dlls/msi/string.c

   1 /*
   2  * String Table Functions
   3  *
   4  * Copyright 2002-2004, Mike McCormack for CodeWeavers
   5  * Copyright 2007 Robert Shearman for CodeWeavers
   6  * Copyright 2010 Hans Leidekker for CodeWeavers
   7  *
   8  * This library is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU Lesser General Public
  10  * License as published by the Free Software Foundation; either
  11  * version 2.1 of the License, or (at your option) any later version.
  12  *
  13  * This library is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public
  19  * License along with this library; if not, write to the Free Software
  20  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
  21  */
  22
  23 #define COBJMACROS
  24
  25 #include <stdarg.h>
  26 #include <assert.h>
  27
  28 #include "windef.h"
  29 #include "winbase.h"
  30 #include "winerror.h"
  31 #include "wine/debug.h"
  32 #include "wine/unicode.h"
  33 #include "msi.h"
  34 #include "msiquery.h"
  35 #include "objbase.h"
  36 #include "objidl.h"
  37 #include "msipriv.h"
  38 #include "winnls.h"
  39
  40 #include "query.h"
  41
  42 WINE_DEFAULT_DEBUG_CHANNEL(msidb);
  43
  44 struct msistring
  45 {
  46     USHORT persistent_refcount;
  47     USHORT nonpersistent_refcount;
  48     WCHAR *data;
  49     int    len;
  50 };
  51
  52 struct string_table
  53 {
  54     UINT maxcount;         /* the number of strings */
  55     UINT freeslot;
  56     UINT codepage;
  57     UINT sortcount;
  58     struct msistring *strings; /* an array of strings */
  59     UINT *sorted;              /* index */
  60 };
  61
  62 static BOOL validate_codepage( UINT codepage )
  63 {
  64     if (codepage != CP_ACP && !IsValidCodePage( codepage ))
  65     {
  66         WARN("invalid codepage %u\n", codepage);
  67         return FALSE;
  68     }
  69     return TRUE;
  70 }
  71
  72 static string_table *init_stringtable( int entries, UINT codepage )
  73 {
  74     string_table *st;
  75
  76     if (!validate_codepage( codepage ))
  77         return NULL;
  78
  79     st = msi_alloc( sizeof (string_table) );
  80     if( !st )
  81         return NULL;
  82     if( entries < 1 )
  83         entries = 1;
  84
  85     st->strings = msi_alloc_zero( sizeof(struct msistring) * entries );
  86     if( !st->strings )
  87     {
  88         msi_free( st );
  89         return NULL;
  90     }
  91
  92     st->sorted = msi_alloc( sizeof (UINT) * entries );
  93     if( !st->sorted )
  94     {
  95         msi_free( st->strings );
  96         msi_free( st );
  97         return NULL;
  98     }
  99
 100     st->maxcount = entries;
 101     st->freeslot = 1;
 102     st->codepage = codepage;
 103     st->sortcount = 0;
 104
 105     return st;
 106 }
 107
 108 VOID msi_destroy_stringtable( string_table *st )
 109 {
 110     UINT i;
 111
 112     for( i=0; i<st->maxcount; i++ )
 113     {
 114         if( st->strings[i].persistent_refcount ||
 115             st->strings[i].nonpersistent_refcount )
 116             msi_free( st->strings[i].data );
 117     }
 118     msi_free( st->strings );
 119     msi_free( st->sorted );
 120     msi_free( st );
 121 }
 122
 123 static int st_find_free_entry( string_table *st )
 124 {
 125     UINT i, sz, *s;
 126     struct msistring *p;
 127
 128     TRACE("%p\n", st);
 129
 130     if( st->freeslot )
 131     {
 132         for( i = st->freeslot; i < st->maxcount; i++ )
 133             if( !st->strings[i].persistent_refcount &&
 134                 !st->strings[i].nonpersistent_refcount )
 135                 return i;
 136     }
 137     for( i = 1; i < st->maxcount; i++ )
 138         if( !st->strings[i].persistent_refcount &&
 139             !st->strings[i].nonpersistent_refcount )
 140             return i;
 141
 142     /* dynamically resize */
 143     sz = st->maxcount + 1 + st->maxcount/2;
 144     p = msi_realloc_zero( st->strings, sz * sizeof(struct msistring) );
 145     if( !p )
 146         return -1;
 147
 148     s = msi_realloc( st->sorted, sz*sizeof(UINT) );
 149     if( !s )
 150     {
 151         msi_free( p );
 152         return -1;
 153     }
 154
 155     st->strings = p;
 156     st->sorted = s;
 157
 158     st->freeslot = st->maxcount;
 159     st->maxcount = sz;
 160     if( st->strings[st->freeslot].persistent_refcount ||
 161         st->strings[st->freeslot].nonpersistent_refcount )
 162         ERR("oops. expected freeslot to be free...\n");
 163     return st->freeslot;
 164 }
 165
 166 static inline int cmp_string( const WCHAR *str1, int len1, const WCHAR *str2, int len2 )
 167 {
 168     if (len1 < len2) return -1;
 169     else if (len1 > len2) return 1;
 170     while (len1)
 171     {
 172         if (*str1 == *str2) { str1++; str2++; }
 173         else return *str1 - *str2;
 174         len1--;
 175     }
 176     return 0;
 177 }
 178
 179 static int find_insert_index( const string_table *st, UINT string_id )
 180 {
 181     int i, c, low = 0, high = st->sortcount - 1;
 182
 183     while (low <= high)
 184     {
 185         i = (low + high) / 2;
 186         c = cmp_string( st->strings[string_id].data, st->strings[string_id].len,
 187                         st->strings[st->sorted[i]].data, st->strings[st->sorted[i]].len );
 188         if (c < 0)
 189             high = i - 1;
 190         else if (c > 0)
 191             low = i + 1;
 192         else
 193             return -1; /* already exists */
 194     }
 195     return high + 1;
 196 }
 197
 198 static void insert_string_sorted( string_table *st, UINT string_id )
 199 {
 200     int i;
 201
 202     i = find_insert_index( st, string_id );
 203     if (i == -1)
 204         return;
 205
 206     memmove( &st->sorted[i] + 1, &st->sorted[i], (st->sortcount - i) * sizeof(UINT) );
 207     st->sorted[i] = string_id;
 208     st->sortcount++;
 209 }
 210
 211 static void set_st_entry( string_table *st, UINT n, WCHAR *str, int len, USHORT refcount,
 212                           enum StringPersistence persistence )
 213 {
 214     if (persistence == StringPersistent)
 215     {
 216         st->strings[n].persistent_refcount = refcount;
 217         st->strings[n].nonpersistent_refcount = 0;
 218     }
 219     else
 220     {
 221         st->strings[n].persistent_refcount = 0;
 222         st->strings[n].nonpersistent_refcount = refcount;
 223     }
 224
 225     st->strings[n].data = str;
 226     st->strings[n].len  = len;
 227
 228     insert_string_sorted( st, n );
 229
 230     if( n < st->maxcount )
 231         st->freeslot = n + 1;
 232 }
 233
 234 static UINT msi_string2idA( const string_table *st, LPCSTR buffer, UINT *id )
 235 {
 236     DWORD sz;
 237     UINT r = ERROR_INVALID_PARAMETER;
 238     LPWSTR str;
 239
 240     TRACE("Finding string %s in string table\n", debugstr_a(buffer) );
 241
 242     if( buffer[0] == 0 )
 243     {
 244         *id = 0;
 245         return ERROR_SUCCESS;
 246     }
 247
 248     sz = MultiByteToWideChar( st->codepage, 0, buffer, -1, NULL, 0 );
 249     if( sz <= 0 )
 250         return r;
 251     str = msi_alloc( sz*sizeof(WCHAR) );
 252     if( !str )
 253         return ERROR_NOT_ENOUGH_MEMORY;
 254     MultiByteToWideChar( st->codepage, 0, buffer, -1, str, sz );
 255
 256     r = msi_string2id( st, str, sz - 1, id );
 257     msi_free( str );
 258     return r;
 259 }
 260
 261 static int msi_addstring( string_table *st, UINT n, const CHAR *data, int len, USHORT refcount, enum StringPersistence persistence )
 262 {
 263     LPWSTR str;
 264     int sz;
 265
 266     if( !data )
 267         return 0;
 268     if( !data[0] )
 269         return 0;
 270     if( n > 0 )
 271     {
 272         if( st->strings[n].persistent_refcount ||
 273             st->strings[n].nonpersistent_refcount )
 274             return -1;
 275     }
 276     else
 277     {
 278         if( ERROR_SUCCESS == msi_string2idA( st, data, &n ) )
 279         {
 280             if (persistence == StringPersistent)
 281                 st->strings[n].persistent_refcount += refcount;
 282             else
 283                 st->strings[n].nonpersistent_refcount += refcount;
 284             return n;
 285         }
 286         n = st_find_free_entry( st );
 287         if( n == -1 )
 288             return -1;
 289     }
 290
 291     if( n < 1 )
 292     {
 293         ERR("invalid index adding %s (%d)\n", debugstr_a( data ), n );
 294         return -1;
 295     }
 296
 297     /* allocate a new string */
 298     if( len < 0 )
 299         len = strlen(data);
 300     sz = MultiByteToWideChar( st->codepage, 0, data, len, NULL, 0 );
 301     str = msi_alloc( (sz+1)*sizeof(WCHAR) );
 302     if( !str )
 303         return -1;
 304     MultiByteToWideChar( st->codepage, 0, data, len, str, sz );
 305     str[sz] = 0;
 306
 307     set_st_entry( st, n, str, sz, refcount, persistence );
 308
 309     return n;
 310 }
 311
 312 int msi_addstringW( string_table *st, const WCHAR *data, int len, USHORT refcount, enum StringPersistence persistence )
 313 {
 314     UINT n;
 315     LPWSTR str;
 316
 317     if( !data )
 318         return 0;
 319     if( !data[0] && !len )
 320         return 0;
 321
 322     if (msi_string2id( st, data, len, &n) == ERROR_SUCCESS )
 323     {
 324         if (persistence == StringPersistent)
 325             st->strings[n].persistent_refcount += refcount;
 326         else
 327             st->strings[n].nonpersistent_refcount += refcount;
 328         return n;
 329     }
 330
 331     n = st_find_free_entry( st );
 332     if( n == -1 )
 333         return -1;
 334
 335     /* allocate a new string */
 336     if(len<0)
 337         len = strlenW(data);
 338     TRACE( "%s, n = %d len = %d\n", debugstr_wn(data, len), n, len );
 339
 340     str = msi_alloc( (len+1)*sizeof(WCHAR) );
 341     if( !str )
 342         return -1;
 343     memcpy( str, data, len*sizeof(WCHAR) );
 344     str[len] = 0;
 345
 346     set_st_entry( st, n, str, len, refcount, persistence );
 347
 348     return n;
 349 }
 350
 351 /* find the string identified by an id - return null if there's none */
 352 const WCHAR *msi_string_lookup( const string_table *st, UINT id, int *len )
 353 {
 354     if( id == 0 )
 355     {
 356         if (len) *len = 0;
 357         return szEmpty;
 358     }
 359     if( id >= st->maxcount )
 360         return NULL;
 361
 362     if( id && !st->strings[id].persistent_refcount && !st->strings[id].nonpersistent_refcount)
 363         return NULL;
 364
 365     if (len) *len = st->strings[id].len;
 366
 367     return st->strings[id].data;
 368 }
 369
 370 /*
 371  *  msi_id2stringA
 372  *
 373  *  [in] st         - pointer to the string table
 374  *  [in] id         - id of the string to retrieve
 375  *  [out] buffer    - destination of the UTF8 string
 376  *  [in/out] sz     - number of bytes available in the buffer on input
 377  *                    number of bytes used on output
 378  *
 379  *  Returned string is not nul terminated.
 380  */
 381 static UINT msi_id2stringA( const string_table *st, UINT id, LPSTR buffer, UINT *sz )
 382 {
 383     int len, lenW;
 384     const WCHAR *str;
 385
 386     TRACE("Finding string %d of %d\n", id, st->maxcount);
 387
 388     str = msi_string_lookup( st, id, &lenW );
 389     if( !str )
 390         return ERROR_FUNCTION_FAILED;
 391
 392     len = WideCharToMultiByte( st->codepage, 0, str, lenW, NULL, 0, NULL, NULL );
 393     if( *sz < len )
 394     {
 395         *sz = len;
 396         return ERROR_MORE_DATA;
 397     }
 398     *sz = WideCharToMultiByte( st->codepage, 0, str, lenW, buffer, *sz, NULL, NULL );
 399     return ERROR_SUCCESS;
 400 }
 401
 402 /*
 403  *  msi_string2id
 404  *
 405  *  [in] st         - pointer to the string table
 406  *  [in] str        - string to find in the string table
 407  *  [out] id        - id of the string, if found
 408  */
 409 UINT msi_string2id( const string_table *st, const WCHAR *str, int len, UINT *id )
 410 {
 411     int i, c, low = 0, high = st->sortcount - 1;
 412
 413     if (len < 0) len = strlenW( str );
 414
 415     while (low <= high)
 416     {
 417         i = (low + high) / 2;
 418         c = cmp_string( str, len, st->strings[st->sorted[i]].data, st->strings[st->sorted[i]].len );
 419
 420         if (c < 0)
 421             high = i - 1;
 422         else if (c > 0)
 423             low = i + 1;
 424         else
 425         {
 426             *id = st->sorted[i];
 427             return ERROR_SUCCESS;
 428         }
 429     }
 430     return ERROR_INVALID_PARAMETER;
 431 }
 432
 433 static void string_totalsize( const string_table *st, UINT *datasize, UINT *poolsize )
 434 {
 435     UINT i, len, holesize;
 436
 437     if( st->strings[0].data || st->strings[0].persistent_refcount || st->strings[0].nonpersistent_refcount)
 438         ERR("oops. element 0 has a string\n");
 439
 440     *poolsize = 4;
 441     *datasize = 0;
 442     holesize = 0;
 443     for( i=1; i<st->maxcount; i++ )
 444     {
 445         if( !st->strings[i].persistent_refcount )
 446         {
 447             TRACE("[%u] nonpersistent = %s\n", i, debugstr_wn(st->strings[i].data, st->strings[i].len));
 448             (*poolsize) += 4;
 449         }
 450         else if( st->strings[i].data )
 451         {
 452             TRACE("[%u] = %s\n", i, debugstr_wn(st->strings[i].data, st->strings[i].len));
 453             len = WideCharToMultiByte( st->codepage, 0, st->strings[i].data, st->strings[i].len + 1,
 454                                        NULL, 0, NULL, NULL);
 455             if( len )
 456                 len--;
 457             (*datasize) += len;
 458             if (len>0xffff)
 459                 (*poolsize) += 4;
 460             (*poolsize) += holesize + 4;
 461             holesize = 0;
 462         }
 463         else
 464             holesize += 4;
 465     }
 466     TRACE("data %u pool %u codepage %x\n", *datasize, *poolsize, st->codepage );
 467 }
 468
 469 HRESULT msi_init_string_table( IStorage *stg )
 470 {
 471     USHORT zero[2] = { 0, 0 };
 472     UINT ret;
 473
 474     /* create the StringPool stream... add the zero string to it*/
 475     ret = write_stream_data(stg, szStringPool, zero, sizeof zero, TRUE);
 476     if (ret != ERROR_SUCCESS)
 477         return E_FAIL;
 478
 479     /* create the StringData stream... make it zero length */
 480     ret = write_stream_data(stg, szStringData, NULL, 0, TRUE);
 481     if (ret != ERROR_SUCCESS)
 482         return E_FAIL;
 483
 484     return S_OK;
 485 }
 486
 487 string_table *msi_load_string_table( IStorage *stg, UINT *bytes_per_strref )
 488 {
 489     string_table *st = NULL;
 490     CHAR *data = NULL;
 491     USHORT *pool = NULL;
 492     UINT r, datasize = 0, poolsize = 0, codepage;
 493     DWORD i, count, offset, len, n, refs;
 494
 495     r = read_stream_data( stg, szStringPool, TRUE, (BYTE **)&pool, &poolsize );
 496     if( r != ERROR_SUCCESS)
 497         goto end;
 498     r = read_stream_data( stg, szStringData, TRUE, (BYTE **)&data, &datasize );
 499     if( r != ERROR_SUCCESS)
 500         goto end;
 501
 502     if ( (poolsize > 4) && (pool[1] & 0x8000) )
 503         *bytes_per_strref = LONG_STR_BYTES;
 504     else
 505         *bytes_per_strref = sizeof(USHORT);
 506
 507     count = poolsize/4;
 508     if( poolsize > 4 )
 509         codepage = pool[0] | ( (pool[1] & ~0x8000) << 16 );
 510     else
 511         codepage = CP_ACP;
 512     st = init_stringtable( count, codepage );
 513     if (!st)
 514         goto end;
 515
 516     offset = 0;
 517     n = 1;
 518     i = 1;
 519     while( i<count )
 520     {
 521         /* the string reference count is always the second word */
 522         refs = pool[i*2+1];
 523
 524         /* empty entries have two zeros, still have a string id */
 525         if (pool[i*2] == 0 && refs == 0)
 526         {
 527             i++;
 528             n++;
 529             continue;
 530         }
 531
 532         /*
 533          * If a string is over 64k, the previous string entry is made null
 534          * and its the high word of the length is inserted in the null string's
 535          * reference count field.
 536          */
 537         if( pool[i*2] == 0)
 538         {
 539             len = (pool[i*2+3] << 16) + pool[i*2+2];
 540             i += 2;
 541         }
 542         else
 543         {
 544             len = pool[i*2];
 545             i += 1;
 546         }
 547
 548         if ( (offset + len) > datasize )
 549         {
 550             ERR("string table corrupt?\n");
 551             break;
 552         }
 553
 554         r = msi_addstring( st, n, data+offset, len, refs, StringPersistent );
 555         if( r != n )
 556             ERR("Failed to add string %d\n", n );
 557         n++;
 558         offset += len;
 559     }
 560
 561     if ( datasize != offset )
 562         ERR("string table load failed! (%08x != %08x), please report\n", datasize, offset );
 563
 564     TRACE("Loaded %d strings\n", count);
 565
 566 end:
 567     msi_free( pool );
 568     msi_free( data );
 569
 570     return st;
 571 }
 572
 573 UINT msi_save_string_table( const string_table *st, IStorage *storage, UINT *bytes_per_strref )
 574 {
 575     UINT i, datasize = 0, poolsize = 0, sz, used, r, codepage, n;
 576     UINT ret = ERROR_FUNCTION_FAILED;
 577     CHAR *data = NULL;
 578     USHORT *pool = NULL;
 579
 580     TRACE("\n");
 581
 582     /* construct the new table in memory first */
 583     string_totalsize( st, &datasize, &poolsize );
 584
 585     TRACE("%u %u %u\n", st->maxcount, datasize, poolsize );
 586
 587     pool = msi_alloc( poolsize );
 588     if( ! pool )
 589     {
 590         WARN("Failed to alloc pool %d bytes\n", poolsize );
 591         goto err;
 592     }
 593     data = msi_alloc( datasize );
 594     if( ! data )
 595     {
 596         WARN("Failed to alloc data %d bytes\n", datasize );
 597         goto err;
 598     }
 599
 600     used = 0;
 601     codepage = st->codepage;
 602     pool[0] = codepage & 0xffff;
 603     pool[1] = codepage >> 16;
 604     if (st->maxcount > 0xffff)
 605     {
 606         pool[1] |= 0x8000;
 607         *bytes_per_strref = LONG_STR_BYTES;
 608     }
 609     else
 610         *bytes_per_strref = sizeof(USHORT);
 611
 612     n = 1;
 613     for( i=1; i<st->maxcount; i++ )
 614     {
 615         if( !st->strings[i].persistent_refcount )
 616         {
 617             pool[ n*2 ] = 0;
 618             pool[ n*2 + 1] = 0;
 619             n++;
 620             continue;
 621         }
 622
 623         sz = datasize - used;
 624         r = msi_id2stringA( st, i, data+used, &sz );
 625         if( r != ERROR_SUCCESS )
 626         {
 627             ERR("failed to fetch string\n");
 628             sz = 0;
 629         }
 630
 631         if (sz)
 632             pool[ n*2 + 1 ] = st->strings[i].persistent_refcount;
 633         else
 634             pool[ n*2 + 1 ] = 0;
 635         if (sz < 0x10000)
 636         {
 637             pool[ n*2 ] = sz;
 638             n++;
 639         }
 640         else
 641         {
 642             pool[ n*2 ] = 0;
 643             pool[ n*2 + 2 ] = sz&0xffff;
 644             pool[ n*2 + 3 ] = (sz>>16);
 645             n += 2;
 646         }
 647         used += sz;
 648         if( used > datasize  )
 649         {
 650             ERR("oops overran %d >= %d\n", used, datasize);
 651             goto err;
 652         }
 653     }
 654
 655     if( used != datasize )
 656     {
 657         ERR("oops used %d != datasize %d\n", used, datasize);
 658         goto err;
 659     }
 660
 661     /* write the streams */
 662     r = write_stream_data( storage, szStringData, data, datasize, TRUE );
 663     TRACE("Wrote StringData r=%08x\n", r);
 664     if( r )
 665         goto err;
 666     r = write_stream_data( storage, szStringPool, pool, poolsize, TRUE );
 667     TRACE("Wrote StringPool r=%08x\n", r);
 668     if( r )
 669         goto err;
 670
 671     ret = ERROR_SUCCESS;
 672
 673 err:
 674     msi_free( data );
 675     msi_free( pool );
 676
 677     return ret;
 678 }
 679
 680 UINT msi_get_string_table_codepage( const string_table *st )
 681 {
 682     return st->codepage;
 683 }
 684
 685 UINT msi_set_string_table_codepage( string_table *st, UINT codepage )
 686 {
 687     if (validate_codepage( codepage ))
 688     {
 689         st->codepage = codepage;
 690         return ERROR_SUCCESS;
 691     }
 692     return ERROR_FUNCTION_FAILED;
 693 }