a','t','a',0
 215 };
 216 static const WCHAR _CHMU_CONTENT[] = {
 217 ':',':','D','a','t','a','S','p','a','c','e','/',
 218         'S','t','o','r','a','g','e','/',
 219         'M','S','C','o','m','p','r','e','s','s','e','d','/',
 220         'C','o','n','t','e','n','t',0
 221 };
 222 static const WCHAR _CHMU_SPANINFO[] = {
 223 ':',':','D','a','t','a','S','p','a','c','e','/',
 224         'S','t','o','r','a','g','e','/',
 225         'M','S','C','o','m','p','r','e','s','s','e','d','/',
 226         'S','p','a','n','I','n','f','o',
 227 };
 228
 229 /*
 230  * structures local to this module
 231  */
 232
 233 /* structure of ITSF headers */
 234 #define _CHM_ITSF_V2_LEN (0x58)
 235 #define _CHM_ITSF_V3_LEN (0x60)
 236 struct chmItsfHeader
 237 {
 238     char        signature[4];           /*  0 (ITSF) */
 239     Int32       version;                /*  4 */
 240     Int32       header_len;             /*  8 */
 241     Int32       unknown_000c;           /*  c */
 242     UInt32      last_modified;          /* 10 */
 243     UInt32      lang_id;                /* 14 */
 244     UChar       dir_uuid[16];           /* 18 */
 245     UChar       stream_uuid[16];        /* 28 */
 246     UInt64      unknown_offset;         /* 38 */
 247     UInt64      unknown_len;            /* 40 */
 248     UInt64      dir_offset;             /* 48 */
 249     UInt64      dir_len;                /* 50 */
 250     UInt64      data_offset;            /* 58 (Not present before V3) */
 251 }; /* __attribute__ ((aligned (1))); */
 252
 253 static int _unmarshal_itsf_header(unsigned char **pData,
 254                                   unsigned int *pDataLen,
 255                                   struct chmItsfHeader *dest)
 256 {
 257     /* we only know how to deal with the 0x58 and 0x60 byte structures */
 258     if (*pDataLen != _CHM_ITSF_V2_LEN  &&  *pDataLen != _CHM_ITSF_V3_LEN)
 259         return 0;
 260
 261     /* unmarshal common fields */
 262     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 263     _unmarshal_int32     (pData, pDataLen, &dest->version);
 264     _unmarshal_int32     (pData, pDataLen, &dest->header_len);
 265     _unmarshal_int32     (pData, pDataLen, &dest->unknown_000c);
 266     _unmarshal_uint32    (pData, pDataLen, &dest->last_modified);
 267     _unmarshal_uint32    (pData, pDataLen, &dest->lang_id);
 268     _unmarshal_uuid      (pData, pDataLen,  dest->dir_uuid);
 269     _unmarshal_uuid      (pData, pDataLen,  dest->stream_uuid);
 270     _unmarshal_uint64    (pData, pDataLen, &dest->unknown_offset);
 271     _unmarshal_uint64    (pData, pDataLen, &dest->unknown_len);
 272     _unmarshal_uint64    (pData, pDataLen, &dest->dir_offset);
 273     _unmarshal_uint64    (pData, pDataLen, &dest->dir_len);
 274
 275     /* error check the data */
 276     /* XXX: should also check UUIDs, probably, though with a version 3 file,
 277      * current MS tools do not seem to use them.
 278      */
 279     if (memcmp(dest->signature, "ITSF", 4) != 0)
 280         return 0;
 281     if (dest->version == 2)
 282     {
 283         if (dest->header_len < _CHM_ITSF_V2_LEN)
 284             return 0;
 285     }
 286     else if (dest->version == 3)
 287     {
 288         if (dest->header_len < _CHM_ITSF_V3_LEN)
 289             return 0;
 290     }
 291     else
 292         return 0;
 293
 294     /* now, if we have a V3 structure, unmarshal the rest.
 295      * otherwise, compute it
 296      */
 297     if (dest->version == 3)
 298     {
 299         if (*pDataLen != 0)
 300             _unmarshal_uint64(pData, pDataLen, &dest->data_offset);
 301         else
 302             return 0;
 303     }
 304     else
 305         dest->data_offset = dest->dir_offset + dest->dir_len;
 306
 307     return 1;
 308 }
 309
 310 /* structure of ITSP headers */
 311 #define _CHM_ITSP_V1_LEN (0x54)
 312 struct chmItspHeader
 313 {
 314     char        signature[4];           /*  0 (ITSP) */
 315     Int32       version;                /*  4 */
 316     Int32       header_len;             /*  8 */
 317     Int32       unknown_000c;           /*  c */
 318     UInt32      block_len;              /* 10 */
 319     Int32       blockidx_intvl;         /* 14 */
 320     Int32       index_depth;            /* 18 */
 321     Int32       index_root;             /* 1c */
 322     Int32       index_head;             /* 20 */
 323     Int32       unknown_0024;           /* 24 */
 324     UInt32      num_blocks;             /* 28 */
 325     Int32       unknown_002c;           /* 2c */
 326     UInt32      lang_id;                /* 30 */
 327     UChar       system_uuid[16];        /* 34 */
 328     UChar       unknown_0044[16];       /* 44 */
 329 }; /* __attribute__ ((aligned (1))); */
 330
 331 static int _unmarshal_itsp_header(unsigned char **pData,
 332                                   unsigned int *pDataLen,
 333                                   struct chmItspHeader *dest)
 334 {
 335     /* we only know how to deal with a 0x54 byte structures */
 336     if (*pDataLen != _CHM_ITSP_V1_LEN)
 337         return 0;
 338
 339     /* unmarshal fields */
 340     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 341     _unmarshal_int32     (pData, pDataLen, &dest->version);
 342     _unmarshal_int32     (pData, pDataLen, &dest->header_len);
 343     _unmarshal_int32     (pData, pDataLen, &dest->unknown_000c);
 344     _unmarshal_uint32    (pData, pDataLen, &dest->block_len);
 345     _unmarshal_int32     (pData, pDataLen, &dest->blockidx_intvl);
 346     _unmarshal_int32     (pData, pDataLen, &dest->index_depth);
 347     _unmarshal_int32     (pData, pDataLen, &dest->index_root);
 348     _unmarshal_int32     (pData, pDataLen, &dest->index_head);
 349     _unmarshal_int32     (pData, pDataLen, &dest->unknown_0024);
 350     _unmarshal_uint32    (pData, pDataLen, &dest->num_blocks);
 351     _unmarshal_int32     (pData, pDataLen, &dest->unknown_002c);
 352     _unmarshal_uint32    (pData, pDataLen, &dest->lang_id);
 353     _unmarshal_uuid      (pData, pDataLen,  dest->system_uuid);
 354     _unmarshal_uchar_array(pData, pDataLen, dest->unknown_0044, 16);
 355
 356     /* error check the data */
 357     if (memcmp(dest->signature, "ITSP", 4) != 0)
 358         return 0;
 359     if (dest->version != 1)
 360         return 0;
 361     if (dest->header_len != _CHM_ITSP_V1_LEN)
 362         return 0;
 363
 364     return 1;
 365 }
 366
 367 /* structure of PMGL headers */
 368 static const char _chm_pmgl_marker[4] = "PMGL";
 369 #define _CHM_PMGL_LEN (0x14)
 370 struct chmPmglHeader
 371 {
 372     char        signature[4];           /*  0 (PMGL) */
 373     UInt32      free_space;             /*  4 */
 374     UInt32      unknown_0008;           /*  8 */
 375     Int32       block_prev;             /*  c */
 376     Int32       block_next;             /* 10 */
 377 }; /* __attribute__ ((aligned (1))); */
 378
 379 static int _unmarshal_pmgl_header(unsigned char **pData,
 380                                   unsigned int *pDataLen,
 381                                   struct chmPmglHeader *dest)
 382 {
 383     /* we only know how to deal with a 0x14 byte structures */
 384     if (*pDataLen != _CHM_PMGL_LEN)
 385         return 0;
 386
 387     /* unmarshal fields */
 388     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 389     _unmarshal_uint32    (pData, pDataLen, &dest->free_space);
 390     _unmarshal_uint32    (pData, pDataLen, &dest->unknown_0008);
 391     _unmarshal_int32     (pData, pDataLen, &dest->block_prev);
 392     _unmarshal_int32     (pData, pDataLen, &dest->block_next);
 393
 394     /* check structure */
 395     if (memcmp(dest->signature, _chm_pmgl_marker, 4) != 0)
 396         return 0;
 397
 398     return 1;
 399 }
 400
 401 /* structure of PMGI headers */
 402 static const char _chm_pmgi_marker[4] = "PMGI";
 403 #define _CHM_PMGI_LEN (0x08)
 404 struct chmPmgiHeader
 405 {
 406     char        signature[4];           /*  0 (PMGI) */
 407     UInt32      free_space;             /*  4 */
 408 }; /* __attribute__ ((aligned (1))); */
 409
 410 static int _unmarshal_pmgi_header(unsigned char **pData,
 411                                   unsigned int *pDataLen,
 412                                   struct chmPmgiHeader *dest)
 413 {
 414     /* we only know how to deal with a 0x8 byte structures */
 415     if (*pDataLen != _CHM_PMGI_LEN)
 416         return 0;
 417
 418     /* unmarshal fields */
 419     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 420     _unmarshal_uint32    (pData, pDataLen, &dest->free_space);
 421
 422     /* check structure */
 423     if (memcmp(dest->signature, _chm_pmgi_marker, 4) != 0)
 424         return 0;
 425
 426     return 1;
 427 }
 428
 429 /* structure of LZXC reset table */
 430 #define _CHM_LZXC_RESETTABLE_V1_LEN (0x28)
 431 struct chmLzxcResetTable
 432 {
 433     UInt32      version;
 434     UInt32      block_count;
 435     UInt32      unknown;
 436     UInt32      table_offset;
 437     UInt64      uncompressed_len;
 438     UInt64      compressed_len;
 439     UInt64      block_len;
 440 }; /* __attribute__ ((aligned (1))); */
 441
 442 static int _unmarshal_lzxc_reset_table(unsigned char **pData,
 443                                        unsigned int *pDataLen,
 444                                        struct chmLzxcResetTable *dest)
 445 {
 446     /* we only know how to deal with a 0x28 byte structures */
 447     if (*pDataLen != _CHM_LZXC_RESETTABLE_V1_LEN)
 448         return 0;
 449
 450     /* unmarshal fields */
 451     _unmarshal_uint32    (pData, pDataLen, &dest->version);
 452     _unmarshal_uint32    (pData, pDataLen, &dest->block_count);
 453     _unmarshal_uint32    (pData, pDataLen, &dest->unknown);
 454     _unmarshal_uint32    (pData, pDataLen, &dest->table_offset);
 455     _unmarshal_uint64    (pData, pDataLen, &dest->uncompressed_len);
 456     _unmarshal_uint64    (pData, pDataLen, &dest->compressed_len);
 457     _unmarshal_uint64    (pData, pDataLen, &dest->block_len);
 458
 459     /* check structure */
 460     if (dest->version != 2)
 461         return 0;
 462
 463     return 1;
 464 }
 465
 466 /* structure of LZXC control data block */
 467 #define _CHM_LZXC_MIN_LEN (0x18)
 468 #define _CHM_LZXC_V2_LEN (0x1c)
 469 struct chmLzxcControlData
 470 {
 471     UInt32      size;                   /*  0        */
 472     char        signature[4];           /*  4 (LZXC) */
 473     UInt32      version;                /*  8        */
 474     UInt32      resetInterval;          /*  c        */
 475     UInt32      windowSize;             /* 10        */
 476     UInt32      windowsPerReset;        /* 14        */
 477     UInt32      unknown_18;             /* 18        */
 478 };
 479
 480 static int _unmarshal_lzxc_control_data(unsigned char **pData,
 481                                         unsigned int *pDataLen,
 482                                         struct chmLzxcControlData *dest)
 483 {
 484     /* we want at least 0x18 bytes */
 485     if (*pDataLen < _CHM_LZXC_MIN_LEN)
 486         return 0;
 487
 488     /* unmarshal fields */
 489     _unmarshal_uint32    (pData, pDataLen, &dest->size);
 490     _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
 491     _unmarshal_uint32    (pData, pDataLen, &dest->version);
 492     _unmarshal_uint32    (pData, pDataLen, &dest->resetInterval);
 493     _unmarshal_uint32    (pData, pDataLen, &dest->windowSize);
 494     _unmarshal_uint32    (pData, pDataLen, &dest->windowsPerReset);
 495
 496     if (*pDataLen >= _CHM_LZXC_V2_LEN)
 497         _unmarshal_uint32    (pData, pDataLen, &dest->unknown_18);
 498     else
 499         dest->unknown_18 = 0;
 500
 501     if (dest->version == 2)
 502     {
 503         dest->resetInterval *= 0x8000;
 504         dest->windowSize *= 0x8000;
 505     }
 506     if (dest->windowSize == 0  ||  dest->resetInterval == 0)
 507         return 0;
 508
 509     /* for now, only support resetInterval a multiple of windowSize/2 */
 510     if (dest->windowSize == 1)
 511         return 0;
 512     if ((dest->resetInterval % (dest->windowSize/2)) != 0)
 513         return 0;
 514
 515     /* check structure */
 516     if (memcmp(dest->signature, "LZXC", 4) != 0)
 517         return 0;
 518
 519     return 1;
 520 }
 521
 522 /* the structure used for chm file handles */
 523 struct chmFile
 524 {
 525     HANDLE              fd;
 526
 527     CRITICAL_SECTION    mutex;
 528     CRITICAL_SECTION    lzx_mutex;
 529     CRITICAL_SECTION    cache_mutex;
 530
 531     UInt64              dir_offset;
 532     UInt64              dir_len;
 533     UInt64              data_offset;
 534     Int32               index_root;
 535     Int32               index_head;
 536     UInt32              block_len;
 537
 538     UInt64              span;
 539     struct chmUnitInfo  rt_unit;
 540     struct chmUnitInfo  cn_unit;
 541     struct chmLzxcResetTable reset_table;
 542
 543     /* LZX control data */
 544     int                 compression_enabled;
 545     UInt32              window_size;
 546     UInt32              reset_interval;
 547     UInt32              reset_blkcount;
 548
 549     /* decompressor state */
 550     struct LZXstate    *lzx_state;
 551     int                 lzx_last_block;
 552
 553     /* cache for decompressed blocks */
 554     UChar             **cache_blocks;
 555     Int64              *cache_block_indices;
 556     Int32               cache_num_blocks;
 557 };
 558
 559 /*
 560  * utility functions local to this module
 561  */
 562
 563 /* utility function to handle differences between {pread,read}(64)? */
 564 static Int64 _chm_fetch_bytes(struct chmFile *h,
 565                               UChar *buf,
 566                               UInt64 os,
 567                               Int64 len)
 568 {
 569     Int64 readLen=0;
 570     if (h->fd  ==  CHM_NULL_FD)
 571         return readLen;
 572
 573     CHM_ACQUIRE_LOCK(h->mutex);
 574     /* NOTE: this might be better done with CreateFileMapping, et cetera... */
 575     {
 576         LARGE_INTEGER old_pos, new_pos;
 577         DWORD actualLen=0;
 578
 579         /* awkward Win32 Seek/Tell */
 580         new_pos.QuadPart = 0;
 581         SetFilePointerEx( h->fd, new_pos, &old_pos, FILE_CURRENT );
 582         new_pos.QuadPart = os;
 583         SetFilePointerEx( h->fd, new_pos, NULL, FILE_BEGIN );
 584
 585         /* read the data */
 586         if (ReadFile(h->fd,
 587                      buf,
 588                      (DWORD)len,
 589                      &actualLen,
 590                      NULL))
 591             readLen = actualLen;
 592         else
 593             readLen = 0;
 594
 595         /* restore original position */
 596         SetFilePointerEx( h->fd, old_pos, NULL, FILE_BEGIN );
 597     }
 598     CHM_RELEASE_LOCK(h->mutex);
 599     return readLen;
 600 }
 601
 602 /* open an ITS archive */
 603 struct chmFile *chm_openW(const WCHAR *filename)
 604 {
 605     unsigned char               sbuffer[256];
 606     unsigned int                sremain;
 607     unsigned char              *sbufpos;
 608     struct chmFile             *newHandle=NULL;
 609     struct chmItsfHeader        itsfHeader;
 610     struct chmItspHeader        itspHeader;
 611 #if 0
 612     struct chmUnitInfo          uiSpan;
 613 #endif
 614     struct chmUnitInfo          uiLzxc;
 615     struct chmLzxcControlData   ctlData;
 616
 617     /* allocate handle */
 618     newHandle = malloc(sizeof(struct chmFile));
 619     newHandle->fd = CHM_NULL_FD;
 620     newHandle->lzx_state = NULL;
 621     newHandle->cache_blocks = NULL;
 622     newHandle->cache_block_indices = NULL;
 623     newHandle->cache_num_blocks = 0;
 624
 625     /* open file */
 626     if ((newHandle->fd=CreateFileW(filename,
 627                                    GENERIC_READ,
 628                                    FILE_SHARE_READ,
 629                                    NULL,
 630                                    OPEN_EXISTING,
 631                                    FILE_ATTRIBUTE_NORMAL,
 632                                    NULL)) == CHM_NULL_FD)
 633     {
 634         free(newHandle);
 635         return NULL;
 636     }
 637
 638     /* initialize mutexes, if needed */
 639     InitializeCriticalSection(&newHandle->mutex);
 640     InitializeCriticalSection(&newHandle->lzx_mutex);
 641     InitializeCriticalSection(&newHandle->cache_mutex);
 642
 643     /* read and verify header */
 644     sremain = _CHM_ITSF_V3_LEN;
 645     sbufpos = sbuffer;
 646     if (_chm_fetch_bytes(newHandle, sbuffer, (UInt64)0, sremain) != sremain    ||
 647         !_unmarshal_itsf_header(&sbufpos, &sremain, &itsfHeader))
 648     {
 649         chm_close(newHandle);
 650         return NULL;
 651     }
 652
 653     /* stash important values from header */
 654     newHandle->dir_offset  = itsfHeader.dir_offset;
 655     newHandle->dir_len     = itsfHeader.dir_len;
 656     newHandle->data_offset = itsfHeader.data_offset;
 657
 658     /* now, read and verify the directory header chunk */
 659     sremain = _CHM_ITSP_V1_LEN;
 660     sbufpos = sbuffer;
 661     if (_chm_fetch_bytes(newHandle, sbuffer,
 662                          (UInt64)itsfHeader.dir_offset, sremain) != sremain       ||
 663         !_unmarshal_itsp_header(&sbufpos, &sremain, &itspHeader))
 664     {
 665         chm_close(newHandle);
 666         return NULL;
 667     }
 668
 669     /* grab essential information from ITSP header */
 670     newHandle->dir_offset += itspHeader.header_len;
 671     newHandle->dir_len    -= itspHeader.header_len;
 672     newHandle->index_root  = itspHeader.index_root;
 673     newHandle->index_head  = itspHeader.index_head;
 674     newHandle->block_len   = itspHeader.block_len;
 675
 676     /* if the index root is -1, this means we don't have any PMGI blocks.
 677      * as a result, we must use the sole PMGL block as the index root
 678      */
 679     if (newHandle->index_root == -1)
 680         newHandle->index_root = newHandle->index_head;
 681
 682     /* By default, compression is enabled. */
 683     newHandle->compression_enabled = 1;
 684
 685 /* Jed, Sun Jun 27: 'span' doesn't seem to be used anywhere?! */
 686 #if 0
 687     /* fetch span */
 688     if (CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
 689                                                   _CHMU_SPANINFO,
 690                                                   &uiSpan)                ||
 691         uiSpan.space == CHM_COMPRESSED)
 692     {
 693         chm_close(newHandle);
 694         return NULL;
 695     }
 696
 697     /* N.B.: we've already checked that uiSpan is in the uncompressed section,
 698      *       so this should not require attempting to decompress, which may
 699      *       rely on having a valid "span"
 700      */
 701     sremain = 8;
 702     sbufpos = sbuffer;
 703     if (chm_retrieve_object(newHandle, &uiSpan, sbuffer,
 704                             0, sremain) != sremain                        ||
 705         !_unmarshal_uint64(&sbufpos, &sremain, &newHandle->span))
 706     {
 707         chm_close(newHandle);
 708         return NULL;
 709     }
 710 #endif
 711
 712     /* prefetch most commonly needed unit infos */
 713     if (CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
 714                                                   _CHMU_RESET_TABLE,
 715                                                   &newHandle->rt_unit)    ||
 716         newHandle->rt_unit.space == CHM_COMPRESSED                        ||
 717         CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
 718                                                   _CHMU_CONTENT,
 719                                                   &newHandle->cn_unit)    ||
 720         newHandle->cn_unit.space == CHM_COMPRESSED                        ||
 721         CHM_RESOLVE_SUCCESS != chm_resolve_object(newHandle,
 722                                                   _CHMU_LZXC_CONTROLDATA,
 723                                                   &uiLzxc)                ||
 724         uiLzxc.space == CHM_COMPRESSED)
 725     {
 726         newHandle->compression_enabled = 0;
 727     }
 728
 729     /* read reset table info */
 730     if (newHandle->compression_enabled)
 731     {
 732         sremain = _CHM_LZXC_RESETTABLE_V1_LEN;
 733         sbufpos = sbuffer;
 734         if (chm_retrieve_object(newHandle, &newHandle->rt_unit, sbuffer,
 735                                 0, sremain) != sremain                        ||
 736             !_unmarshal_lzxc_reset_table(&sbufpos, &sremain,
 737                                          &newHandle->reset_table))
 738         {
 739             newHandle->compression_enabled = 0;
 740         }
 741     }
 742
 743     /* read control data */
 744     if (newHandle->compression_enabled)
 745     {
 746         sremain = (unsigned long)uiLzxc.length;
 747         sbufpos = sbuffer;
 748         if (chm_retrieve_object(newHandle, &uiLzxc, sbuffer,
 749                                 0, sremain) != sremain                       ||
 750             !_unmarshal_lzxc_control_data(&sbufpos, &sremain,
 751                                           &ctlData))
 752         {
 753             newHandle->compression_enabled = 0;
 754         }
 755
 756         newHandle->window_size = ctlData.windowSize;
 757         newHandle->reset_interval = ctlData.resetInterval;
 758
 759 /* Jed, Mon Jun 28: Experimentally, it appears that the reset block count */
 760 /*       must be multiplied by this formerly unknown ctrl data field in   */
 761 /*       order to decompress some files.                                  */
 762 #if 0
 763         newHandle->reset_blkcount = newHandle->reset_interval /
 764                     (newHandle->window_size / 2);
 765 #else
 766         newHandle->reset_blkcount = newHandle->reset_interval    /
 767                                     (newHandle->window_size / 2) *
 768                                     ctlData.windowsPerReset;
 769 #endif
 770     }
 771
 772     /* initialize cache */
 773     chm_set_param(newHandle, CHM_PARAM_MAX_BLOCKS_CACHED,
 774                   CHM_MAX_BLOCKS_CACHED);
 775
 776     return newHandle;
 777 }
 778
 779 /* close an ITS archive */
 780 void chm_close(struct chmFile *h)
 781 {
 782     if (h != NULL)
 783     {
 784         if (h->fd != CHM_NULL_FD)
 785             CHM_CLOSE_FILE(h->fd);
 786         h->fd = CHM_NULL_FD;
 787
 788         DeleteCriticalSection(&h->mutex);
 789         DeleteCriticalSection(&h->lzx_mutex);
 790         DeleteCriticalSection(&h->cache_mutex);
 791
 792         if (h->lzx_state)
 793             LZXteardown(h->lzx_state);
 794         h->lzx_state = NULL;
 795
 796         if (h->cache_blocks)
 797         {
 798             int i;
 799             for (i=0; i<h->cache_num_blocks; i++)
 800             {
 801                 if (h->cache_blocks[i])
 802                     free(h->cache_blocks[i]);
 803             }
 804             free(h->cache_blocks);
 805             h->cache_blocks = NULL;
 806         }
 807
 808         free(h->cache_block_indices);
 809         h->cache_block_indices = NULL;
 810
 811         free(h);
 812     }
 813 }
 814
 815 /*
 816  * set a parameter on the file handle.
 817  * valid parameter types:
 818  *          CHM_PARAM_MAX_BLOCKS_CACHED:
 819  *                 how many decompressed blocks should be cached?  A simple
 820  *                 caching scheme is used, wherein the index of the block is
 821  *                 used as a hash value, and hash collision results in the
 822  *                 invalidation of the previously cached block.
 823  */
 824 void chm_set_param(struct chmFile *h,
 825                    int paramType,
 826                    int paramVal)
 827 {
 828     switch (paramType)
 829     {
 830         case CHM_PARAM_MAX_BLOCKS_CACHED:
 831             CHM_ACQUIRE_LOCK(h->cache_mutex);
 832             if (paramVal != h->cache_num_blocks)
 833             {
 834                 UChar **newBlocks;
 835                 Int64 *newIndices;
 836                 int     i;
 837
 838                 /* allocate new cached blocks */
 839                 newBlocks = malloc(paramVal * sizeof (UChar *));
 840                 newIndices = malloc(paramVal * sizeof (UInt64));
 841                 for (i=0; i<paramVal; i++)
 842                 {
 843                     newBlocks[i] = NULL;
 844                     newIndices[i] = 0;
 845                 }
 846
 847                 /* re-distribute old cached blocks */
 848                 if (h->cache_blocks)
 849                 {
 850                     for (i=0; i<h->cache_num_blocks; i++)
 851                     {
 852                         int newSlot = (int)(h->cache_block_indices[i] % paramVal);
 853
 854                         if (h->cache_blocks[i])
 855                         {
 856                             /* in case of collision, destroy newcomer */
 857                             if (newBlocks[newSlot])
 858                             {
 859                                 free(h->cache_blocks[i]);
 860                                 h->cache_blocks[i] = NULL;
 861                             }
 862                             else
 863                             {
 864                                 newBlocks[newSlot] = h->cache_blocks[i];
 865                                 newIndices[newSlot] =
 866                                             h->cache_block_indices[i];
 867                             }
 868                         }
 869                     }
 870
 871                     free(h->cache_blocks);
 872                     free(h->cache_block_indices);
 873                 }
 874
 875                 /* now, set new values */
 876                 h->cache_blocks = newBlocks;
 877                 h->cache_block_indices = newIndices;
 878                 h->cache_num_blocks = paramVal;
 879             }
 880             CHM_RELEASE_LOCK(h->cache_mutex);
 881             break;
 882
 883         default:
 884             break;
 885     }
 886 }
 887
 888 /*
 889  * helper methods for chm_resolve_object
 890  */
 891
 892 /* skip a compressed dword */
 893 static void _chm_skip_cword(UChar **pEntry)
 894 {
 895     while (*(*pEntry)++ >= 0x80)
 896         ;
 897 }
 898
 899 /* skip the data from a PMGL entry */
 900 static void _chm_skip_PMGL_entry_data(UChar **pEntry)
 901 {
 902     _chm_skip_cword(pEntry);
 903     _chm_skip_cword(pEntry);
 904     _chm_skip_cword(pEntry);
 905 }
 906
 907 /* parse a compressed dword */
 908 static UInt64 _chm_parse_cword(UChar **pEntry)
 909 {
 910     UInt64 accum = 0;
 911     UChar temp;
 912     while ((temp=*(*pEntry)++) >= 0x80)
 913     {
 914         accum <<= 7;
 915         accum += temp & 0x7f;
 916     }
 917
 918     return (accum << 7) + temp;
 919 }
 920
 921 /* parse a utf-8 string into an ASCII char buffer */
 922 static int _chm_parse_UTF8(UChar **pEntry, UInt64 count, WCHAR *path)
 923 {
 924     /* MJM - Modified to return real Unicode strings */
 925     while (count != 0)
 926     {
 927         *path++ = (*(*pEntry)++);
 928         --count;
 929     }
 930
 931     *path = '\0';
 932     return 1;
 933 }
 934
 935 /* parse a PMGL entry into a chmUnitInfo struct; return 1 on success. */
 936 static int _chm_parse_PMGL_entry(UChar **pEntry, struct chmUnitInfo *ui)
 937 {
 938     UInt64 strLen;
 939
 940     /* parse str len */
 941     strLen = _chm_parse_cword(pEntry);
 942     if (strLen > CHM_MAX_PATHLEN)
 943         return 0;
 944
 945     /* parse path */
 946     if (! _chm_parse_UTF8(pEntry, strLen, ui->path))
 947         return 0;
 948
 949     /* parse info */
 950     ui->space  = (int)_chm_parse_cword(pEntry);
 951     ui->start  = _chm_parse_cword(pEntry);
 952     ui->length = _chm_parse_cword(pEntry);
 953     return 1;
 954 }
 955
 956 /* find an exact entry in PMGL; return NULL if we fail */
 957 static UChar *_chm_find_in_PMGL(UChar *page_buf,
 958                          UInt32 block_len,
 959                          const WCHAR *objPath)
 960 {
 961     /* XXX: modify this to do a binary search using the nice index structure
 962      *      that is provided for us.
 963      */
 964     struct chmPmglHeader header;
 965     UInt32 hremain;
 966     UChar *end;
 967     UChar *cur;
 968     UChar *temp;
 969     UInt64 strLen;
 970     WCHAR buffer[CHM_MAX_PATHLEN+1];
 971
 972     /* figure out where to start and end */
 973     cur = page_buf;
 974     hremain = _CHM_PMGL_LEN;
 975     if (! _unmarshal_pmgl_header(&cur, &hremain, &header))
 976         return NULL;
 977     end = page_buf + block_len - (header.free_space);
 978
 979     /* now, scan progressively */
 980     while (cur < end)
 981     {
 982         /* grab the name */
 983         temp = cur;
 984         strLen = _chm_parse_cword(&cur);
 985         if (! _chm_parse_UTF8(&cur, strLen, buffer))
 986             return NULL;
 987
 988         /* check if it is the right name */
 989         if (! strcmpiW(buffer, objPath))
 990             return temp;
 991
 992         _chm_skip_PMGL_entry_data(&cur);
 993     }
 994
 995     return NULL;
 996 }
 997
 998 /* find which block should be searched next for the entry; -1 if no block */
 999 static Int32 _chm_find_in_PMGI(UChar *page_buf,
1000                         UInt32 block_len,
1001                         const WCHAR *objPath)
1002 {
1003     /* XXX: modify this to do a binary search using the nice index structure
1004      *      that is provided for us
1005      */
1006     struct chmPmgiHeader header;
1007     UInt32 hremain;
1008     int page=-1;
1009     UChar *end;
1010     UChar *cur;
1011     UInt64 strLen;
1012     WCHAR buffer[CHM_MAX_PATHLEN+1];
1013
1014     /* figure out where to start and end */
1015     cur = page_buf;
1016     hremain = _CHM_PMGI_LEN;
1017     if (! _unmarshal_pmgi_header(&cur, &hremain, &header))
1018         return -1;
1019     end = page_buf + block_len - (header.free_space);
1020
1021     /* now, scan progressively */
1022     while (cur < end)
1023     {
1024         /* grab the name */
1025         strLen = _chm_parse_cword(&cur);
1026         if (! _chm_parse_UTF8(&cur, strLen, buffer))
1027             return -1;
1028
1029         /* check if it is the right name */
1030         if (strcmpiW(buffer, objPath) > 0)
1031             return page;
1032
1033         /* load next value for path */
1034         page = (int)_chm_parse_cword(&cur);
1035     }
1036
1037     return page;
1038 }
1039
1040 /* resolve a particular object from the archive */
1041 int chm_resolve_object(struct chmFile *h,
1042                        const WCHAR *objPath,
1043                        struct chmUnitInfo *ui)
1044 {
1045     /*
1046      * XXX: implement caching scheme for dir pages
1047      */
1048
1049     Int32 curPage;
1050
1051     /* buffer to hold whatever page we're looking at */
1052     UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, h->block_len);
1053
1054     /* starting page */
1055     curPage = h->index_root;
1056
1057     /* until we have either returned or given up */
1058     while (curPage != -1)
1059     {
1060
1061         /* try to fetch the index page */
1062         if (_chm_fetch_bytes(h, page_buf,
1063                              (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1064                              h->block_len) != h->block_len)
1065         {
1066             HeapFree(GetProcessHeap(), 0, page_buf);
1067             return CHM_RESOLVE_FAILURE;
1068         }
1069
1070         /* now, if it is a leaf node: */
1071         if (memcmp(page_buf, _chm_pmgl_marker, 4) == 0)
1072         {
1073             /* scan block */
1074             UChar *pEntry = _chm_find_in_PMGL(page_buf,
1075                                               h->block_len,
1076                                               objPath);
1077             if (pEntry == NULL)
1078             {
1079                 HeapFree(GetProcessHeap(), 0, page_buf);
1080                 return CHM_RESOLVE_FAILURE;
1081             }
1082
1083             /* parse entry and return */
1084             _chm_parse_PMGL_entry(&pEntry, ui);
1085             HeapFree(GetProcessHeap(), 0, page_buf);
1086             return CHM_RESOLVE_SUCCESS;
1087         }
1088
1089         /* else, if it is a branch node: */
1090         else if (memcmp(page_buf, _chm_pmgi_marker, 4) == 0)
1091             curPage = _chm_find_in_PMGI(page_buf, h->block_len, objPath);
1092
1093         /* else, we are confused.  give up. */
1094         else
1095         {
1096             HeapFree(GetProcessHeap(), 0, page_buf);
1097             return CHM_RESOLVE_FAILURE;
1098         }
1099     }
1100
1101     /* didn't find anything.  fail. */
1102     HeapFree(GetProcessHeap(), 0, page_buf);
1103     return CHM_RESOLVE_FAILURE;
1104 }
1105
1106 /*
1107  * utility methods for dealing with compressed data
1108  */
1109
1110 /* get the bounds of a compressed block.  return 0 on failure */
1111 static int _chm_get_cmpblock_bounds(struct chmFile *h,
1112                              UInt64 block,
1113                              UInt64 *start,
1114                              Int64 *len)
1115 {
1116     UChar buffer[8], *dummy;
1117     UInt32 remain;
1118
1119     /* for all but the last block, use the reset table */
1120     if (block < h->reset_table.block_count-1)
1121     {
1122         /* unpack the start address */
1123         dummy = buffer;
1124         remain = 8;
1125         if (_chm_fetch_bytes(h, buffer,
1126                              (UInt64)h->data_offset
1127                                 + (UInt64)h->rt_unit.start
1128                                 + (UInt64)h->reset_table.table_offset
1129                                 + (UInt64)block*8,
1130                              remain) != remain                            ||
1131             !_unmarshal_uint64(&dummy, &remain, start))
1132             return 0;
1133
1134         /* unpack the end address */
1135         dummy = buffer;
1136         remain = 8;
1137         if (_chm_fetch_bytes(h, buffer,
1138                          (UInt64)h->data_offset
1139                                 + (UInt64)h->rt_unit.start
1140                                 + (UInt64)h->reset_table.table_offset
1141                                 + (UInt64)block*8 + 8,
1142                          remain) != remain                                ||
1143             !_unmarshal_int64(&dummy, &remain, len))
1144             return 0;
1145     }
1146
1147     /* for the last block, use the span in addition to the reset table */
1148     else
1149     {
1150         /* unpack the start address */
1151         dummy = buffer;
1152         remain = 8;
1153         if (_chm_fetch_bytes(h, buffer,
1154                              (UInt64)h->data_offset
1155                                 + (UInt64)h->rt_unit.start
1156                                 + (UInt64)h->reset_table.table_offset
1157                                 + (UInt64)block*8,
1158                              remain) != remain                            ||
1159             !_unmarshal_uint64(&dummy, &remain, start))
1160             return 0;
1161
1162         *len = h->reset_table.compressed_len;
1163     }
1164
1165     /* compute the length and absolute start address */
1166     *len -= *start;
1167     *start += h->data_offset + h->cn_unit.start;
1168
1169     return 1;
1170 }
1171
1172 /* decompress the block.  must have lzx_mutex. */
1173 static Int64 _chm_decompress_block(struct chmFile *h,
1174                                    UInt64 block,
1175                                    UChar **ubuffer)
1176 {
1177     UChar *cbuffer = HeapAlloc( GetProcessHeap(), 0,
1178                               ((unsigned int)h->reset_table.block_len + 6144));
1179     UInt64 cmpStart;                                    /* compressed start  */
1180     Int64 cmpLen;                                       /* compressed len    */
1181     int indexSlot;                                      /* cache index slot  */
1182     UChar *lbuffer;                                     /* local buffer ptr  */
1183     UInt32 blockAlign = (UInt32)(block % h->reset_blkcount); /* reset intvl. aln. */
1184     UInt32 i;                                           /* local loop index  */
1185
1186     /* let the caching system pull its weight! */
1187     if (block - blockAlign <= h->lzx_last_block  &&
1188         block              >= h->lzx_last_block)
1189         blockAlign = (block - h->lzx_last_block);
1190
1191     /* check if we need previous blocks */
1192     if (blockAlign != 0)
1193     {
1194         /* fetch all required previous blocks since last reset */
1195         for (i = blockAlign; i > 0; i--)
1196         {
1197             UInt32 curBlockIdx = block - i;
1198
1199             /* check if we most recently decompressed the previous block */
1200             if (h->lzx_last_block != curBlockIdx)
1201             {
1202                 if ((curBlockIdx % h->reset_blkcount) == 0)
1203                 {
1204 #ifdef CHM_DEBUG
1205                     fprintf(stderr, "***RESET (1)***\n");
1206 #endif
1207                     LZXreset(h->lzx_state);
1208                 }
1209
1210                 indexSlot = (int)((curBlockIdx) % h->cache_num_blocks);
1211                 h->cache_block_indices[indexSlot] = curBlockIdx;
1212                 if (! h->cache_blocks[indexSlot])
1213                     h->cache_blocks[indexSlot] = malloc( (unsigned int)(h->reset_table.block_len));
1214                 lbuffer = h->cache_blocks[indexSlot];
1215
1216                 /* decompress the previous block */
1217 #ifdef CHM_DEBUG
1218                 fprintf(stderr, "Decompressing block #%4d (EXTRA)\n", curBlockIdx);
1219 #endif
1220                 if (!_chm_get_cmpblock_bounds(h, curBlockIdx, &cmpStart, &cmpLen) ||
1221                     _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen      ||
1222                     LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1223                                   (int)h->reset_table.block_len) != DECR_OK)
1224                 {
1225 #ifdef CHM_DEBUG
1226                     fprintf(stderr, "   (DECOMPRESS FAILED!)\n");
1227 #endif
1228                     HeapFree(GetProcessHeap(), 0, cbuffer);
1229                     return (Int64)0;
1230                 }
1231
1232                 h->lzx_last_block = (int)curBlockIdx;
1233             }
1234         }
1235     }
1236     else
1237     {
1238         if ((block % h->reset_blkcount) == 0)
1239         {
1240 #ifdef CHM_DEBUG
1241             fprintf(stderr, "***RESET (2)***\n");
1242 #endif
1243             LZXreset(h->lzx_state);
1244         }
1245     }
1246
1247     /* allocate slot in cache */
1248     indexSlot = (int)(block % h->cache_num_blocks);
1249     h->cache_block_indices[indexSlot] = block;
1250     if (! h->cache_blocks[indexSlot])
1251         h->cache_blocks[indexSlot] = malloc( ((unsigned int)h->reset_table.block_len));
1252     lbuffer = h->cache_blocks[indexSlot];
1253     *ubuffer = lbuffer;
1254
1255     /* decompress the block we actually want */
1256 #ifdef CHM_DEBUG
1257     fprintf(stderr, "Decompressing block #%4d (REAL )\n", block);
1258 #endif
1259     if (! _chm_get_cmpblock_bounds(h, block, &cmpStart, &cmpLen)          ||
1260         _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen          ||
1261         LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1262                       (int)h->reset_table.block_len) != DECR_OK)
1263     {
1264 #ifdef CHM_DEBUG
1265         fprintf(stderr, "   (DECOMPRESS FAILED!)\n");
1266 #endif
1267         HeapFree(GetProcessHeap(), 0, cbuffer);
1268         return (Int64)0;
1269     }
1270     h->lzx_last_block = (int)block;
1271
1272     /* XXX: modify LZX routines to return the length of the data they
1273      * decompressed and return that instead, for an extra sanity check.
1274      */
1275     HeapFree(GetProcessHeap(), 0, cbuffer);
1276     return h->reset_table.block_len;
1277 }
1278
1279 /* grab a region from a compressed block */
1280 static Int64 _chm_decompress_region(struct chmFile *h,
1281                                     UChar *buf,
1282                                     UInt64 start,
1283                                     Int64 len)
1284 {
1285     UInt64 nBlock, nOffset;
1286     UInt64 nLen;
1287     UInt64 gotLen;
1288     UChar *ubuffer = NULL;
1289
1290         if (len <= 0)
1291                 return (Int64)0;
1292
1293     /* figure out what we need to read */
1294     nBlock = start / h->reset_table.block_len;
1295     nOffset = start % h->reset_table.block_len;
1296     nLen = len;
1297     if (nLen > (h->reset_table.block_len - nOffset))
1298         nLen = h->reset_table.block_len - nOffset;
1299
1300     /* if block is cached, return data from it. */
1301     CHM_ACQUIRE_LOCK(h->lzx_mutex);
1302     CHM_ACQUIRE_LOCK(h->cache_mutex);
1303     if (h->cache_block_indices[nBlock % h->cache_num_blocks] == nBlock    &&
1304         h->cache_blocks[nBlock % h->cache_num_blocks] != NULL)
1305     {
1306         memcpy(buf,
1307                h->cache_blocks[nBlock % h->cache_num_blocks] + nOffset,
1308                (unsigned int)nLen);
1309         CHM_RELEASE_LOCK(h->cache_mutex);
1310         CHM_RELEASE_LOCK(h->lzx_mutex);
1311         return nLen;
1312     }
1313     CHM_RELEASE_LOCK(h->cache_mutex);
1314
1315     /* data request not satisfied, so... start up the decompressor machine */
1316     if (! h->lzx_state)
1317     {
1318         int window_size = ffs(h->window_size) - 1;
1319         h->lzx_last_block = -1;
1320         h->lzx_state = LZXinit(window_size);
1321     }
1322
1323     /* decompress some data */
1324     gotLen = _chm_decompress_block(h, nBlock, &ubuffer);
1325     if (gotLen < nLen)
1326         nLen = gotLen;
1327     memcpy(buf, ubuffer+nOffset, (unsigned int)nLen);
1328     CHM_RELEASE_LOCK(h->lzx_mutex);
1329     return nLen;
1330 }
1331
1332 /* retrieve (part of) an object */
1333 LONGINT64 chm_retrieve_object(struct chmFile *h,
1334                                struct chmUnitInfo *ui,
1335                                unsigned char *buf,
1336                                LONGUINT64 addr,
1337                                LONGINT64 len)
1338 {
1339     /* must be valid file handle */
1340     if (h == NULL)
1341         return (Int64)0;
1342
1343     /* starting address must be in correct range */
1344     if (addr < 0  ||  addr >= ui->length)
1345         return (Int64)0;
1346
1347     /* clip length */
1348     if (addr + len > ui->length)
1349         len = ui->length - addr;
1350
1351     /* if the file is uncompressed, it's simple */
1352     if (ui->space == CHM_UNCOMPRESSED)
1353     {
1354         /* read data */
1355         return _chm_fetch_bytes(h,
1356                                 buf,
1357                                 (UInt64)h->data_offset + (UInt64)ui->start + (UInt64)addr,
1358                                 len);
1359     }
1360
1361     /* else if the file is compressed, it's a little trickier */
1362     else /* ui->space == CHM_COMPRESSED */
1363     {
1364         Int64 swath=0, total=0;
1365
1366         /* if compression is not enabled for this file... */
1367         if (! h->compression_enabled)
1368             return total;
1369
1370         do {
1371
1372             /* swill another mouthful */
1373             swath = _chm_decompress_region(h, buf, ui->start + addr, len);
1374
1375             /* if we didn't get any... */
1376             if (swath == 0)
1377                 return total;
1378
1379             /* update stats */
1380             total += swath;
1381             len -= swath;
1382             addr += swath;
1383             buf += swath;
1384
1385         } while (len != 0);
1386
1387         return total;
1388     }
1389 }
1390
1391 /* enumerate the objects in the .chm archive */
1392 int chm_enumerate(struct chmFile *h,
1393                   int what,
1394                   CHM_ENUMERATOR e,
1395                   void *context)
1396 {
1397     Int32 curPage;
1398
1399     /* buffer to hold whatever page we're looking at */
1400     UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, (unsigned int)h->block_len);
1401     struct chmPmglHeader header;
1402     UChar *end;
1403     UChar *cur;
1404     unsigned int lenRemain;
1405     UInt64 ui_path_len;
1406
1407     /* the current ui */
1408     struct chmUnitInfo ui;
1409     int flag;
1410
1411     /* starting page */
1412     curPage = h->index_head;
1413
1414     /* until we have either returned or given up */
1415     while (curPage != -1)
1416     {
1417
1418         /* try to fetch the index page */
1419         if (_chm_fetch_bytes(h,
1420                              page_buf,
1421                              (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1422                              h->block_len) != h->block_len)
1423         {
1424             HeapFree(GetProcessHeap(), 0, page_buf);
1425             return 0;
1426         }
1427
1428         /* figure out start and end for this page */
1429         cur = page_buf;
1430         lenRemain = _CHM_PMGL_LEN;
1431         if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1432         {
1433             HeapFree(GetProcessHeap(), 0, page_buf);
1434             return 0;
1435         }
1436         end = page_buf + h->block_len - (header.free_space);
1437
1438         /* loop over this page */
1439         while (cur < end)
1440         {
1441             if (! _chm_parse_PMGL_entry(&cur, &ui))
1442             {
1443                 HeapFree(GetProcessHeap(), 0, page_buf);
1444                 return 0;
1445             }
1446
1447             /* get the length of the path */
1448             ui_path_len = strlenW(ui.path)-1;
1449
1450             /* check for DIRS */
1451             if (ui.path[ui_path_len] == '/'  &&  !(what & CHM_ENUMERATE_DIRS))
1452                 continue;
1453
1454             /* check for FILES */
1455             if (ui.path[ui_path_len] != '/'  &&  !(what & CHM_ENUMERATE_FILES))
1456                 continue;
1457
1458             /* check for NORMAL vs. META */
1459             if (ui.path[0] == '/')
1460             {
1461
1462                 /* check for NORMAL vs. SPECIAL */
1463                 if (ui.path[1] == '#'  ||  ui.path[1] == '$')
1464                     flag = CHM_ENUMERATE_SPECIAL;
1465                 else
1466                     flag = CHM_ENUMERATE_NORMAL;
1467             }
1468             else
1469                 flag = CHM_ENUMERATE_META;
1470             if (! (what & flag))
1471                 continue;
1472
1473             /* call the enumerator */
1474             {
1475                 int status = (*e)(h, &ui, context);
1476                 switch (status)
1477                 {
1478                     case CHM_ENUMERATOR_FAILURE:
1479                         HeapFree(GetProcessHeap(), 0, page_buf);
1480                         return 0;
1481                     case CHM_ENUMERATOR_CONTINUE:
1482                         break;
1483                     case CHM_ENUMERATOR_SUCCESS:
1484                         HeapFree(GetProcessHeap(), 0, page_buf);
1485                         return 1;
1486                     default:
1487                         break;
1488                 }
1489             }
1490         }
1491
1492         /* advance to next page */
1493         curPage = header.block_next;
1494     }
1495
1496     HeapFree(GetProcessHeap(), 0, page_buf);
1497     return 1;
1498 }
1499
1500 int chm_enumerate_dir(struct chmFile *h,
1501                       const WCHAR *prefix,
1502                       int what,
1503                       CHM_ENUMERATOR e,
1504                       void *context)
1505 {
1506     /*
1507      * XXX: do this efficiently (i.e. using the tree index)
1508      */
1509
1510     Int32 curPage;
1511
1512     /* buffer to hold whatever page we're looking at */
1513     UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, (unsigned int)h->block_len);
1514     struct chmPmglHeader header;
1515     UChar *end;
1516     UChar *cur;
1517     unsigned int lenRemain;
1518
1519     /* set to 1 once we've started */
1520     int it_has_begun=0;
1521
1522     /* the current ui */
1523     struct chmUnitInfo ui;
1524     int flag;
1525     UInt64 ui_path_len;
1526
1527     /* the length of the prefix */
1528     WCHAR prefixRectified[CHM_MAX_PATHLEN+1];
1529     int prefixLen;
1530     WCHAR lastPath[CHM_MAX_PATHLEN];
1531     int lastPathLen;
1532
1533     /* starting page */
1534     curPage = h->index_head;
1535
1536     /* initialize pathname state */
1537     lstrcpynW(prefixRectified, prefix, CHM_MAX_PATHLEN);
1538     prefixLen = strlenW(prefixRectified);
1539     if (prefixLen != 0)
1540     {
1541         if (prefixRectified[prefixLen-1] != '/')
1542         {
1543             prefixRectified[prefixLen] = '/';
1544             prefixRectified[prefixLen+1] = '\0';
1545             ++prefixLen;
1546         }
1547     }
1548     lastPath[0] = '\0';
1549     lastPathLen = -1;
1550
1551     /* until we have either returned or given up */
1552     while (curPage != -1)
1553     {
1554
1555         /* try to fetch the index page */
1556         if (_chm_fetch_bytes(h,
1557                              page_buf,
1558                              (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1559                              h->block_len) != h->block_len)
1560         {
1561             HeapFree(GetProcessHeap(), 0, page_buf);
1562             return 0;
1563         }
1564
1565         /* figure out start and end for this page */
1566         cur = page_buf;
1567         lenRemain = _CHM_PMGL_LEN;
1568         if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1569         {
1570             HeapFree(GetProcessHeap(), 0, page_buf);
1571             return 0;
1572         }
1573         end = page_buf + h->block_len - (header.free_space);
1574
1575         /* loop over this page */
1576         while (cur < end)
1577         {
1578             if (! _chm_parse_PMGL_entry(&cur, &ui))
1579             {
1580                 HeapFree(GetProcessHeap(), 0, page_buf);
1581                 return 0;
1582             }
1583
1584             /* check if we should start */
1585             if (! it_has_begun)
1586             {
1587                 if (ui.length == 0  &&  strncmpiW(ui.path, prefixRectified, prefixLen) == 0)
1588                     it_has_begun = 1;
1589                 else
1590                     continue;
1591
1592                 if (ui.path[prefixLen] == '\0')
1593                     continue;
1594             }
1595
1596             /* check if we should stop */
1597             else
1598             {
1599                 if (strncmpiW(ui.path, prefixRectified, prefixLen) != 0)
1600                 {
1601                     HeapFree(GetProcessHeap(), 0, page_buf);
1602                     return 1;
1603                 }
1604             }
1605
1606             /* check if we should include this path */
1607             if (lastPathLen != -1)
1608             {
1609                 if (strncmpiW(ui.path, lastPath, lastPathLen) == 0)
1610                     continue;
1611             }
1612             strcpyW(lastPath, ui.path);
1613             lastPathLen = strlenW(lastPath);
1614
1615             /* get the length of the path */
1616             ui_path_len = strlenW(ui.path)-1;
1617
1618             /* check for DIRS */
1619             if (ui.path[ui_path_len] == '/'  &&  !(what & CHM_ENUMERATE_DIRS))
1620                 continue;
1621
1622             /* check for FILES */
1623             if (ui.path[ui_path_len] != '/'  &&  !(what & CHM_ENUMERATE_FILES))
1624                 continue;
1625
1626             /* check for NORMAL vs. META */
1627             if (ui.path[0] == '/')
1628             {
1629
1630                 /* check for NORMAL vs. SPECIAL */
1631                 if (ui.path[1] == '#'  ||  ui.path[1] == '$')
1632                     flag = CHM_ENUMERATE_SPECIAL;
1633                 else
1634                     flag = CHM_ENUMERATE_NORMAL;
1635             }
1636             else
1637                 flag = CHM_ENUMERATE_META;
1638             if (! (what & flag))
1639                 continue;
1640
1641             /* call the enumerator */
1642             {
1643                 int status = (*e)(h, &ui, context);
1644                 switch (status)
1645                 {
1646                     case CHM_ENUMERATOR_FAILURE:
1647                         HeapFree(GetProcessHeap(), 0, page_buf);
1648                         return 0;
1649                     case CHM_ENUMERATOR_CONTINUE:
1650                         break;
1651                     case CHM_ENUMERATOR_SUCCESS:
1652                         HeapFree(GetProcessHeap(), 0, page_buf);
1653                         return 1;
1654                     default:
1655                         break;
1656                 }
1657             }
1658         }
1659
1660         /* advance to next page */
1661         curPage = header.block_next;
1662     }
1663
1664     HeapFree(GetProcessHeap(), 0, page_buf);
1665     return 1;
1666 }