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 long *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 long *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 long *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 long *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 long *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 long *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 long *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 long *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 long *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 long *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 long *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 long *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 long *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 long               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         if (h->cache_block_indices)
 809             free(h->cache_block_indices);
 810         h->cache_block_indices = NULL;
 811
 812         free(h);
 813     }
 814 }
 815
 816 /*
 817  * set a parameter on the file handle.
 818  * valid parameter types:
 819  *          CHM_PARAM_MAX_BLOCKS_CACHED:
 820  *                 how many decompressed blocks should be cached?  A simple
 821  *                 caching scheme is used, wherein the index of the block is
 822  *                 used as a hash value, and hash collision results in the
 823  *                 invalidation of the previously cached block.
 824  */
 825 void chm_set_param(struct chmFile *h,
 826                    int paramType,
 827                    int paramVal)
 828 {
 829     switch (paramType)
 830     {
 831         case CHM_PARAM_MAX_BLOCKS_CACHED:
 832             CHM_ACQUIRE_LOCK(h->cache_mutex);
 833             if (paramVal != h->cache_num_blocks)
 834             {
 835                 UChar **newBlocks;
 836                 Int64 *newIndices;
 837                 int     i;
 838
 839                 /* allocate new cached blocks */
 840                 newBlocks = malloc(paramVal * sizeof (UChar *));
 841                 newIndices = malloc(paramVal * sizeof (UInt64));
 842                 for (i=0; i<paramVal; i++)
 843                 {
 844                     newBlocks[i] = NULL;
 845                     newIndices[i] = 0;
 846                 }
 847
 848                 /* re-distribute old cached blocks */
 849                 if (h->cache_blocks)
 850                 {
 851                     for (i=0; i<h->cache_num_blocks; i++)
 852                     {
 853                         int newSlot = (int)(h->cache_block_indices[i] % paramVal);
 854
 855                         if (h->cache_blocks[i])
 856                         {
 857                             /* in case of collision, destroy newcomer */
 858                             if (newBlocks[newSlot])
 859                             {
 860                                 free(h->cache_blocks[i]);
 861                                 h->cache_blocks[i] = NULL;
 862                             }
 863                             else
 864                             {
 865                                 newBlocks[newSlot] = h->cache_blocks[i];
 866                                 newIndices[newSlot] =
 867                                             h->cache_block_indices[i];
 868                             }
 869                         }
 870                     }
 871
 872                     free(h->cache_blocks);
 873                     free(h->cache_block_indices);
 874                 }
 875
 876                 /* now, set new values */
 877                 h->cache_blocks = newBlocks;
 878                 h->cache_block_indices = newIndices;
 879                 h->cache_num_blocks = paramVal;
 880             }
 881             CHM_RELEASE_LOCK(h->cache_mutex);
 882             break;
 883
 884         default:
 885             break;
 886     }
 887 }
 888
 889 /*
 890  * helper methods for chm_resolve_object
 891  */
 892
 893 /* skip a compressed dword */
 894 static void _chm_skip_cword(UChar **pEntry)
 895 {
 896     while (*(*pEntry)++ >= 0x80)
 897         ;
 898 }
 899
 900 /* skip the data from a PMGL entry */
 901 static void _chm_skip_PMGL_entry_data(UChar **pEntry)
 902 {
 903     _chm_skip_cword(pEntry);
 904     _chm_skip_cword(pEntry);
 905     _chm_skip_cword(pEntry);
 906 }
 907
 908 /* parse a compressed dword */
 909 static UInt64 _chm_parse_cword(UChar **pEntry)
 910 {
 911     UInt64 accum = 0;
 912     UChar temp;
 913     while ((temp=*(*pEntry)++) >= 0x80)
 914     {
 915         accum <<= 7;
 916         accum += temp & 0x7f;
 917     }
 918
 919     return (accum << 7) + temp;
 920 }
 921
 922 /* parse a utf-8 string into an ASCII char buffer */
 923 static int _chm_parse_UTF8(UChar **pEntry, UInt64 count, WCHAR *path)
 924 {
 925     /* MJM - Modified to return real Unicode strings */
 926     while (count != 0)
 927     {
 928         *path++ = (*(*pEntry)++);
 929         --count;
 930     }
 931
 932     *path = '\0';
 933     return 1;
 934 }
 935
 936 /* parse a PMGL entry into a chmUnitInfo struct; return 1 on success. */
 937 static int _chm_parse_PMGL_entry(UChar **pEntry, struct chmUnitInfo *ui)
 938 {
 939     UInt64 strLen;
 940
 941     /* parse str len */
 942     strLen = _chm_parse_cword(pEntry);
 943     if (strLen > CHM_MAX_PATHLEN)
 944         return 0;
 945
 946     /* parse path */
 947     if (! _chm_parse_UTF8(pEntry, strLen, ui->path))
 948         return 0;
 949
 950     /* parse info */
 951     ui->space  = (int)_chm_parse_cword(pEntry);
 952     ui->start  = _chm_parse_cword(pEntry);
 953     ui->length = _chm_parse_cword(pEntry);
 954     return 1;
 955 }
 956
 957 /* find an exact entry in PMGL; return NULL if we fail */
 958 static UChar *_chm_find_in_PMGL(UChar *page_buf,
 959                          UInt32 block_len,
 960                          const WCHAR *objPath)
 961 {
 962     /* XXX: modify this to do a binary search using the nice index structure
 963      *      that is provided for us.
 964      */
 965     struct chmPmglHeader header;
 966     UInt32 hremain;
 967     UChar *end;
 968     UChar *cur;
 969     UChar *temp;
 970     UInt64 strLen;
 971     WCHAR buffer[CHM_MAX_PATHLEN+1];
 972
 973     /* figure out where to start and end */
 974     cur = page_buf;
 975     hremain = _CHM_PMGL_LEN;
 976     if (! _unmarshal_pmgl_header(&cur, &hremain, &header))
 977         return NULL;
 978     end = page_buf + block_len - (header.free_space);
 979
 980     /* now, scan progressively */
 981     while (cur < end)
 982     {
 983         /* grab the name */
 984         temp = cur;
 985         strLen = _chm_parse_cword(&cur);
 986         if (! _chm_parse_UTF8(&cur, strLen, buffer))
 987             return NULL;
 988
 989         /* check if it is the right name */
 990         if (! strcmpiW(buffer, objPath))
 991             return temp;
 992
 993         _chm_skip_PMGL_entry_data(&cur);
 994     }
 995
 996     return NULL;
 997 }
 998
 999 /* find which block should be searched next for the entry; -1 if no block */
1000 static Int32 _chm_find_in_PMGI(UChar *page_buf,
1001                         UInt32 block_len,
1002                         const WCHAR *objPath)
1003 {
1004     /* XXX: modify this to do a binary search using the nice index structure
1005      *      that is provided for us
1006      */
1007     struct chmPmgiHeader header;
1008     UInt32 hremain;
1009     int page=-1;
1010     UChar *end;
1011     UChar *cur;
1012     UInt64 strLen;
1013     WCHAR buffer[CHM_MAX_PATHLEN+1];
1014
1015     /* figure out where to start and end */
1016     cur = page_buf;
1017     hremain = _CHM_PMGI_LEN;
1018     if (! _unmarshal_pmgi_header(&cur, &hremain, &header))
1019         return -1;
1020     end = page_buf + block_len - (header.free_space);
1021
1022     /* now, scan progressively */
1023     while (cur < end)
1024     {
1025         /* grab the name */
1026         strLen = _chm_parse_cword(&cur);
1027         if (! _chm_parse_UTF8(&cur, strLen, buffer))
1028             return -1;
1029
1030         /* check if it is the right name */
1031         if (strcmpiW(buffer, objPath) > 0)
1032             return page;
1033
1034         /* load next value for path */
1035         page = (int)_chm_parse_cword(&cur);
1036     }
1037
1038     return page;
1039 }
1040
1041 /* resolve a particular object from the archive */
1042 int chm_resolve_object(struct chmFile *h,
1043                        const WCHAR *objPath,
1044                        struct chmUnitInfo *ui)
1045 {
1046     /*
1047      * XXX: implement caching scheme for dir pages
1048      */
1049
1050     Int32 curPage;
1051
1052     /* buffer to hold whatever page we're looking at */
1053     UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, h->block_len);
1054
1055     /* starting page */
1056     curPage = h->index_root;
1057
1058     /* until we have either returned or given up */
1059     while (curPage != -1)
1060     {
1061
1062         /* try to fetch the index page */
1063         if (_chm_fetch_bytes(h, page_buf,
1064                              (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1065                              h->block_len) != h->block_len)
1066         {
1067             HeapFree(GetProcessHeap(), 0, page_buf);
1068             return CHM_RESOLVE_FAILURE;
1069         }
1070
1071         /* now, if it is a leaf node: */
1072         if (memcmp(page_buf, _chm_pmgl_marker, 4) == 0)
1073         {
1074             /* scan block */
1075             UChar *pEntry = _chm_find_in_PMGL(page_buf,
1076                                               h->block_len,
1077                                               objPath);
1078             if (pEntry == NULL)
1079             {
1080                 HeapFree(GetProcessHeap(), 0, page_buf);
1081                 return CHM_RESOLVE_FAILURE;
1082             }
1083
1084             /* parse entry and return */
1085             _chm_parse_PMGL_entry(&pEntry, ui);
1086             HeapFree(GetProcessHeap(), 0, page_buf);
1087             return CHM_RESOLVE_SUCCESS;
1088         }
1089
1090         /* else, if it is a branch node: */
1091         else if (memcmp(page_buf, _chm_pmgi_marker, 4) == 0)
1092             curPage = _chm_find_in_PMGI(page_buf, h->block_len, objPath);
1093
1094         /* else, we are confused.  give up. */
1095         else
1096         {
1097             HeapFree(GetProcessHeap(), 0, page_buf);
1098             return CHM_RESOLVE_FAILURE;
1099         }
1100     }
1101
1102     /* didn't find anything.  fail. */
1103     HeapFree(GetProcessHeap(), 0, page_buf);
1104     return CHM_RESOLVE_FAILURE;
1105 }
1106
1107 /*
1108  * utility methods for dealing with compressed data
1109  */
1110
1111 /* get the bounds of a compressed block.  return 0 on failure */
1112 static int _chm_get_cmpblock_bounds(struct chmFile *h,
1113                              UInt64 block,
1114                              UInt64 *start,
1115                              Int64 *len)
1116 {
1117     UChar buffer[8], *dummy;
1118     UInt32 remain;
1119
1120     /* for all but the last block, use the reset table */
1121     if (block < h->reset_table.block_count-1)
1122     {
1123         /* unpack the start address */
1124         dummy = buffer;
1125         remain = 8;
1126         if (_chm_fetch_bytes(h, buffer,
1127                              (UInt64)h->data_offset
1128                                 + (UInt64)h->rt_unit.start
1129                                 + (UInt64)h->reset_table.table_offset
1130                                 + (UInt64)block*8,
1131                              remain) != remain                            ||
1132             !_unmarshal_uint64(&dummy, &remain, start))
1133             return 0;
1134
1135         /* unpack the end address */
1136         dummy = buffer;
1137         remain = 8;
1138         if (_chm_fetch_bytes(h, buffer,
1139                          (UInt64)h->data_offset
1140                                 + (UInt64)h->rt_unit.start
1141                                 + (UInt64)h->reset_table.table_offset
1142                                 + (UInt64)block*8 + 8,
1143                          remain) != remain                                ||
1144             !_unmarshal_int64(&dummy, &remain, len))
1145             return 0;
1146     }
1147
1148     /* for the last block, use the span in addition to the reset table */
1149     else
1150     {
1151         /* unpack the start address */
1152         dummy = buffer;
1153         remain = 8;
1154         if (_chm_fetch_bytes(h, buffer,
1155                              (UInt64)h->data_offset
1156                                 + (UInt64)h->rt_unit.start
1157                                 + (UInt64)h->reset_table.table_offset
1158                                 + (UInt64)block*8,
1159                              remain) != remain                            ||
1160             !_unmarshal_uint64(&dummy, &remain, start))
1161             return 0;
1162
1163         *len = h->reset_table.compressed_len;
1164     }
1165
1166     /* compute the length and absolute start address */
1167     *len -= *start;
1168     *start += h->data_offset + h->cn_unit.start;
1169
1170     return 1;
1171 }
1172
1173 /* decompress the block.  must have lzx_mutex. */
1174 static Int64 _chm_decompress_block(struct chmFile *h,
1175                                    UInt64 block,
1176                                    UChar **ubuffer)
1177 {
1178     UChar *cbuffer = HeapAlloc( GetProcessHeap(), 0,
1179                               ((unsigned int)h->reset_table.block_len + 6144));
1180     UInt64 cmpStart;                                    /* compressed start  */
1181     Int64 cmpLen;                                       /* compressed len    */
1182     int indexSlot;                                      /* cache index slot  */
1183     UChar *lbuffer;                                     /* local buffer ptr  */
1184     UInt32 blockAlign = (UInt32)(block % h->reset_blkcount); /* reset intvl. aln. */
1185     UInt32 i;                                           /* local loop index  */
1186
1187     /* let the caching system pull its weight! */
1188     if (block - blockAlign <= h->lzx_last_block  &&
1189         block              >= h->lzx_last_block)
1190         blockAlign = (block - h->lzx_last_block);
1191
1192     /* check if we need previous blocks */
1193     if (blockAlign != 0)
1194     {
1195         /* fetch all required previous blocks since last reset */
1196         for (i = blockAlign; i > 0; i--)
1197         {
1198             UInt32 curBlockIdx = block - i;
1199
1200             /* check if we most recently decompressed the previous block */
1201             if (h->lzx_last_block != curBlockIdx)
1202             {
1203                 if ((curBlockIdx % h->reset_blkcount) == 0)
1204                 {
1205 #ifdef CHM_DEBUG
1206                     fprintf(stderr, "***RESET (1)***\n");
1207 #endif
1208                     LZXreset(h->lzx_state);
1209                 }
1210
1211                 indexSlot = (int)((curBlockIdx) % h->cache_num_blocks);
1212                 h->cache_block_indices[indexSlot] = curBlockIdx;
1213                 if (! h->cache_blocks[indexSlot])
1214                     h->cache_blocks[indexSlot] = malloc( (unsigned int)(h->reset_table.block_len));
1215                 lbuffer = h->cache_blocks[indexSlot];
1216
1217                 /* decompress the previous block */
1218 #ifdef CHM_DEBUG
1219                 fprintf(stderr, "Decompressing block #%4d (EXTRA)\n", curBlockIdx);
1220 #endif
1221                 if (!_chm_get_cmpblock_bounds(h, curBlockIdx, &cmpStart, &cmpLen) ||
1222                     _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen      ||
1223                     LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1224                                   (int)h->reset_table.block_len) != DECR_OK)
1225                 {
1226 #ifdef CHM_DEBUG
1227                     fprintf(stderr, "   (DECOMPRESS FAILED!)\n");
1228 #endif
1229                     HeapFree(GetProcessHeap(), 0, cbuffer);
1230                     return (Int64)0;
1231                 }
1232
1233                 h->lzx_last_block = (int)curBlockIdx;
1234             }
1235         }
1236     }
1237     else
1238     {
1239         if ((block % h->reset_blkcount) == 0)
1240         {
1241 #ifdef CHM_DEBUG
1242             fprintf(stderr, "***RESET (2)***\n");
1243 #endif
1244             LZXreset(h->lzx_state);
1245         }
1246     }
1247
1248     /* allocate slot in cache */
1249     indexSlot = (int)(block % h->cache_num_blocks);
1250     h->cache_block_indices[indexSlot] = block;
1251     if (! h->cache_blocks[indexSlot])
1252         h->cache_blocks[indexSlot] = malloc( ((unsigned int)h->reset_table.block_len));
1253     lbuffer = h->cache_blocks[indexSlot];
1254     *ubuffer = lbuffer;
1255
1256     /* decompress the block we actually want */
1257 #ifdef CHM_DEBUG
1258     fprintf(stderr, "Decompressing block #%4d (REAL )\n", block);
1259 #endif
1260     if (! _chm_get_cmpblock_bounds(h, block, &cmpStart, &cmpLen)          ||
1261         _chm_fetch_bytes(h, cbuffer, cmpStart, cmpLen) != cmpLen          ||
1262         LZXdecompress(h->lzx_state, cbuffer, lbuffer, (int)cmpLen,
1263                       (int)h->reset_table.block_len) != DECR_OK)
1264     {
1265 #ifdef CHM_DEBUG
1266         fprintf(stderr, "   (DECOMPRESS FAILED!)\n");
1267 #endif
1268         HeapFree(GetProcessHeap(), 0, cbuffer);
1269         return (Int64)0;
1270     }
1271     h->lzx_last_block = (int)block;
1272
1273     /* XXX: modify LZX routines to return the length of the data they
1274      * decompressed and return that instead, for an extra sanity check.
1275      */
1276     HeapFree(GetProcessHeap(), 0, cbuffer);
1277     return h->reset_table.block_len;
1278 }
1279
1280 /* grab a region from a compressed block */
1281 static Int64 _chm_decompress_region(struct chmFile *h,
1282                                     UChar *buf,
1283                                     UInt64 start,
1284                                     Int64 len)
1285 {
1286     UInt64 nBlock, nOffset;
1287     UInt64 nLen;
1288     UInt64 gotLen;
1289     UChar *ubuffer = NULL;
1290
1291         if (len <= 0)
1292                 return (Int64)0;
1293
1294     /* figure out what we need to read */
1295     nBlock = start / h->reset_table.block_len;
1296     nOffset = start % h->reset_table.block_len;
1297     nLen = len;
1298     if (nLen > (h->reset_table.block_len - nOffset))
1299         nLen = h->reset_table.block_len - nOffset;
1300
1301     /* if block is cached, return data from it. */
1302     CHM_ACQUIRE_LOCK(h->lzx_mutex);
1303     CHM_ACQUIRE_LOCK(h->cache_mutex);
1304     if (h->cache_block_indices[nBlock % h->cache_num_blocks] == nBlock    &&
1305         h->cache_blocks[nBlock % h->cache_num_blocks] != NULL)
1306     {
1307         memcpy(buf,
1308                h->cache_blocks[nBlock % h->cache_num_blocks] + nOffset,
1309                (unsigned int)nLen);
1310         CHM_RELEASE_LOCK(h->cache_mutex);
1311         CHM_RELEASE_LOCK(h->lzx_mutex);
1312         return nLen;
1313     }
1314     CHM_RELEASE_LOCK(h->cache_mutex);
1315
1316     /* data request not satisfied, so... start up the decompressor machine */
1317     if (! h->lzx_state)
1318     {
1319         int window_size = ffs(h->window_size) - 1;
1320         h->lzx_last_block = -1;
1321         h->lzx_state = LZXinit(window_size);
1322     }
1323
1324     /* decompress some data */
1325     gotLen = _chm_decompress_block(h, nBlock, &ubuffer);
1326     if (gotLen < nLen)
1327         nLen = gotLen;
1328     memcpy(buf, ubuffer+nOffset, (unsigned int)nLen);
1329     CHM_RELEASE_LOCK(h->lzx_mutex);
1330     return nLen;
1331 }
1332
1333 /* retrieve (part of) an object */
1334 LONGINT64 chm_retrieve_object(struct chmFile *h,
1335                                struct chmUnitInfo *ui,
1336                                unsigned char *buf,
1337                                LONGUINT64 addr,
1338                                LONGINT64 len)
1339 {
1340     /* must be valid file handle */
1341     if (h == NULL)
1342         return (Int64)0;
1343
1344     /* starting address must be in correct range */
1345     if (addr < 0  ||  addr >= ui->length)
1346         return (Int64)0;
1347
1348     /* clip length */
1349     if (addr + len > ui->length)
1350         len = ui->length - addr;
1351
1352     /* if the file is uncompressed, it's simple */
1353     if (ui->space == CHM_UNCOMPRESSED)
1354     {
1355         /* read data */
1356         return _chm_fetch_bytes(h,
1357                                 buf,
1358                                 (UInt64)h->data_offset + (UInt64)ui->start + (UInt64)addr,
1359                                 len);
1360     }
1361
1362     /* else if the file is compressed, it's a little trickier */
1363     else /* ui->space == CHM_COMPRESSED */
1364     {
1365         Int64 swath=0, total=0;
1366
1367         /* if compression is not enabled for this file... */
1368         if (! h->compression_enabled)
1369             return total;
1370
1371         do {
1372
1373             /* swill another mouthful */
1374             swath = _chm_decompress_region(h, buf, ui->start + addr, len);
1375
1376             /* if we didn't get any... */
1377             if (swath == 0)
1378                 return total;
1379
1380             /* update stats */
1381             total += swath;
1382             len -= swath;
1383             addr += swath;
1384             buf += swath;
1385
1386         } while (len != 0);
1387
1388         return total;
1389     }
1390 }
1391
1392 /* enumerate the objects in the .chm archive */
1393 int chm_enumerate(struct chmFile *h,
1394                   int what,
1395                   CHM_ENUMERATOR e,
1396                   void *context)
1397 {
1398     Int32 curPage;
1399
1400     /* buffer to hold whatever page we're looking at */
1401     UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, (unsigned int)h->block_len);
1402     struct chmPmglHeader header;
1403     UChar *end;
1404     UChar *cur;
1405     unsigned long lenRemain;
1406     UInt64 ui_path_len;
1407
1408     /* the current ui */
1409     struct chmUnitInfo ui;
1410     int flag;
1411
1412     /* starting page */
1413     curPage = h->index_head;
1414
1415     /* until we have either returned or given up */
1416     while (curPage != -1)
1417     {
1418
1419         /* try to fetch the index page */
1420         if (_chm_fetch_bytes(h,
1421                              page_buf,
1422                              (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1423                              h->block_len) != h->block_len)
1424         {
1425             HeapFree(GetProcessHeap(), 0, page_buf);
1426             return 0;
1427         }
1428
1429         /* figure out start and end for this page */
1430         cur = page_buf;
1431         lenRemain = _CHM_PMGL_LEN;
1432         if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1433         {
1434             HeapFree(GetProcessHeap(), 0, page_buf);
1435             return 0;
1436         }
1437         end = page_buf + h->block_len - (header.free_space);
1438
1439         /* loop over this page */
1440         while (cur < end)
1441         {
1442             if (! _chm_parse_PMGL_entry(&cur, &ui))
1443             {
1444                 HeapFree(GetProcessHeap(), 0, page_buf);
1445                 return 0;
1446             }
1447
1448             /* get the length of the path */
1449             ui_path_len = strlenW(ui.path)-1;
1450
1451             /* check for DIRS */
1452             if (ui.path[ui_path_len] == '/'  &&  !(what & CHM_ENUMERATE_DIRS))
1453                 continue;
1454
1455             /* check for FILES */
1456             if (ui.path[ui_path_len] != '/'  &&  !(what & CHM_ENUMERATE_FILES))
1457                 continue;
1458
1459             /* check for NORMAL vs. META */
1460             if (ui.path[0] == '/')
1461             {
1462
1463                 /* check for NORMAL vs. SPECIAL */
1464                 if (ui.path[1] == '#'  ||  ui.path[1] == '$')
1465                     flag = CHM_ENUMERATE_SPECIAL;
1466                 else
1467                     flag = CHM_ENUMERATE_NORMAL;
1468             }
1469             else
1470                 flag = CHM_ENUMERATE_META;
1471             if (! (what & flag))
1472                 continue;
1473
1474             /* call the enumerator */
1475             {
1476                 int status = (*e)(h, &ui, context);
1477                 switch (status)
1478                 {
1479                     case CHM_ENUMERATOR_FAILURE:
1480                         HeapFree(GetProcessHeap(), 0, page_buf);
1481                         return 0;
1482                     case CHM_ENUMERATOR_CONTINUE:
1483                         break;
1484                     case CHM_ENUMERATOR_SUCCESS:
1485                         HeapFree(GetProcessHeap(), 0, page_buf);
1486                         return 1;
1487                     default:
1488                         break;
1489                 }
1490             }
1491         }
1492
1493         /* advance to next page */
1494         curPage = header.block_next;
1495     }
1496
1497     HeapFree(GetProcessHeap(), 0, page_buf);
1498     return 1;
1499 }
1500
1501 int chm_enumerate_dir(struct chmFile *h,
1502                       const WCHAR *prefix,
1503                       int what,
1504                       CHM_ENUMERATOR e,
1505                       void *context)
1506 {
1507     /*
1508      * XXX: do this efficiently (i.e. using the tree index)
1509      */
1510
1511     Int32 curPage;
1512
1513     /* buffer to hold whatever page we're looking at */
1514     UChar *page_buf = HeapAlloc(GetProcessHeap(), 0, (unsigned int)h->block_len);
1515     struct chmPmglHeader header;
1516     UChar *end;
1517     UChar *cur;
1518     unsigned long lenRemain;
1519
1520     /* set to 1 once we've started */
1521     int it_has_begun=0;
1522
1523     /* the current ui */
1524     struct chmUnitInfo ui;
1525     int flag;
1526     UInt64 ui_path_len;
1527
1528     /* the length of the prefix */
1529     WCHAR prefixRectified[CHM_MAX_PATHLEN+1];
1530     int prefixLen;
1531     WCHAR lastPath[CHM_MAX_PATHLEN];
1532     int lastPathLen;
1533
1534     /* starting page */
1535     curPage = h->index_head;
1536
1537     /* initialize pathname state */
1538     lstrcpynW(prefixRectified, prefix, CHM_MAX_PATHLEN);
1539     prefixLen = strlenW(prefixRectified);
1540     if (prefixLen != 0)
1541     {
1542         if (prefixRectified[prefixLen-1] != '/')
1543         {
1544             prefixRectified[prefixLen] = '/';
1545             prefixRectified[prefixLen+1] = '\0';
1546             ++prefixLen;
1547         }
1548     }
1549     lastPath[0] = '\0';
1550     lastPathLen = -1;
1551
1552     /* until we have either returned or given up */
1553     while (curPage != -1)
1554     {
1555
1556         /* try to fetch the index page */
1557         if (_chm_fetch_bytes(h,
1558                              page_buf,
1559                              (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
1560                              h->block_len) != h->block_len)
1561         {
1562             HeapFree(GetProcessHeap(), 0, page_buf);
1563             return 0;
1564         }
1565
1566         /* figure out start and end for this page */
1567         cur = page_buf;
1568         lenRemain = _CHM_PMGL_LEN;
1569         if (! _unmarshal_pmgl_header(&cur, &lenRemain, &header))
1570         {
1571             HeapFree(GetProcessHeap(), 0, page_buf);
1572             return 0;
1573         }
1574         end = page_buf + h->block_len - (header.free_space);
1575
1576         /* loop over this page */
1577         while (cur < end)
1578         {
1579             if (! _chm_parse_PMGL_entry(&cur, &ui))
1580             {
1581                 HeapFree(GetProcessHeap(), 0, page_buf);
1582                 return 0;
1583             }
1584
1585             /* check if we should start */
1586             if (! it_has_begun)
1587             {
1588                 if (ui.length == 0  &&  strncmpiW(ui.path, prefixRectified, prefixLen) == 0)
1589                     it_has_begun = 1;
1590                 else
1591                     continue;
1592
1593                 if (ui.path[prefixLen] == '\0')
1594                     continue;
1595             }
1596
1597             /* check if we should stop */
1598             else
1599             {
1600                 if (strncmpiW(ui.path, prefixRectified, prefixLen) != 0)
1601                 {
1602                     HeapFree(GetProcessHeap(), 0, page_buf);
1603                     return 1;
1604                 }
1605             }
1606
1607             /* check if we should include this path */
1608             if (lastPathLen != -1)
1609             {
1610                 if (strncmpiW(ui.path, lastPath, lastPathLen) == 0)
1611                     continue;
1612             }
1613             strcpyW(lastPath, ui.path);
1614             lastPathLen = strlenW(lastPath);
1615
1616             /* get the length of the path */
1617             ui_path_len = strlenW(ui.path)-1;
1618
1619             /* check for DIRS */
1620             if (ui.path[ui_path_len] == '/'  &&  !(what & CHM_ENUMERATE_DIRS))
1621                 continue;
1622
1623             /* check for FILES */
1624             if (ui.path[ui_path_len] != '/'  &&  !(what & CHM_ENUMERATE_FILES))
1625                 continue;
1626
1627             /* check for NORMAL vs. META */
1628             if (ui.path[0] == '/')
1629             {
1630
1631                 /* check for NORMAL vs. SPECIAL */
1632                 if (ui.path[1] == '#'  ||  ui.path[1] == '$')
1633                     flag = CHM_ENUMERATE_SPECIAL;
1634                 else
1635                     flag = CHM_ENUMERATE_NORMAL;
1636             }
1637             else
1638                 flag = CHM_ENUMERATE_META;
1639             if (! (what & flag))
1640                 continue;
1641
1642             /* call the enumerator */
1643             {
1644                 int status = (*e)(h, &ui, context);
1645                 switch (status)
1646                 {
1647                     case CHM_ENUMERATOR_FAILURE:
1648                         HeapFree(GetProcessHeap(), 0, page_buf);
1649                         return 0;
1650                     case CHM_ENUMERATOR_CONTINUE:
1651                         break;
1652                     case CHM_ENUMERATOR_SUCCESS:
1653                         HeapFree(GetProcessHeap(), 0, page_buf);
1654                         return 1;
1655                     default:
1656                         break;
1657                 }
1658             }
1659         }
1660
1661         /* advance to next page */
1662         curPage = header.block_next;
1663     }
1664
1665     HeapFree(GetProcessHeap(), 0, page_buf);
1666     return 1;
1667 }