icuSources/common/ucol_swp.c

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2003-2006, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  ucol_swp.c
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2003sep10
  14 *   created by: Markus W. Scherer
  15 *
  16 *   Swap collation binaries.
  17 */
  18
  19 #include "unicode/udata.h" /* UDataInfo */
  20 #include "utrie.h"
  21 #include "udataswp.h"
  22 #include "cmemory.h"
  23 #include "ucol_imp.h"
  24 #include "ucol_swp.h"
  25
  26 /* swapping ----------------------------------------------------------------- */
  27
  28 /*
  29  * This performs data swapping for a folded trie (see utrie.c for details).
  30  */
  31
  32 U_CAPI int32_t U_EXPORT2
  33 utrie_swap(const UDataSwapper *ds,
  34            const void *inData, int32_t length, void *outData,
  35            UErrorCode *pErrorCode) {
  36     const UTrieHeader *inTrie;
  37     UTrieHeader trie;
  38     int32_t size;
  39     UBool dataIs32;
  40
  41     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
  42         return 0;
  43     }
  44     if(ds==NULL || inData==NULL || (length>=0 && outData==NULL)) {
  45         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
  46         return 0;
  47     }
  48
  49     /* setup and swapping */
  50     if(length>=0 && length<sizeof(UTrieHeader)) {
  51         *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
  52         return 0;
  53     }
  54
  55     inTrie=(const UTrieHeader *)inData;
  56     trie.signature=ds->readUInt32(inTrie->signature);
  57     trie.options=ds->readUInt32(inTrie->options);
  58     trie.indexLength=udata_readInt32(ds, inTrie->indexLength);
  59     trie.dataLength=udata_readInt32(ds, inTrie->dataLength);
  60
  61     if( trie.signature!=0x54726965 ||
  62         (trie.options&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_SHIFT ||
  63         ((trie.options>>UTRIE_OPTIONS_INDEX_SHIFT)&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_INDEX_SHIFT ||
  64         trie.indexLength<UTRIE_BMP_INDEX_LENGTH ||
  65         (trie.indexLength&(UTRIE_SURROGATE_BLOCK_COUNT-1))!=0 ||
  66         trie.dataLength<UTRIE_DATA_BLOCK_LENGTH ||
  67         (trie.dataLength&(UTRIE_DATA_GRANULARITY-1))!=0 ||
  68         ((trie.options&UTRIE_OPTIONS_LATIN1_IS_LINEAR)!=0 && trie.dataLength<(UTRIE_DATA_BLOCK_LENGTH+0x100))
  69     ) {
  70         *pErrorCode=U_INVALID_FORMAT_ERROR; /* not a UTrie */
  71         return 0;
  72     }
  73
  74     dataIs32=(UBool)((trie.options&UTRIE_OPTIONS_DATA_IS_32_BIT)!=0);
  75     size=sizeof(UTrieHeader)+trie.indexLength*2+trie.dataLength*(dataIs32?4:2);
  76
  77     if(length>=0) {
  78         UTrieHeader *outTrie;
  79
  80         if(length<size) {
  81             *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
  82             return 0;
  83         }
  84
  85         outTrie=(UTrieHeader *)outData;
  86
  87         /* swap the header */
  88         ds->swapArray32(ds, inTrie, sizeof(UTrieHeader), outTrie, pErrorCode);
  89
  90         /* swap the index and the data */
  91         if(dataIs32) {
  92             ds->swapArray16(ds, inTrie+1, trie.indexLength*2, outTrie+1, pErrorCode);
  93             ds->swapArray32(ds, (const uint16_t *)(inTrie+1)+trie.indexLength, trie.dataLength*4,
  94                                      (uint16_t *)(outTrie+1)+trie.indexLength, pErrorCode);
  95         } else {
  96             ds->swapArray16(ds, inTrie+1, (trie.indexLength+trie.dataLength)*2, outTrie+1, pErrorCode);
  97         }
  98     }
  99
 100     return size;
 101 }
 102
 103 #if !UCONFIG_NO_COLLATION
 104
 105 /* swap a header-less collation binary, inside a resource bundle or ucadata.icu */
 106 U_CAPI int32_t U_EXPORT2
 107 ucol_swapBinary(const UDataSwapper *ds,
 108                 const void *inData, int32_t length, void *outData,
 109                 UErrorCode *pErrorCode) {
 110     const uint8_t *inBytes;
 111     uint8_t *outBytes;
 112
 113     const UCATableHeader *inHeader;
 114     UCATableHeader *outHeader;
 115     UCATableHeader header={ 0 };
 116
 117     uint32_t count;
 118
 119     /* argument checking in case we were not called from ucol_swap() */
 120     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 121         return 0;
 122     }
 123     if(ds==NULL || inData==NULL || length<-1 || (length>0 && outData==NULL)) {
 124         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 125         return 0;
 126     }
 127
 128     inBytes=(const uint8_t *)inData;
 129     outBytes=(uint8_t *)outData;
 130
 131     inHeader=(const UCATableHeader *)inData;
 132     outHeader=(UCATableHeader *)outData;
 133
 134     /*
 135      * The collation binary must contain at least the UCATableHeader,
 136      * starting with its size field.
 137      * sizeof(UCATableHeader)==42*4 in ICU 2.8
 138      * check the length against the header size before reading the size field
 139      */
 140     if(length<0) {
 141         header.size=udata_readInt32(ds, inHeader->size);
 142     } else if((length<(42*4) || length<(header.size=udata_readInt32(ds, inHeader->size)))) {
 143         udata_printError(ds, "ucol_swapBinary(): too few bytes (%d after header) for collation data\n",
 144                          length);
 145         *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
 146         return 0;
 147     }
 148
 149     header.magic=ds->readUInt32(inHeader->magic);
 150     if(!(
 151         header.magic==UCOL_HEADER_MAGIC &&
 152         inHeader->formatVersion[0]==2 &&
 153         inHeader->formatVersion[1]>=3
 154     )) {
 155         udata_printError(ds, "ucol_swapBinary(): magic 0x%08x or format version %02x.%02x is not a collation binary\n",
 156                          header.magic,
 157                          inHeader->formatVersion[0], inHeader->formatVersion[1]);
 158         *pErrorCode=U_UNSUPPORTED_ERROR;
 159         return 0;
 160     }
 161
 162     if(inHeader->isBigEndian!=ds->inIsBigEndian || inHeader->charSetFamily!=ds->inCharset) {
 163         udata_printError(ds, "ucol_swapBinary(): endianness %d or charset %d does not match the swapper\n",
 164                          inHeader->isBigEndian, inHeader->charSetFamily);
 165         *pErrorCode=U_INVALID_FORMAT_ERROR;
 166         return 0;
 167     }
 168
 169     if(length>=0) {
 170         /* copy everything, takes care of data that needs no swapping */
 171         if(inBytes!=outBytes) {
 172             uprv_memcpy(outBytes, inBytes, header.size);
 173         }
 174
 175         /* swap the necessary pieces in the order of their occurrence in the data */
 176
 177         /* read more of the UCATableHeader (the size field was read above) */
 178         header.options=                 ds->readUInt32(inHeader->options);
 179         header.UCAConsts=               ds->readUInt32(inHeader->UCAConsts);
 180         header.contractionUCACombos=    ds->readUInt32(inHeader->contractionUCACombos);
 181         header.mappingPosition=         ds->readUInt32(inHeader->mappingPosition);
 182         header.expansion=               ds->readUInt32(inHeader->expansion);
 183         header.contractionIndex=        ds->readUInt32(inHeader->contractionIndex);
 184         header.contractionCEs=          ds->readUInt32(inHeader->contractionCEs);
 185         header.contractionSize=         ds->readUInt32(inHeader->contractionSize);
 186         header.endExpansionCE=          ds->readUInt32(inHeader->endExpansionCE);
 187         header.expansionCESize=         ds->readUInt32(inHeader->expansionCESize);
 188         header.endExpansionCECount=     udata_readInt32(ds, inHeader->endExpansionCECount);
 189         header.contractionUCACombosSize=udata_readInt32(ds, inHeader->contractionUCACombosSize);
 190
 191         /* swap the 32-bit integers in the header */
 192         ds->swapArray32(ds, inHeader, (int32_t)((const char *)&inHeader->jamoSpecial-(const char *)inHeader),
 193                            outHeader, pErrorCode);
 194
 195         /* set the output platform properties */
 196         outHeader->isBigEndian=ds->outIsBigEndian;
 197         outHeader->charSetFamily=ds->outCharset;
 198
 199         /* swap the options */
 200         if(header.options!=0) {
 201             ds->swapArray32(ds, inBytes+header.options, header.expansion-header.options,
 202                                outBytes+header.options, pErrorCode);
 203         }
 204
 205         /* swap the expansions */
 206         if(header.mappingPosition!=0 && header.expansion!=0) {
 207             if(header.contractionIndex!=0) {
 208                 /* expansions bounded by contractions */
 209                 count=header.contractionIndex-header.expansion;
 210             } else {
 211                 /* no contractions: expansions bounded by the main trie */
 212                 count=header.mappingPosition-header.expansion;
 213             }
 214             ds->swapArray32(ds, inBytes+header.expansion, (int32_t)count,
 215                                outBytes+header.expansion, pErrorCode);
 216         }
 217
 218         /* swap the contractions */
 219         if(header.contractionSize!=0) {
 220             /* contractionIndex: UChar[] */
 221             ds->swapArray16(ds, inBytes+header.contractionIndex, header.contractionSize*2,
 222                                outBytes+header.contractionIndex, pErrorCode);
 223
 224             /* contractionCEs: CEs[] */
 225             ds->swapArray32(ds, inBytes+header.contractionCEs, header.contractionSize*4,
 226                                outBytes+header.contractionCEs, pErrorCode);
 227         }
 228
 229         /* swap the main trie */
 230         if(header.mappingPosition!=0) {
 231             count=header.endExpansionCE-header.mappingPosition;
 232             utrie_swap(ds, inBytes+header.mappingPosition, (int32_t)count,
 233                           outBytes+header.mappingPosition, pErrorCode);
 234         }
 235
 236         /* swap the max expansion table */
 237         if(header.endExpansionCECount!=0) {
 238             ds->swapArray32(ds, inBytes+header.endExpansionCE, header.endExpansionCECount*4,
 239                                outBytes+header.endExpansionCE, pErrorCode);
 240         }
 241
 242         /* expansionCESize, unsafeCP, contrEndCP: uint8_t[], no need to swap */
 243
 244         /* swap UCA constants */
 245         if(header.UCAConsts!=0) {
 246             /*
 247              * if UCAConsts!=0 then contractionUCACombos because we are swapping
 248              * the UCA data file, and we know that the UCA contains contractions
 249              */
 250             count=header.contractionUCACombos-header.UCAConsts;
 251             ds->swapArray32(ds, inBytes+header.UCAConsts, header.contractionUCACombos-header.UCAConsts,
 252                                outBytes+header.UCAConsts, pErrorCode);
 253         }
 254
 255         /* swap UCA contractions */
 256         if(header.contractionUCACombosSize!=0) {
 257             count=header.contractionUCACombosSize*inHeader->contractionUCACombosWidth*U_SIZEOF_UCHAR;
 258             ds->swapArray16(ds, inBytes+header.contractionUCACombos, (int32_t)count,
 259                                outBytes+header.contractionUCACombos, pErrorCode);
 260         }
 261     }
 262
 263     return header.size;
 264 }
 265
 266 /* swap ICU collation data like ucadata.icu */
 267 U_CAPI int32_t U_EXPORT2
 268 ucol_swap(const UDataSwapper *ds,
 269           const void *inData, int32_t length, void *outData,
 270           UErrorCode *pErrorCode) {
 271     const UDataInfo *pInfo;
 272     int32_t headerSize, collationSize;
 273
 274     /* udata_swapDataHeader checks the arguments */
 275     headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
 276     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 277         return 0;
 278     }
 279
 280     /* check data format and format version */
 281     pInfo=(const UDataInfo *)((const char *)inData+4);
 282     if(!(
 283         pInfo->dataFormat[0]==0x55 &&   /* dataFormat="UCol" */
 284         pInfo->dataFormat[1]==0x43 &&
 285         pInfo->dataFormat[2]==0x6f &&
 286         pInfo->dataFormat[3]==0x6c &&
 287         pInfo->formatVersion[0]==2 &&
 288         pInfo->formatVersion[1]>=3
 289     )) {
 290         udata_printError(ds, "ucol_swap(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not a collation file\n",
 291                          pInfo->dataFormat[0], pInfo->dataFormat[1],
 292                          pInfo->dataFormat[2], pInfo->dataFormat[3],
 293                          pInfo->formatVersion[0], pInfo->formatVersion[1]);
 294         *pErrorCode=U_UNSUPPORTED_ERROR;
 295         return 0;
 296     }
 297
 298     collationSize=ucol_swapBinary(ds,
 299                         (const char *)inData+headerSize,
 300                         length>=0 ? length-headerSize : -1,
 301                         (char *)outData+headerSize,
 302                         pErrorCode);
 303     if(U_SUCCESS(*pErrorCode)) {
 304         return headerSize+collationSize;
 305     } else {
 306         return 0;
 307     }
 308 }
 309
 310 /* swap inverse UCA collation data (invuca.icu) */
 311 U_CAPI int32_t U_EXPORT2
 312 ucol_swapInverseUCA(const UDataSwapper *ds,
 313                     const void *inData, int32_t length, void *outData,
 314                     UErrorCode *pErrorCode) {
 315     const UDataInfo *pInfo;
 316     int32_t headerSize;
 317
 318     const uint8_t *inBytes;
 319     uint8_t *outBytes;
 320
 321     const InverseUCATableHeader *inHeader;
 322     InverseUCATableHeader *outHeader;
 323     InverseUCATableHeader header={ 0 };
 324
 325     /* udata_swapDataHeader checks the arguments */
 326     headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
 327     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 328         return 0;
 329     }
 330
 331     /* check data format and format version */
 332     pInfo=(const UDataInfo *)((const char *)inData+4);
 333     if(!(
 334         pInfo->dataFormat[0]==0x49 &&   /* dataFormat="InvC" */
 335         pInfo->dataFormat[1]==0x6e &&
 336         pInfo->dataFormat[2]==0x76 &&
 337         pInfo->dataFormat[3]==0x43 &&
 338         pInfo->formatVersion[0]==2 &&
 339         pInfo->formatVersion[1]>=1
 340     )) {
 341         udata_printError(ds, "ucol_swapInverseUCA(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not an inverse UCA collation file\n",
 342                          pInfo->dataFormat[0], pInfo->dataFormat[1],
 343                          pInfo->dataFormat[2], pInfo->dataFormat[3],
 344                          pInfo->formatVersion[0], pInfo->formatVersion[1]);
 345         *pErrorCode=U_UNSUPPORTED_ERROR;
 346         return 0;
 347     }
 348
 349     inBytes=(const uint8_t *)inData+headerSize;
 350     outBytes=(uint8_t *)outData+headerSize;
 351
 352     inHeader=(const InverseUCATableHeader *)inBytes;
 353     outHeader=(InverseUCATableHeader *)outBytes;
 354
 355     /*
 356      * The inverse UCA collation binary must contain at least the InverseUCATableHeader,
 357      * starting with its size field.
 358      * sizeof(UCATableHeader)==8*4 in ICU 2.8
 359      * check the length against the header size before reading the size field
 360      */
 361     if(length<0) {
 362         header.byteSize=udata_readInt32(ds, inHeader->byteSize);
 363     } else if(
 364         ((length-headerSize)<(8*4) ||
 365          (uint32_t)(length-headerSize)<(header.byteSize=udata_readInt32(ds, inHeader->byteSize)))
 366     ) {
 367         udata_printError(ds, "ucol_swapInverseUCA(): too few bytes (%d after header) for inverse UCA collation data\n",
 368                          length);
 369         *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
 370         return 0;
 371     }
 372
 373     if(length>=0) {
 374         /* copy everything, takes care of data that needs no swapping */
 375         if(inBytes!=outBytes) {
 376             uprv_memcpy(outBytes, inBytes, header.byteSize);
 377         }
 378
 379         /* swap the necessary pieces in the order of their occurrence in the data */
 380
 381         /* read more of the InverseUCATableHeader (the byteSize field was read above) */
 382         header.tableSize=   ds->readUInt32(inHeader->tableSize);
 383         header.contsSize=   ds->readUInt32(inHeader->contsSize);
 384         header.table=       ds->readUInt32(inHeader->table);
 385         header.conts=       ds->readUInt32(inHeader->conts);
 386
 387         /* swap the 32-bit integers in the header */
 388         ds->swapArray32(ds, inHeader, 5*4, outHeader, pErrorCode);
 389
 390         /* swap the inverse table; tableSize counts uint32_t[3] rows */
 391         ds->swapArray32(ds, inBytes+header.table, header.tableSize*3*4,
 392                            outBytes+header.table, pErrorCode);
 393
 394         /* swap the continuation table; contsSize counts UChars */
 395         ds->swapArray16(ds, inBytes+header.conts, header.contsSize*U_SIZEOF_UCHAR,
 396                            outBytes+header.conts, pErrorCode);
 397     }
 398
 399     return headerSize+header.byteSize;
 400 }
 401
 402 #endif /* #if !UCONFIG_NO_COLLATION */