icuSources/common/uresdata.c

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 1999-2014, International Business Machines Corporation
   4 *               and others. All Rights Reserved.
   5 *******************************************************************************
   6 *   file name:  uresdata.c
   7 *   encoding:   US-ASCII
   8 *   tab size:   8 (not used)
   9 *   indentation:4
  10 *
  11 *   created on: 1999dec08
  12 *   created by: Markus W. Scherer
  13 * Modification History:
  14 *
  15 *   Date        Name        Description
  16 *   06/20/2000  helena      OS/400 port changes; mostly typecast.
  17 *   06/24/02    weiv        Added support for resource sharing
  18 */
  19
  20 #include "unicode/utypes.h"
  21 #include "unicode/udata.h"
  22 #include "unicode/ustring.h"
  23 #include "unicode/utf16.h"
  24 #include "cmemory.h"
  25 #include "cstring.h"
  26 #include "uarrsort.h"
  27 #include "udataswp.h"
  28 #include "ucol_swp.h"
  29 #include "uinvchar.h"
  30 #include "uresdata.h"
  31 #include "uresimp.h"
  32 #include "uassert.h"
  33
  34 #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
  35
  36 /*
  37  * Resource access helpers
  38  */
  39
  40 /* get a const char* pointer to the key with the keyOffset byte offset from pRoot */
  41 #define RES_GET_KEY16(pResData, keyOffset) \
  42     ((keyOffset)<(pResData)->localKeyLimit ? \
  43         (const char *)(pResData)->pRoot+(keyOffset) : \
  44         (pResData)->poolBundleKeys+(keyOffset)-(pResData)->localKeyLimit)
  45
  46 #define RES_GET_KEY32(pResData, keyOffset) \
  47     ((keyOffset)>=0 ? \
  48         (const char *)(pResData)->pRoot+(keyOffset) : \
  49         (pResData)->poolBundleKeys+((keyOffset)&0x7fffffff))
  50
  51 #define URESDATA_ITEM_NOT_FOUND -1
  52
  53 /* empty resources, returned when the resource offset is 0 */
  54 static const uint16_t gEmpty16=0;
  55
  56 static const struct {
  57     int32_t length;
  58     int32_t res;
  59 } gEmpty32={ 0, 0 };
  60
  61 static const struct {
  62     int32_t length;
  63     UChar nul;
  64     UChar pad;
  65 } gEmptyString={ 0, 0, 0 };
  66
  67 /*
  68  * All the type-access functions assume that
  69  * the resource is of the expected type.
  70  */
  71
  72 static int32_t
  73 _res_findTableItem(const ResourceData *pResData, const uint16_t *keyOffsets, int32_t length,
  74                    const char *key, const char **realKey) {
  75     const char *tableKey;
  76     int32_t mid, start, limit;
  77     int result;
  78
  79     /* do a binary search for the key */
  80     start=0;
  81     limit=length;
  82     while(start<limit) {
  83         mid = (start + limit) / 2;
  84         tableKey = RES_GET_KEY16(pResData, keyOffsets[mid]);
  85         if (pResData->useNativeStrcmp) {
  86             result = uprv_strcmp(key, tableKey);
  87         } else {
  88             result = uprv_compareInvCharsAsAscii(key, tableKey);
  89         }
  90         if (result < 0) {
  91             limit = mid;
  92         } else if (result > 0) {
  93             start = mid + 1;
  94         } else {
  95             /* We found it! */
  96             *realKey=tableKey;
  97             return mid;
  98         }
  99     }
 100     return URESDATA_ITEM_NOT_FOUND;  /* not found or table is empty. */
 101 }
 102
 103 static int32_t
 104 _res_findTable32Item(const ResourceData *pResData, const int32_t *keyOffsets, int32_t length,
 105                      const char *key, const char **realKey) {
 106     const char *tableKey;
 107     int32_t mid, start, limit;
 108     int result;
 109
 110     /* do a binary search for the key */
 111     start=0;
 112     limit=length;
 113     while(start<limit) {
 114         mid = (start + limit) / 2;
 115         tableKey = RES_GET_KEY32(pResData, keyOffsets[mid]);
 116         if (pResData->useNativeStrcmp) {
 117             result = uprv_strcmp(key, tableKey);
 118         } else {
 119             result = uprv_compareInvCharsAsAscii(key, tableKey);
 120         }
 121         if (result < 0) {
 122             limit = mid;
 123         } else if (result > 0) {
 124             start = mid + 1;
 125         } else {
 126             /* We found it! */
 127             *realKey=tableKey;
 128             return mid;
 129         }
 130     }
 131     return URESDATA_ITEM_NOT_FOUND;  /* not found or table is empty. */
 132 }
 133
 134 /* helper for res_load() ---------------------------------------------------- */
 135
 136 static UBool U_CALLCONV
 137 isAcceptable(void *context,
 138              const char *type, const char *name,
 139              const UDataInfo *pInfo) {
 140     uprv_memcpy(context, pInfo->formatVersion, 4);
 141     return (UBool)(
 142         pInfo->size>=20 &&
 143         pInfo->isBigEndian==U_IS_BIG_ENDIAN &&
 144         pInfo->charsetFamily==U_CHARSET_FAMILY &&
 145         pInfo->sizeofUChar==U_SIZEOF_UCHAR &&
 146         pInfo->dataFormat[0]==0x52 &&   /* dataFormat="ResB" */
 147         pInfo->dataFormat[1]==0x65 &&
 148         pInfo->dataFormat[2]==0x73 &&
 149         pInfo->dataFormat[3]==0x42 &&
 150         (pInfo->formatVersion[0]==1 || pInfo->formatVersion[0]==2));
 151 }
 152
 153 /* semi-public functions ---------------------------------------------------- */
 154
 155 static void
 156 res_init(ResourceData *pResData,
 157          UVersionInfo formatVersion, const void *inBytes, int32_t length,
 158          UErrorCode *errorCode) {
 159     UResType rootType;
 160
 161     /* get the root resource */
 162     pResData->pRoot=(const int32_t *)inBytes;
 163     pResData->rootRes=(Resource)*pResData->pRoot;
 164     pResData->p16BitUnits=&gEmpty16;
 165
 166     /* formatVersion 1.1 must have a root item and at least 5 indexes */
 167     if(length>=0 && (length/4)<((formatVersion[0]==1 && formatVersion[1]==0) ? 1 : 1+5)) {
 168         *errorCode=U_INVALID_FORMAT_ERROR;
 169         res_unload(pResData);
 170         return;
 171     }
 172
 173     /* currently, we accept only resources that have a Table as their roots */
 174     rootType=(UResType)RES_GET_TYPE(pResData->rootRes);
 175     if(!URES_IS_TABLE(rootType)) {
 176         *errorCode=U_INVALID_FORMAT_ERROR;
 177         res_unload(pResData);
 178         return;
 179     }
 180
 181     if(formatVersion[0]==1 && formatVersion[1]==0) {
 182         pResData->localKeyLimit=0x10000;  /* greater than any 16-bit key string offset */
 183     } else {
 184         /* bundles with formatVersion 1.1 and later contain an indexes[] array */
 185         const int32_t *indexes=pResData->pRoot+1;
 186         int32_t indexLength=indexes[URES_INDEX_LENGTH]&0xff;
 187         if(indexLength<=URES_INDEX_MAX_TABLE_LENGTH) {
 188             *errorCode=U_INVALID_FORMAT_ERROR;
 189             res_unload(pResData);
 190             return;
 191         }
 192         if( length>=0 &&
 193             (length<((1+indexLength)<<2) ||
 194              length<(indexes[URES_INDEX_BUNDLE_TOP]<<2))
 195         ) {
 196             *errorCode=U_INVALID_FORMAT_ERROR;
 197             res_unload(pResData);
 198             return;
 199         }
 200         if(indexes[URES_INDEX_KEYS_TOP]>(1+indexLength)) {
 201             pResData->localKeyLimit=indexes[URES_INDEX_KEYS_TOP]<<2;
 202         }
 203         if(indexLength>URES_INDEX_ATTRIBUTES) {
 204             int32_t att=indexes[URES_INDEX_ATTRIBUTES];
 205             pResData->noFallback=(UBool)(att&URES_ATT_NO_FALLBACK);
 206             pResData->isPoolBundle=(UBool)((att&URES_ATT_IS_POOL_BUNDLE)!=0);
 207             pResData->usesPoolBundle=(UBool)((att&URES_ATT_USES_POOL_BUNDLE)!=0);
 208         }
 209         if((pResData->isPoolBundle || pResData->usesPoolBundle) && indexLength<=URES_INDEX_POOL_CHECKSUM) {
 210             *errorCode=U_INVALID_FORMAT_ERROR;
 211             res_unload(pResData);
 212             return;
 213         }
 214         if( indexLength>URES_INDEX_16BIT_TOP &&
 215             indexes[URES_INDEX_16BIT_TOP]>indexes[URES_INDEX_KEYS_TOP]
 216         ) {
 217             pResData->p16BitUnits=(const uint16_t *)(pResData->pRoot+indexes[URES_INDEX_KEYS_TOP]);
 218         }
 219     }
 220
 221     if(formatVersion[0]==1 || U_CHARSET_FAMILY==U_ASCII_FAMILY) {
 222         /*
 223          * formatVersion 1: compare key strings in native-charset order
 224          * formatVersion 2 and up: compare key strings in ASCII order
 225          */
 226         pResData->useNativeStrcmp=TRUE;
 227     }
 228 }
 229
 230 U_CAPI void U_EXPORT2
 231 res_read(ResourceData *pResData,
 232          const UDataInfo *pInfo, const void *inBytes, int32_t length,
 233          UErrorCode *errorCode) {
 234     UVersionInfo formatVersion;
 235
 236     uprv_memset(pResData, 0, sizeof(ResourceData));
 237     if(U_FAILURE(*errorCode)) {
 238         return;
 239     }
 240     if(!isAcceptable(formatVersion, NULL, NULL, pInfo)) {
 241         *errorCode=U_INVALID_FORMAT_ERROR;
 242         return;
 243     }
 244     res_init(pResData, formatVersion, inBytes, length, errorCode);
 245 }
 246
 247 U_CFUNC void
 248 res_load(ResourceData *pResData,
 249          const char *path, const char *name, UErrorCode *errorCode) {
 250     UVersionInfo formatVersion;
 251
 252     uprv_memset(pResData, 0, sizeof(ResourceData));
 253
 254     /* load the ResourceBundle file */
 255     pResData->data=udata_openChoice(path, "res", name, isAcceptable, formatVersion, errorCode);
 256     if(U_FAILURE(*errorCode)) {
 257         return;
 258     }
 259
 260     /* get its memory and initialize *pResData */
 261     res_init(pResData, formatVersion, udata_getMemory(pResData->data), -1, errorCode);
 262 }
 263
 264 U_CFUNC void
 265 res_unload(ResourceData *pResData) {
 266     if(pResData->data!=NULL) {
 267         udata_close(pResData->data);
 268         pResData->data=NULL;
 269     }
 270 }
 271
 272 static const int8_t gPublicTypes[URES_LIMIT] = {
 273     URES_STRING,
 274     URES_BINARY,
 275     URES_TABLE,
 276     URES_ALIAS,
 277
 278     URES_TABLE,     /* URES_TABLE32 */
 279     URES_TABLE,     /* URES_TABLE16 */
 280     URES_STRING,    /* URES_STRING_V2 */
 281     URES_INT,
 282
 283     URES_ARRAY,
 284     URES_ARRAY,     /* URES_ARRAY16 */
 285     URES_NONE,
 286     URES_NONE,
 287
 288     URES_NONE,
 289     URES_NONE,
 290     URES_INT_VECTOR,
 291     URES_NONE
 292 };
 293
 294 U_CAPI UResType U_EXPORT2
 295 res_getPublicType(Resource res) {
 296     return (UResType)gPublicTypes[RES_GET_TYPE(res)];
 297 }
 298
 299 U_CAPI const UChar * U_EXPORT2
 300 res_getString(const ResourceData *pResData, Resource res, int32_t *pLength) {
 301     const UChar *p;
 302     uint32_t offset=RES_GET_OFFSET(res);
 303     int32_t length;
 304     if(RES_GET_TYPE(res)==URES_STRING_V2) {
 305         int32_t first;
 306         p=(const UChar *)(pResData->p16BitUnits+offset);
 307         first=*p;
 308         if(!U16_IS_TRAIL(first)) {
 309             length=u_strlen(p);
 310         } else if(first<0xdfef) {
 311             length=first&0x3ff;
 312             ++p;
 313         } else if(first<0xdfff) {
 314             length=((first-0xdfef)<<16)|p[1];
 315             p+=2;
 316         } else {
 317             length=((int32_t)p[1]<<16)|p[2];
 318             p+=3;
 319         }
 320     } else if(res==offset) /* RES_GET_TYPE(res)==URES_STRING */ {
 321         const int32_t *p32= res==0 ? &gEmptyString.length : pResData->pRoot+res;
 322         length=*p32++;
 323         p=(const UChar *)p32;
 324     } else {
 325         p=NULL;
 326         length=0;
 327     }
 328     if(pLength) {
 329         *pLength=length;
 330     }
 331     return p;
 332 }
 333
 334 U_CAPI const UChar * U_EXPORT2
 335 res_getAlias(const ResourceData *pResData, Resource res, int32_t *pLength) {
 336     const UChar *p;
 337     uint32_t offset=RES_GET_OFFSET(res);
 338     int32_t length;
 339     if(RES_GET_TYPE(res)==URES_ALIAS) {
 340         const int32_t *p32= offset==0 ? &gEmptyString.length : pResData->pRoot+offset;
 341         length=*p32++;
 342         p=(const UChar *)p32;
 343     } else {
 344         p=NULL;
 345         length=0;
 346     }
 347     if(pLength) {
 348         *pLength=length;
 349     }
 350     return p;
 351 }
 352
 353 U_CAPI const uint8_t * U_EXPORT2
 354 res_getBinary(const ResourceData *pResData, Resource res, int32_t *pLength) {
 355     const uint8_t *p;
 356     uint32_t offset=RES_GET_OFFSET(res);
 357     int32_t length;
 358     if(RES_GET_TYPE(res)==URES_BINARY) {
 359         const int32_t *p32= offset==0 ? (const int32_t*)&gEmpty32 : pResData->pRoot+offset;
 360         length=*p32++;
 361         p=(const uint8_t *)p32;
 362     } else {
 363         p=NULL;
 364         length=0;
 365     }
 366     if(pLength) {
 367         *pLength=length;
 368     }
 369     return p;
 370 }
 371
 372
 373 U_CAPI const int32_t * U_EXPORT2
 374 res_getIntVector(const ResourceData *pResData, Resource res, int32_t *pLength) {
 375     const int32_t *p;
 376     uint32_t offset=RES_GET_OFFSET(res);
 377     int32_t length;
 378     if(RES_GET_TYPE(res)==URES_INT_VECTOR) {
 379         p= offset==0 ? (const int32_t *)&gEmpty32 : pResData->pRoot+offset;
 380         length=*p++;
 381     } else {
 382         p=NULL;
 383         length=0;
 384     }
 385     if(pLength) {
 386         *pLength=length;
 387     }
 388     return p;
 389 }
 390
 391 U_CAPI int32_t U_EXPORT2
 392 res_countArrayItems(const ResourceData *pResData, Resource res) {
 393     uint32_t offset=RES_GET_OFFSET(res);
 394     switch(RES_GET_TYPE(res)) {
 395     case URES_STRING:
 396     case URES_STRING_V2:
 397     case URES_BINARY:
 398     case URES_ALIAS:
 399     case URES_INT:
 400     case URES_INT_VECTOR:
 401         return 1;
 402     case URES_ARRAY:
 403     case URES_TABLE32:
 404         return offset==0 ? 0 : *(pResData->pRoot+offset);
 405     case URES_TABLE:
 406         return offset==0 ? 0 : *((const uint16_t *)(pResData->pRoot+offset));
 407     case URES_ARRAY16:
 408     case URES_TABLE16:
 409         return pResData->p16BitUnits[offset];
 410     default:
 411         return 0;
 412     }
 413 }
 414
 415 U_CAPI Resource U_EXPORT2
 416 res_getTableItemByKey(const ResourceData *pResData, Resource table,
 417                       int32_t *indexR, const char **key) {
 418     uint32_t offset=RES_GET_OFFSET(table);
 419     int32_t length;
 420     int32_t idx;
 421     if(key == NULL || *key == NULL) {
 422         return RES_BOGUS;
 423     }
 424     switch(RES_GET_TYPE(table)) {
 425     case URES_TABLE: {
 426         if (offset!=0) { /* empty if offset==0 */
 427             const uint16_t *p= (const uint16_t *)(pResData->pRoot+offset);
 428             length=*p++;
 429             *indexR=idx=_res_findTableItem(pResData, p, length, *key, key);
 430             if(idx>=0) {
 431                 const Resource *p32=(const Resource *)(p+length+(~length&1));
 432                 return p32[idx];
 433             }
 434         }
 435         break;
 436     }
 437     case URES_TABLE16: {
 438         const uint16_t *p=pResData->p16BitUnits+offset;
 439         length=*p++;
 440         *indexR=idx=_res_findTableItem(pResData, p, length, *key, key);
 441         if(idx>=0) {
 442             return URES_MAKE_RESOURCE(URES_STRING_V2, p[length+idx]);
 443         }
 444         break;
 445     }
 446     case URES_TABLE32: {
 447         if (offset!=0) { /* empty if offset==0 */
 448             const int32_t *p= pResData->pRoot+offset;
 449             length=*p++;
 450             *indexR=idx=_res_findTable32Item(pResData, p, length, *key, key);
 451             if(idx>=0) {
 452                 return (Resource)p[length+idx];
 453             }
 454         }
 455         break;
 456     }
 457     default:
 458         break;
 459     }
 460     return RES_BOGUS;
 461 }
 462
 463 U_CAPI Resource U_EXPORT2
 464 res_getTableItemByIndex(const ResourceData *pResData, Resource table,
 465                         int32_t indexR, const char **key) {
 466     uint32_t offset=RES_GET_OFFSET(table);
 467     int32_t length;
 468     U_ASSERT(indexR>=0); /* to ensure the index is not negative */
 469     switch(RES_GET_TYPE(table)) {
 470     case URES_TABLE: {
 471         if (offset != 0) { /* empty if offset==0 */
 472             const uint16_t *p= (const uint16_t *)(pResData->pRoot+offset);
 473             length=*p++;
 474             if(indexR<length) {
 475                 const Resource *p32=(const Resource *)(p+length+(~length&1));
 476                 if(key!=NULL) {
 477                     *key=RES_GET_KEY16(pResData, p[indexR]);
 478                 }
 479                 return p32[indexR];
 480             }
 481         }
 482         break;
 483     }
 484     case URES_TABLE16: {
 485         const uint16_t *p=pResData->p16BitUnits+offset;
 486         length=*p++;
 487         if(indexR<length) {
 488             if(key!=NULL) {
 489                 *key=RES_GET_KEY16(pResData, p[indexR]);
 490             }
 491             return URES_MAKE_RESOURCE(URES_STRING_V2, p[length+indexR]);
 492         }
 493         break;
 494     }
 495     case URES_TABLE32: {
 496         if (offset != 0) { /* empty if offset==0 */
 497             const int32_t *p= pResData->pRoot+offset;
 498             length=*p++;
 499             if(indexR<length) {
 500                 if(key!=NULL) {
 501                     *key=RES_GET_KEY32(pResData, p[indexR]);
 502                 }
 503                 return (Resource)p[length+indexR];
 504             }
 505         }
 506         break;
 507     }
 508     default:
 509         break;
 510     }
 511     return RES_BOGUS;
 512 }
 513
 514 U_CAPI Resource U_EXPORT2
 515 res_getResource(const ResourceData *pResData, const char *key) {
 516     const char *realKey=key;
 517     int32_t idx;
 518     return res_getTableItemByKey(pResData, pResData->rootRes, &idx, &realKey);
 519 }
 520
 521 U_CAPI Resource U_EXPORT2
 522 res_getArrayItem(const ResourceData *pResData, Resource array, int32_t indexR) {
 523     uint32_t offset=RES_GET_OFFSET(array);
 524     U_ASSERT(indexR>=0); /* to ensure the index is not negative */
 525     switch(RES_GET_TYPE(array)) {
 526     case URES_ARRAY: {
 527         if (offset!=0) { /* empty if offset==0 */
 528             const int32_t *p= pResData->pRoot+offset;
 529             if(indexR<*p) {
 530                 return (Resource)p[1+indexR];
 531             }
 532         }
 533         break;
 534     }
 535     case URES_ARRAY16: {
 536         const uint16_t *p=pResData->p16BitUnits+offset;
 537         if(indexR<*p) {
 538             return URES_MAKE_RESOURCE(URES_STRING_V2, p[1+indexR]);
 539         }
 540         break;
 541     }
 542     default:
 543         break;
 544     }
 545     return RES_BOGUS;
 546 }
 547
 548 U_CFUNC Resource
 549 res_findResource(const ResourceData *pResData, Resource r, char** path, const char** key) {
 550   /* we pass in a path. CollationElements/Sequence or zoneStrings/3/2 etc.
 551    * iterates over a path and stops when a scalar resource is found. This
 552    * CAN be an alias. Path gets set to the part that has not yet been processed.
 553    */
 554
 555   char *pathP = *path, *nextSepP = *path;
 556   char *closeIndex = NULL;
 557   Resource t1 = r;
 558   Resource t2;
 559   int32_t indexR = 0;
 560   UResType type = (UResType)RES_GET_TYPE(t1);
 561
 562   /* if you come in with an empty path, you'll be getting back the same resource */
 563   if(!uprv_strlen(pathP)) {
 564       return r;
 565   }
 566
 567   /* one needs to have an aggregate resource in order to search in it */
 568   if(!URES_IS_CONTAINER(type)) {
 569       return RES_BOGUS;
 570   }
 571
 572   while(nextSepP && *pathP && t1 != RES_BOGUS && URES_IS_CONTAINER(type)) {
 573     /* Iteration stops if: the path has been consumed, we found a non-existing
 574      * resource (t1 == RES_BOGUS) or we found a scalar resource (including alias)
 575      */
 576     nextSepP = uprv_strchr(pathP, RES_PATH_SEPARATOR);
 577     /* if there are more separators, terminate string
 578      * and set path to the remaining part of the string
 579      */
 580     if(nextSepP != NULL) {
 581       *nextSepP = 0; /* overwrite the separator with a NUL to terminate the key */
 582       *path = nextSepP+1;
 583     } else {
 584       *path = uprv_strchr(pathP, 0);
 585     }
 586
 587     /* if the resource is a table */
 588     /* try the key based access */
 589     if(URES_IS_TABLE(type)) {
 590       *key = pathP;
 591       t2 = res_getTableItemByKey(pResData, t1, &indexR, key);
 592       if(t2 == RES_BOGUS) {
 593         /* if we fail to get the resource by key, maybe we got an index */
 594         indexR = uprv_strtol(pathP, &closeIndex, 10);
 595         if(closeIndex != pathP) {
 596           /* if we indeed have an index, try to get the item by index */
 597           t2 = res_getTableItemByIndex(pResData, t1, indexR, key);
 598         }
 599       }
 600     } else if(URES_IS_ARRAY(type)) {
 601       indexR = uprv_strtol(pathP, &closeIndex, 10);
 602       if(closeIndex != pathP) {
 603         t2 = res_getArrayItem(pResData, t1, indexR);
 604       } else {
 605         t2 = RES_BOGUS; /* have an array, but don't have a valid index */
 606       }
 607       *key = NULL;
 608     } else { /* can't do much here, except setting t2 to bogus */
 609       t2 = RES_BOGUS;
 610     }
 611     t1 = t2;
 612     type = (UResType)RES_GET_TYPE(t1);
 613     /* position pathP to next resource key/index */
 614     pathP = *path;
 615   }
 616
 617   return t1;
 618 }
 619
 620 /* resource bundle swapping ------------------------------------------------- */
 621
 622 /*
 623  * Need to always enumerate the entire item tree,
 624  * track the lowest address of any item to use as the limit for char keys[],
 625  * track the highest address of any item to return the size of the data.
 626  *
 627  * We should have thought of storing those in the data...
 628  * It is possible to extend the data structure by putting additional values
 629  * in places that are inaccessible by ordinary enumeration of the item tree.
 630  * For example, additional integers could be stored at the beginning or
 631  * end of the key strings; this could be indicated by a minor version number,
 632  * and the data swapping would have to know about these values.
 633  *
 634  * The data structure does not forbid keys to be shared, so we must swap
 635  * all keys once instead of each key when it is referenced.
 636  *
 637  * These swapping functions assume that a resource bundle always has a length
 638  * that is a multiple of 4 bytes.
 639  * Currently, this is trivially true because genrb writes bundle tree leaves
 640  * physically first, before their branches, so that the root table with its
 641  * array of resource items (uint32_t values) is always last.
 642  */
 643
 644 /* definitions for table sorting ------------------------ */
 645
 646 /*
 647  * row of a temporary array
 648  *
 649  * gets platform-endian key string indexes and sorting indexes;
 650  * after sorting this array by keys, the actual key/value arrays are permutated
 651  * according to the sorting indexes
 652  */
 653 typedef struct Row {
 654     int32_t keyIndex, sortIndex;
 655 } Row;
 656
 657 static int32_t
 658 ures_compareRows(const void *context, const void *left, const void *right) {
 659     const char *keyChars=(const char *)context;
 660     return (int32_t)uprv_strcmp(keyChars+((const Row *)left)->keyIndex,
 661                                 keyChars+((const Row *)right)->keyIndex);
 662 }
 663
 664 typedef struct TempTable {
 665     const char *keyChars;
 666     Row *rows;
 667     int32_t *resort;
 668     uint32_t *resFlags;
 669     int32_t localKeyLimit;
 670     uint8_t majorFormatVersion;
 671 } TempTable;
 672
 673 enum {
 674     STACK_ROW_CAPACITY=200
 675 };
 676
 677 /* The table item key string is not locally available. */
 678 static const char *const gUnknownKey="";
 679
 680 /* resource table key for collation binaries: "%%CollationBin" */
 681 static const UChar gCollationBinKey[]={
 682     0x25, 0x25,
 683     0x43, 0x6f, 0x6c, 0x6c, 0x61, 0x74, 0x69, 0x6f, 0x6e,
 684     0x42, 0x69, 0x6e,
 685     0
 686 };
 687
 688 /*
 689  * swap one resource item
 690  */
 691 static void
 692 ures_swapResource(const UDataSwapper *ds,
 693                   const Resource *inBundle, Resource *outBundle,
 694                   Resource res, /* caller swaps res itself */
 695                   const char *key,
 696                   TempTable *pTempTable,
 697                   UErrorCode *pErrorCode) {
 698     const Resource *p;
 699     Resource *q;
 700     int32_t offset, count;
 701
 702     switch(RES_GET_TYPE(res)) {
 703     case URES_TABLE16:
 704     case URES_STRING_V2:
 705     case URES_INT:
 706     case URES_ARRAY16:
 707         /* integer, or points to 16-bit units, nothing to do here */
 708         return;
 709     default:
 710         break;
 711     }
 712
 713     /* all other types use an offset to point to their data */
 714     offset=(int32_t)RES_GET_OFFSET(res);
 715     if(offset==0) {
 716         /* special offset indicating an empty item */
 717         return;
 718     }
 719     if(pTempTable->resFlags[offset>>5]&((uint32_t)1<<(offset&0x1f))) {
 720         /* we already swapped this resource item */
 721         return;
 722     } else {
 723         /* mark it as swapped now */
 724         pTempTable->resFlags[offset>>5]|=((uint32_t)1<<(offset&0x1f));
 725     }
 726
 727     p=inBundle+offset;
 728     q=outBundle+offset;
 729
 730     switch(RES_GET_TYPE(res)) {
 731     case URES_ALIAS:
 732         /* physically same value layout as string, fall through */
 733     case URES_STRING:
 734         count=udata_readInt32(ds, (int32_t)*p);
 735         /* swap length */
 736         ds->swapArray32(ds, p, 4, q, pErrorCode);
 737         /* swap each UChar (the terminating NUL would not change) */
 738         ds->swapArray16(ds, p+1, 2*count, q+1, pErrorCode);
 739         break;
 740     case URES_BINARY:
 741         count=udata_readInt32(ds, (int32_t)*p);
 742         /* swap length */
 743         ds->swapArray32(ds, p, 4, q, pErrorCode);
 744         /* no need to swap or copy bytes - ures_swap() copied them all */
 745
 746         /* swap known formats */
 747 #if !UCONFIG_NO_COLLATION
 748         if( key!=NULL &&  /* the binary is in a table */
 749             (key!=gUnknownKey ?
 750                 /* its table key string is "%%CollationBin" */
 751                 0==ds->compareInvChars(ds, key, -1,
 752                                        gCollationBinKey, LENGTHOF(gCollationBinKey)-1) :
 753                 /* its table key string is unknown but it looks like a collation binary */
 754                 ucol_looksLikeCollationBinary(ds, p+1, count))
 755         ) {
 756             ucol_swap(ds, p+1, count, q+1, pErrorCode);
 757         }
 758 #endif
 759         break;
 760     case URES_TABLE:
 761     case URES_TABLE32:
 762         {
 763             const uint16_t *pKey16;
 764             uint16_t *qKey16;
 765
 766             const int32_t *pKey32;
 767             int32_t *qKey32;
 768
 769             Resource item;
 770             int32_t i, oldIndex;
 771
 772             if(RES_GET_TYPE(res)==URES_TABLE) {
 773                 /* get table item count */
 774                 pKey16=(const uint16_t *)p;
 775                 qKey16=(uint16_t *)q;
 776                 count=ds->readUInt16(*pKey16);
 777
 778                 pKey32=qKey32=NULL;
 779
 780                 /* swap count */
 781                 ds->swapArray16(ds, pKey16++, 2, qKey16++, pErrorCode);
 782
 783                 offset+=((1+count)+1)/2;
 784             } else {
 785                 /* get table item count */
 786                 pKey32=(const int32_t *)p;
 787                 qKey32=(int32_t *)q;
 788                 count=udata_readInt32(ds, *pKey32);
 789
 790                 pKey16=qKey16=NULL;
 791
 792                 /* swap count */
 793                 ds->swapArray32(ds, pKey32++, 4, qKey32++, pErrorCode);
 794
 795                 offset+=1+count;
 796             }
 797
 798             if(count==0) {
 799                 break;
 800             }
 801
 802             p=inBundle+offset; /* pointer to table resources */
 803             q=outBundle+offset;
 804
 805             /* recurse */
 806             for(i=0; i<count; ++i) {
 807                 const char *itemKey=gUnknownKey;
 808                 if(pKey16!=NULL) {
 809                     int32_t keyOffset=ds->readUInt16(pKey16[i]);
 810                     if(keyOffset<pTempTable->localKeyLimit) {
 811                         itemKey=(const char *)outBundle+keyOffset;
 812                     }
 813                 } else {
 814                     int32_t keyOffset=udata_readInt32(ds, pKey32[i]);
 815                     if(keyOffset>=0) {
 816                         itemKey=(const char *)outBundle+keyOffset;
 817                     }
 818                 }
 819                 item=ds->readUInt32(p[i]);
 820                 ures_swapResource(ds, inBundle, outBundle, item, itemKey, pTempTable, pErrorCode);
 821                 if(U_FAILURE(*pErrorCode)) {
 822                     udata_printError(ds, "ures_swapResource(table res=%08x)[%d].recurse(%08x) failed\n",
 823                                      res, i, item);
 824                     return;
 825                 }
 826             }
 827
 828             if(pTempTable->majorFormatVersion>1 || ds->inCharset==ds->outCharset) {
 829                 /* no need to sort, just swap the offset/value arrays */
 830                 if(pKey16!=NULL) {
 831                     ds->swapArray16(ds, pKey16, count*2, qKey16, pErrorCode);
 832                     ds->swapArray32(ds, p, count*4, q, pErrorCode);
 833                 } else {
 834                     /* swap key offsets and items as one array */
 835                     ds->swapArray32(ds, pKey32, count*2*4, qKey32, pErrorCode);
 836                 }
 837                 break;
 838             }
 839
 840             /*
 841              * We need to sort tables by outCharset key strings because they
 842              * sort differently for different charset families.
 843              * ures_swap() already set pTempTable->keyChars appropriately.
 844              * First we set up a temporary table with the key indexes and
 845              * sorting indexes and sort that.
 846              * Then we permutate and copy/swap the actual values.
 847              */
 848             if(pKey16!=NULL) {
 849                 for(i=0; i<count; ++i) {
 850                     pTempTable->rows[i].keyIndex=ds->readUInt16(pKey16[i]);
 851                     pTempTable->rows[i].sortIndex=i;
 852                 }
 853             } else {
 854                 for(i=0; i<count; ++i) {
 855                     pTempTable->rows[i].keyIndex=udata_readInt32(ds, pKey32[i]);
 856                     pTempTable->rows[i].sortIndex=i;
 857                 }
 858             }
 859             uprv_sortArray(pTempTable->rows, count, sizeof(Row),
 860                            ures_compareRows, pTempTable->keyChars,
 861                            FALSE, pErrorCode);
 862             if(U_FAILURE(*pErrorCode)) {
 863                 udata_printError(ds, "ures_swapResource(table res=%08x).uprv_sortArray(%d items) failed\n",
 864                                  res, count);
 865                 return;
 866             }
 867
 868             /*
 869              * copy/swap/permutate items
 870              *
 871              * If we swap in-place, then the permutation must use another
 872              * temporary array (pTempTable->resort)
 873              * before the results are copied to the outBundle.
 874              */
 875             /* keys */
 876             if(pKey16!=NULL) {
 877                 uint16_t *rKey16;
 878
 879                 if(pKey16!=qKey16) {
 880                     rKey16=qKey16;
 881                 } else {
 882                     rKey16=(uint16_t *)pTempTable->resort;
 883                 }
 884                 for(i=0; i<count; ++i) {
 885                     oldIndex=pTempTable->rows[i].sortIndex;
 886                     ds->swapArray16(ds, pKey16+oldIndex, 2, rKey16+i, pErrorCode);
 887                 }
 888                 if(qKey16!=rKey16) {
 889                     uprv_memcpy(qKey16, rKey16, 2*count);
 890                 }
 891             } else {
 892                 int32_t *rKey32;
 893
 894                 if(pKey32!=qKey32) {
 895                     rKey32=qKey32;
 896                 } else {
 897                     rKey32=pTempTable->resort;
 898                 }
 899                 for(i=0; i<count; ++i) {
 900                     oldIndex=pTempTable->rows[i].sortIndex;
 901                     ds->swapArray32(ds, pKey32+oldIndex, 4, rKey32+i, pErrorCode);
 902                 }
 903                 if(qKey32!=rKey32) {
 904                     uprv_memcpy(qKey32, rKey32, 4*count);
 905                 }
 906             }
 907
 908             /* resources */
 909             {
 910                 Resource *r;
 911
 912
 913                 if(p!=q) {
 914                     r=q;
 915                 } else {
 916                     r=(Resource *)pTempTable->resort;
 917                 }
 918                 for(i=0; i<count; ++i) {
 919                     oldIndex=pTempTable->rows[i].sortIndex;
 920                     ds->swapArray32(ds, p+oldIndex, 4, r+i, pErrorCode);
 921                 }
 922                 if(q!=r) {
 923                     uprv_memcpy(q, r, 4*count);
 924                 }
 925             }
 926         }
 927         break;
 928     case URES_ARRAY:
 929         {
 930             Resource item;
 931             int32_t i;
 932
 933             count=udata_readInt32(ds, (int32_t)*p);
 934             /* swap length */
 935             ds->swapArray32(ds, p++, 4, q++, pErrorCode);
 936
 937             /* recurse */
 938             for(i=0; i<count; ++i) {
 939                 item=ds->readUInt32(p[i]);
 940                 ures_swapResource(ds, inBundle, outBundle, item, NULL, pTempTable, pErrorCode);
 941                 if(U_FAILURE(*pErrorCode)) {
 942                     udata_printError(ds, "ures_swapResource(array res=%08x)[%d].recurse(%08x) failed\n",
 943                                      res, i, item);
 944                     return;
 945                 }
 946             }
 947
 948             /* swap items */
 949             ds->swapArray32(ds, p, 4*count, q, pErrorCode);
 950         }
 951         break;
 952     case URES_INT_VECTOR:
 953         count=udata_readInt32(ds, (int32_t)*p);
 954         /* swap length and each integer */
 955         ds->swapArray32(ds, p, 4*(1+count), q, pErrorCode);
 956         break;
 957     default:
 958         /* also catches RES_BOGUS */
 959         *pErrorCode=U_UNSUPPORTED_ERROR;
 960         break;
 961     }
 962 }
 963
 964 U_CAPI int32_t U_EXPORT2
 965 ures_swap(const UDataSwapper *ds,
 966           const void *inData, int32_t length, void *outData,
 967           UErrorCode *pErrorCode) {
 968     const UDataInfo *pInfo;
 969     const Resource *inBundle;
 970     Resource rootRes;
 971     int32_t headerSize, maxTableLength;
 972
 973     Row rows[STACK_ROW_CAPACITY];
 974     int32_t resort[STACK_ROW_CAPACITY];
 975     TempTable tempTable;
 976
 977     const int32_t *inIndexes;
 978
 979     /* the following integers count Resource item offsets (4 bytes each), not bytes */
 980     int32_t bundleLength, indexLength, keysBottom, keysTop, resBottom, top;
 981
 982     /* udata_swapDataHeader checks the arguments */
 983     headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
 984     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 985         return 0;
 986     }
 987
 988     /* check data format and format version */
 989     pInfo=(const UDataInfo *)((const char *)inData+4);
 990     if(!(
 991         pInfo->dataFormat[0]==0x52 &&   /* dataFormat="ResB" */
 992         pInfo->dataFormat[1]==0x65 &&
 993         pInfo->dataFormat[2]==0x73 &&
 994         pInfo->dataFormat[3]==0x42 &&
 995         ((pInfo->formatVersion[0]==1 && pInfo->formatVersion[1]>=1) ||  /* formatVersion 1.1+ or 2.x */
 996          pInfo->formatVersion[0]==2)
 997     )) {
 998         udata_printError(ds, "ures_swap(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not a resource bundle\n",
 999                          pInfo->dataFormat[0], pInfo->dataFormat[1],
1000                          pInfo->dataFormat[2], pInfo->dataFormat[3],
1001                          pInfo->formatVersion[0], pInfo->formatVersion[1]);
1002         *pErrorCode=U_UNSUPPORTED_ERROR;
1003         return 0;
1004     }
1005     tempTable.majorFormatVersion=pInfo->formatVersion[0];
1006
1007     /* a resource bundle must contain at least one resource item */
1008     if(length<0) {
1009         bundleLength=-1;
1010     } else {
1011         bundleLength=(length-headerSize)/4;
1012
1013         /* formatVersion 1.1 must have a root item and at least 5 indexes */
1014         if(bundleLength<(1+5)) {
1015             udata_printError(ds, "ures_swap(): too few bytes (%d after header) for a resource bundle\n",
1016                              length-headerSize);
1017             *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1018             return 0;
1019         }
1020     }
1021
1022     inBundle=(const Resource *)((const char *)inData+headerSize);
1023     rootRes=ds->readUInt32(*inBundle);
1024
1025     /* formatVersion 1.1 adds the indexes[] array */
1026     inIndexes=(const int32_t *)(inBundle+1);
1027
1028     indexLength=udata_readInt32(ds, inIndexes[URES_INDEX_LENGTH])&0xff;
1029     if(indexLength<=URES_INDEX_MAX_TABLE_LENGTH) {
1030         udata_printError(ds, "ures_swap(): too few indexes for a 1.1+ resource bundle\n");
1031         *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1032         return 0;
1033     }
1034     keysBottom=1+indexLength;
1035     keysTop=udata_readInt32(ds, inIndexes[URES_INDEX_KEYS_TOP]);
1036     if(indexLength>URES_INDEX_16BIT_TOP) {
1037         resBottom=udata_readInt32(ds, inIndexes[URES_INDEX_16BIT_TOP]);
1038     } else {
1039         resBottom=keysTop;
1040     }
1041     top=udata_readInt32(ds, inIndexes[URES_INDEX_BUNDLE_TOP]);
1042     maxTableLength=udata_readInt32(ds, inIndexes[URES_INDEX_MAX_TABLE_LENGTH]);
1043
1044     if(0<=bundleLength && bundleLength<top) {
1045         udata_printError(ds, "ures_swap(): resource top %d exceeds bundle length %d\n",
1046                          top, bundleLength);
1047         *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1048         return 0;
1049     }
1050     if(keysTop>(1+indexLength)) {
1051         tempTable.localKeyLimit=keysTop<<2;
1052     } else {
1053         tempTable.localKeyLimit=0;
1054     }
1055
1056     if(length>=0) {
1057         Resource *outBundle=(Resource *)((char *)outData+headerSize);
1058
1059         /* track which resources we have already swapped */
1060         uint32_t stackResFlags[STACK_ROW_CAPACITY];
1061         int32_t resFlagsLength;
1062
1063         /*
1064          * We need one bit per 4 resource bundle bytes so that we can track
1065          * every possible Resource for whether we have swapped it already.
1066          * Multiple Resource words can refer to the same bundle offsets
1067          * for sharing identical values.
1068          * We could optimize this by allocating only for locations above
1069          * where Resource values are stored (above keys & strings).
1070          */
1071         resFlagsLength=(length+31)>>5;          /* number of bytes needed */
1072         resFlagsLength=(resFlagsLength+3)&~3;   /* multiple of 4 bytes for uint32_t */
1073         if(resFlagsLength<=sizeof(stackResFlags)) {
1074             tempTable.resFlags=stackResFlags;
1075         } else {
1076             tempTable.resFlags=(uint32_t *)uprv_malloc(resFlagsLength);
1077             if(tempTable.resFlags==NULL) {
1078                 udata_printError(ds, "ures_swap(): unable to allocate memory for tracking resources\n");
1079                 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
1080                 return 0;
1081             }
1082         }
1083         uprv_memset(tempTable.resFlags, 0, resFlagsLength);
1084
1085         /* copy the bundle for binary and inaccessible data */
1086         if(inData!=outData) {
1087             uprv_memcpy(outBundle, inBundle, 4*top);
1088         }
1089
1090         /* swap the key strings, but not the padding bytes (0xaa) after the last string and its NUL */
1091         udata_swapInvStringBlock(ds, inBundle+keysBottom, 4*(keysTop-keysBottom),
1092                                     outBundle+keysBottom, pErrorCode);
1093         if(U_FAILURE(*pErrorCode)) {
1094             udata_printError(ds, "ures_swap().udata_swapInvStringBlock(keys[%d]) failed\n", 4*(keysTop-keysBottom));
1095             return 0;
1096         }
1097
1098         /* swap the 16-bit units (strings, table16, array16) */
1099         if(keysTop<resBottom) {
1100             ds->swapArray16(ds, inBundle+keysTop, (resBottom-keysTop)*4, outBundle+keysTop, pErrorCode);
1101             if(U_FAILURE(*pErrorCode)) {
1102                 udata_printError(ds, "ures_swap().swapArray16(16-bit units[%d]) failed\n", 2*(resBottom-keysTop));
1103                 return 0;
1104             }
1105         }
1106
1107         /* allocate the temporary table for sorting resource tables */
1108         tempTable.keyChars=(const char *)outBundle; /* sort by outCharset */
1109         if(tempTable.majorFormatVersion>1 || maxTableLength<=STACK_ROW_CAPACITY) {
1110             tempTable.rows=rows;
1111             tempTable.resort=resort;
1112         } else {
1113             tempTable.rows=(Row *)uprv_malloc(maxTableLength*sizeof(Row)+maxTableLength*4);
1114             if(tempTable.rows==NULL) {
1115                 udata_printError(ds, "ures_swap(): unable to allocate memory for sorting tables (max length: %d)\n",
1116                                  maxTableLength);
1117                 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
1118                 if(tempTable.resFlags!=stackResFlags) {
1119                     uprv_free(tempTable.resFlags);
1120                 }
1121                 return 0;
1122             }
1123             tempTable.resort=(int32_t *)(tempTable.rows+maxTableLength);
1124         }
1125
1126         /* swap the resources */
1127         ures_swapResource(ds, inBundle, outBundle, rootRes, NULL, &tempTable, pErrorCode);
1128         if(U_FAILURE(*pErrorCode)) {
1129             udata_printError(ds, "ures_swapResource(root res=%08x) failed\n",
1130                              rootRes);
1131         }
1132
1133         if(tempTable.rows!=rows) {
1134             uprv_free(tempTable.rows);
1135         }
1136         if(tempTable.resFlags!=stackResFlags) {
1137             uprv_free(tempTable.resFlags);
1138         }
1139
1140         /* swap the root resource and indexes */
1141         ds->swapArray32(ds, inBundle, keysBottom*4, outBundle, pErrorCode);
1142     }
1143
1144     return headerSize+4*top;
1145 }