1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/ucol_swp.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,467 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 2003-2012, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: ucol_swp.cpp
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 2003sep10
1.17 +* created by: Markus W. Scherer
1.18 +*
1.19 +* Swap collation binaries.
1.20 +*/
1.21 +
1.22 +#include "unicode/udata.h" /* UDataInfo */
1.23 +#include "utrie.h"
1.24 +#include "udataswp.h"
1.25 +#include "cmemory.h"
1.26 +#include "ucol_data.h"
1.27 +#include "ucol_swp.h"
1.28 +
1.29 +/* swapping ----------------------------------------------------------------- */
1.30 +
1.31 +/*
1.32 + * This performs data swapping for a folded trie (see utrie.c for details).
1.33 + */
1.34 +
1.35 +U_CAPI int32_t U_EXPORT2
1.36 +utrie_swap(const UDataSwapper *ds,
1.37 + const void *inData, int32_t length, void *outData,
1.38 + UErrorCode *pErrorCode) {
1.39 + const UTrieHeader *inTrie;
1.40 + UTrieHeader trie;
1.41 + int32_t size;
1.42 + UBool dataIs32;
1.43 +
1.44 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.45 + return 0;
1.46 + }
1.47 + if(ds==NULL || inData==NULL || (length>=0 && outData==NULL)) {
1.48 + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1.49 + return 0;
1.50 + }
1.51 +
1.52 + /* setup and swapping */
1.53 + if(length>=0 && (uint32_t)length<sizeof(UTrieHeader)) {
1.54 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.55 + return 0;
1.56 + }
1.57 +
1.58 + inTrie=(const UTrieHeader *)inData;
1.59 + trie.signature=ds->readUInt32(inTrie->signature);
1.60 + trie.options=ds->readUInt32(inTrie->options);
1.61 + trie.indexLength=udata_readInt32(ds, inTrie->indexLength);
1.62 + trie.dataLength=udata_readInt32(ds, inTrie->dataLength);
1.63 +
1.64 + if( trie.signature!=0x54726965 ||
1.65 + (trie.options&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_SHIFT ||
1.66 + ((trie.options>>UTRIE_OPTIONS_INDEX_SHIFT)&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_INDEX_SHIFT ||
1.67 + trie.indexLength<UTRIE_BMP_INDEX_LENGTH ||
1.68 + (trie.indexLength&(UTRIE_SURROGATE_BLOCK_COUNT-1))!=0 ||
1.69 + trie.dataLength<UTRIE_DATA_BLOCK_LENGTH ||
1.70 + (trie.dataLength&(UTRIE_DATA_GRANULARITY-1))!=0 ||
1.71 + ((trie.options&UTRIE_OPTIONS_LATIN1_IS_LINEAR)!=0 && trie.dataLength<(UTRIE_DATA_BLOCK_LENGTH+0x100))
1.72 + ) {
1.73 + *pErrorCode=U_INVALID_FORMAT_ERROR; /* not a UTrie */
1.74 + return 0;
1.75 + }
1.76 +
1.77 + dataIs32=(UBool)((trie.options&UTRIE_OPTIONS_DATA_IS_32_BIT)!=0);
1.78 + size=sizeof(UTrieHeader)+trie.indexLength*2+trie.dataLength*(dataIs32?4:2);
1.79 +
1.80 + if(length>=0) {
1.81 + UTrieHeader *outTrie;
1.82 +
1.83 + if(length<size) {
1.84 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.85 + return 0;
1.86 + }
1.87 +
1.88 + outTrie=(UTrieHeader *)outData;
1.89 +
1.90 + /* swap the header */
1.91 + ds->swapArray32(ds, inTrie, sizeof(UTrieHeader), outTrie, pErrorCode);
1.92 +
1.93 + /* swap the index and the data */
1.94 + if(dataIs32) {
1.95 + ds->swapArray16(ds, inTrie+1, trie.indexLength*2, outTrie+1, pErrorCode);
1.96 + ds->swapArray32(ds, (const uint16_t *)(inTrie+1)+trie.indexLength, trie.dataLength*4,
1.97 + (uint16_t *)(outTrie+1)+trie.indexLength, pErrorCode);
1.98 + } else {
1.99 + ds->swapArray16(ds, inTrie+1, (trie.indexLength+trie.dataLength)*2, outTrie+1, pErrorCode);
1.100 + }
1.101 + }
1.102 +
1.103 + return size;
1.104 +}
1.105 +
1.106 +#if !UCONFIG_NO_COLLATION
1.107 +
1.108 +/* Modified copy of the beginning of ucol_swapBinary(). */
1.109 +U_CAPI UBool U_EXPORT2
1.110 +ucol_looksLikeCollationBinary(const UDataSwapper *ds,
1.111 + const void *inData, int32_t length) {
1.112 + const UCATableHeader *inHeader;
1.113 + UCATableHeader header;
1.114 +
1.115 + if(ds==NULL || inData==NULL || length<-1) {
1.116 + return FALSE;
1.117 + }
1.118 +
1.119 + inHeader=(const UCATableHeader *)inData;
1.120 +
1.121 + /*
1.122 + * The collation binary must contain at least the UCATableHeader,
1.123 + * starting with its size field.
1.124 + * sizeof(UCATableHeader)==42*4 in ICU 2.8
1.125 + * check the length against the header size before reading the size field
1.126 + */
1.127 + uprv_memset(&header, 0, sizeof(header));
1.128 + if(length<0) {
1.129 + header.size=udata_readInt32(ds, inHeader->size);
1.130 + } else if((length<(42*4) || length<(header.size=udata_readInt32(ds, inHeader->size)))) {
1.131 + return FALSE;
1.132 + }
1.133 +
1.134 + header.magic=ds->readUInt32(inHeader->magic);
1.135 + if(!(
1.136 + header.magic==UCOL_HEADER_MAGIC &&
1.137 + inHeader->formatVersion[0]==3 /*&&
1.138 + inHeader->formatVersion[1]>=0*/
1.139 + )) {
1.140 + return FALSE;
1.141 + }
1.142 +
1.143 + if(inHeader->isBigEndian!=ds->inIsBigEndian || inHeader->charSetFamily!=ds->inCharset) {
1.144 + return FALSE;
1.145 + }
1.146 +
1.147 + return TRUE;
1.148 +}
1.149 +
1.150 +/* swap a header-less collation binary, inside a resource bundle or ucadata.icu */
1.151 +U_CAPI int32_t U_EXPORT2
1.152 +ucol_swapBinary(const UDataSwapper *ds,
1.153 + const void *inData, int32_t length, void *outData,
1.154 + UErrorCode *pErrorCode) {
1.155 + const uint8_t *inBytes;
1.156 + uint8_t *outBytes;
1.157 +
1.158 + const UCATableHeader *inHeader;
1.159 + UCATableHeader *outHeader;
1.160 + UCATableHeader header;
1.161 +
1.162 + uint32_t count;
1.163 +
1.164 + /* argument checking in case we were not called from ucol_swap() */
1.165 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.166 + return 0;
1.167 + }
1.168 + if(ds==NULL || inData==NULL || length<-1 || (length>0 && outData==NULL)) {
1.169 + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1.170 + return 0;
1.171 + }
1.172 +
1.173 + inBytes=(const uint8_t *)inData;
1.174 + outBytes=(uint8_t *)outData;
1.175 +
1.176 + inHeader=(const UCATableHeader *)inData;
1.177 + outHeader=(UCATableHeader *)outData;
1.178 +
1.179 + /*
1.180 + * The collation binary must contain at least the UCATableHeader,
1.181 + * starting with its size field.
1.182 + * sizeof(UCATableHeader)==42*4 in ICU 2.8
1.183 + * check the length against the header size before reading the size field
1.184 + */
1.185 + uprv_memset(&header, 0, sizeof(header));
1.186 + if(length<0) {
1.187 + header.size=udata_readInt32(ds, inHeader->size);
1.188 + } else if((length<(42*4) || length<(header.size=udata_readInt32(ds, inHeader->size)))) {
1.189 + udata_printError(ds, "ucol_swapBinary(): too few bytes (%d after header) for collation data\n",
1.190 + length);
1.191 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.192 + return 0;
1.193 + }
1.194 +
1.195 + header.magic=ds->readUInt32(inHeader->magic);
1.196 + if(!(
1.197 + header.magic==UCOL_HEADER_MAGIC &&
1.198 + inHeader->formatVersion[0]==3 /*&&
1.199 + inHeader->formatVersion[1]>=0*/
1.200 + )) {
1.201 + udata_printError(ds, "ucol_swapBinary(): magic 0x%08x or format version %02x.%02x is not a collation binary\n",
1.202 + header.magic,
1.203 + inHeader->formatVersion[0], inHeader->formatVersion[1]);
1.204 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.205 + return 0;
1.206 + }
1.207 +
1.208 + if(inHeader->isBigEndian!=ds->inIsBigEndian || inHeader->charSetFamily!=ds->inCharset) {
1.209 + udata_printError(ds, "ucol_swapBinary(): endianness %d or charset %d does not match the swapper\n",
1.210 + inHeader->isBigEndian, inHeader->charSetFamily);
1.211 + *pErrorCode=U_INVALID_FORMAT_ERROR;
1.212 + return 0;
1.213 + }
1.214 +
1.215 + if(length>=0) {
1.216 + /* copy everything, takes care of data that needs no swapping */
1.217 + if(inBytes!=outBytes) {
1.218 + uprv_memcpy(outBytes, inBytes, header.size);
1.219 + }
1.220 +
1.221 + /* swap the necessary pieces in the order of their occurrence in the data */
1.222 +
1.223 + /* read more of the UCATableHeader (the size field was read above) */
1.224 + header.options= ds->readUInt32(inHeader->options);
1.225 + header.UCAConsts= ds->readUInt32(inHeader->UCAConsts);
1.226 + header.contractionUCACombos= ds->readUInt32(inHeader->contractionUCACombos);
1.227 + header.mappingPosition= ds->readUInt32(inHeader->mappingPosition);
1.228 + header.expansion= ds->readUInt32(inHeader->expansion);
1.229 + header.contractionIndex= ds->readUInt32(inHeader->contractionIndex);
1.230 + header.contractionCEs= ds->readUInt32(inHeader->contractionCEs);
1.231 + header.contractionSize= ds->readUInt32(inHeader->contractionSize);
1.232 + header.endExpansionCE= ds->readUInt32(inHeader->endExpansionCE);
1.233 + header.expansionCESize= ds->readUInt32(inHeader->expansionCESize);
1.234 + header.endExpansionCECount= udata_readInt32(ds, inHeader->endExpansionCECount);
1.235 + header.contractionUCACombosSize=udata_readInt32(ds, inHeader->contractionUCACombosSize);
1.236 + header.scriptToLeadByte= ds->readUInt32(inHeader->scriptToLeadByte);
1.237 + header.leadByteToScript= ds->readUInt32(inHeader->leadByteToScript);
1.238 +
1.239 + /* swap the 32-bit integers in the header */
1.240 + ds->swapArray32(ds, inHeader, (int32_t)((const char *)&inHeader->jamoSpecial-(const char *)inHeader),
1.241 + outHeader, pErrorCode);
1.242 + ds->swapArray32(ds, &(inHeader->scriptToLeadByte), sizeof(header.scriptToLeadByte) + sizeof(header.leadByteToScript),
1.243 + &(outHeader->scriptToLeadByte), pErrorCode);
1.244 + /* set the output platform properties */
1.245 + outHeader->isBigEndian=ds->outIsBigEndian;
1.246 + outHeader->charSetFamily=ds->outCharset;
1.247 +
1.248 + /* swap the options */
1.249 + if(header.options!=0) {
1.250 + ds->swapArray32(ds, inBytes+header.options, header.expansion-header.options,
1.251 + outBytes+header.options, pErrorCode);
1.252 + }
1.253 +
1.254 + /* swap the expansions */
1.255 + if(header.mappingPosition!=0 && header.expansion!=0) {
1.256 + if(header.contractionIndex!=0) {
1.257 + /* expansions bounded by contractions */
1.258 + count=header.contractionIndex-header.expansion;
1.259 + } else {
1.260 + /* no contractions: expansions bounded by the main trie */
1.261 + count=header.mappingPosition-header.expansion;
1.262 + }
1.263 + ds->swapArray32(ds, inBytes+header.expansion, (int32_t)count,
1.264 + outBytes+header.expansion, pErrorCode);
1.265 + }
1.266 +
1.267 + /* swap the contractions */
1.268 + if(header.contractionSize!=0) {
1.269 + /* contractionIndex: UChar[] */
1.270 + ds->swapArray16(ds, inBytes+header.contractionIndex, header.contractionSize*2,
1.271 + outBytes+header.contractionIndex, pErrorCode);
1.272 +
1.273 + /* contractionCEs: CEs[] */
1.274 + ds->swapArray32(ds, inBytes+header.contractionCEs, header.contractionSize*4,
1.275 + outBytes+header.contractionCEs, pErrorCode);
1.276 + }
1.277 +
1.278 + /* swap the main trie */
1.279 + if(header.mappingPosition!=0) {
1.280 + count=header.endExpansionCE-header.mappingPosition;
1.281 + utrie_swap(ds, inBytes+header.mappingPosition, (int32_t)count,
1.282 + outBytes+header.mappingPosition, pErrorCode);
1.283 + }
1.284 +
1.285 + /* swap the max expansion table */
1.286 + if(header.endExpansionCECount!=0) {
1.287 + ds->swapArray32(ds, inBytes+header.endExpansionCE, header.endExpansionCECount*4,
1.288 + outBytes+header.endExpansionCE, pErrorCode);
1.289 + }
1.290 +
1.291 + /* expansionCESize, unsafeCP, contrEndCP: uint8_t[], no need to swap */
1.292 +
1.293 + /* swap UCA constants */
1.294 + if(header.UCAConsts!=0) {
1.295 + /*
1.296 + * if UCAConsts!=0 then contractionUCACombos because we are swapping
1.297 + * the UCA data file, and we know that the UCA contains contractions
1.298 + */
1.299 + count=header.contractionUCACombos-header.UCAConsts;
1.300 + ds->swapArray32(ds, inBytes+header.UCAConsts, header.contractionUCACombos-header.UCAConsts,
1.301 + outBytes+header.UCAConsts, pErrorCode);
1.302 + }
1.303 +
1.304 + /* swap UCA contractions */
1.305 + if(header.contractionUCACombosSize!=0) {
1.306 + count=header.contractionUCACombosSize*inHeader->contractionUCACombosWidth*U_SIZEOF_UCHAR;
1.307 + ds->swapArray16(ds, inBytes+header.contractionUCACombos, (int32_t)count,
1.308 + outBytes+header.contractionUCACombos, pErrorCode);
1.309 + }
1.310 +
1.311 + /* swap the script to lead bytes */
1.312 + if(header.scriptToLeadByte!=0) {
1.313 + int indexCount = ds->readUInt16(*((uint16_t*)(inBytes+header.scriptToLeadByte))); // each entry = 2 * uint16
1.314 + int dataCount = ds->readUInt16(*((uint16_t*)(inBytes+header.scriptToLeadByte + 2))); // each entry = uint16
1.315 + ds->swapArray16(ds, inBytes+header.scriptToLeadByte,
1.316 + 4 + (4 * indexCount) + (2 * dataCount),
1.317 + outBytes+header.scriptToLeadByte, pErrorCode);
1.318 + }
1.319 +
1.320 + /* swap the lead byte to scripts */
1.321 + if(header.leadByteToScript!=0) {
1.322 + int indexCount = ds->readUInt16(*((uint16_t*)(inBytes+header.leadByteToScript))); // each entry = uint16
1.323 + int dataCount = ds->readUInt16(*((uint16_t*)(inBytes+header.leadByteToScript + 2))); // each entry = uint16
1.324 + ds->swapArray16(ds, inBytes+header.leadByteToScript,
1.325 + 4 + (2 * indexCount) + (2 * dataCount),
1.326 + outBytes+header.leadByteToScript, pErrorCode);
1.327 + }
1.328 + }
1.329 +
1.330 + return header.size;
1.331 +}
1.332 +
1.333 +/* swap ICU collation data like ucadata.icu */
1.334 +U_CAPI int32_t U_EXPORT2
1.335 +ucol_swap(const UDataSwapper *ds,
1.336 + const void *inData, int32_t length, void *outData,
1.337 + UErrorCode *pErrorCode) {
1.338 +
1.339 + const UDataInfo *pInfo;
1.340 + int32_t headerSize, collationSize;
1.341 +
1.342 + /* udata_swapDataHeader checks the arguments */
1.343 + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1.344 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.345 + return 0;
1.346 + }
1.347 +
1.348 + /* check data format and format version */
1.349 + pInfo=(const UDataInfo *)((const char *)inData+4);
1.350 + if(!(
1.351 + pInfo->dataFormat[0]==0x55 && /* dataFormat="UCol" */
1.352 + pInfo->dataFormat[1]==0x43 &&
1.353 + pInfo->dataFormat[2]==0x6f &&
1.354 + pInfo->dataFormat[3]==0x6c &&
1.355 + pInfo->formatVersion[0]==3 /*&&
1.356 + pInfo->formatVersion[1]>=0*/
1.357 + )) {
1.358 + udata_printError(ds, "ucol_swap(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not a collation file\n",
1.359 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.360 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.361 + pInfo->formatVersion[0], pInfo->formatVersion[1]);
1.362 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.363 + return 0;
1.364 + }
1.365 +
1.366 + collationSize=ucol_swapBinary(ds,
1.367 + (const char *)inData+headerSize,
1.368 + length>=0 ? length-headerSize : -1,
1.369 + (char *)outData+headerSize,
1.370 + pErrorCode);
1.371 + if(U_SUCCESS(*pErrorCode)) {
1.372 + return headerSize+collationSize;
1.373 + } else {
1.374 + return 0;
1.375 + }
1.376 +}
1.377 +
1.378 +/* swap inverse UCA collation data (invuca.icu) */
1.379 +U_CAPI int32_t U_EXPORT2
1.380 +ucol_swapInverseUCA(const UDataSwapper *ds,
1.381 + const void *inData, int32_t length, void *outData,
1.382 + UErrorCode *pErrorCode) {
1.383 + const UDataInfo *pInfo;
1.384 + int32_t headerSize;
1.385 +
1.386 + const uint8_t *inBytes;
1.387 + uint8_t *outBytes;
1.388 +
1.389 + const InverseUCATableHeader *inHeader;
1.390 + InverseUCATableHeader *outHeader;
1.391 + InverseUCATableHeader header={ 0,0,0,0,0,{0,0,0,0},{0,0,0,0,0,0,0,0} };
1.392 +
1.393 + /* udata_swapDataHeader checks the arguments */
1.394 + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1.395 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.396 + return 0;
1.397 + }
1.398 +
1.399 + /* check data format and format version */
1.400 + pInfo=(const UDataInfo *)((const char *)inData+4);
1.401 + if(!(
1.402 + pInfo->dataFormat[0]==0x49 && /* dataFormat="InvC" */
1.403 + pInfo->dataFormat[1]==0x6e &&
1.404 + pInfo->dataFormat[2]==0x76 &&
1.405 + pInfo->dataFormat[3]==0x43 &&
1.406 + pInfo->formatVersion[0]==2 &&
1.407 + pInfo->formatVersion[1]>=1
1.408 + )) {
1.409 + udata_printError(ds, "ucol_swapInverseUCA(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not an inverse UCA collation file\n",
1.410 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.411 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.412 + pInfo->formatVersion[0], pInfo->formatVersion[1]);
1.413 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.414 + return 0;
1.415 + }
1.416 +
1.417 + inBytes=(const uint8_t *)inData+headerSize;
1.418 + outBytes=(uint8_t *)outData+headerSize;
1.419 +
1.420 + inHeader=(const InverseUCATableHeader *)inBytes;
1.421 + outHeader=(InverseUCATableHeader *)outBytes;
1.422 +
1.423 + /*
1.424 + * The inverse UCA collation binary must contain at least the InverseUCATableHeader,
1.425 + * starting with its size field.
1.426 + * sizeof(UCATableHeader)==8*4 in ICU 2.8
1.427 + * check the length against the header size before reading the size field
1.428 + */
1.429 + if(length<0) {
1.430 + header.byteSize=udata_readInt32(ds, inHeader->byteSize);
1.431 + } else if(
1.432 + ((length-headerSize)<(8*4) ||
1.433 + (uint32_t)(length-headerSize)<(header.byteSize=udata_readInt32(ds, inHeader->byteSize)))
1.434 + ) {
1.435 + udata_printError(ds, "ucol_swapInverseUCA(): too few bytes (%d after header) for inverse UCA collation data\n",
1.436 + length);
1.437 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.438 + return 0;
1.439 + }
1.440 +
1.441 + if(length>=0) {
1.442 + /* copy everything, takes care of data that needs no swapping */
1.443 + if(inBytes!=outBytes) {
1.444 + uprv_memcpy(outBytes, inBytes, header.byteSize);
1.445 + }
1.446 +
1.447 + /* swap the necessary pieces in the order of their occurrence in the data */
1.448 +
1.449 + /* read more of the InverseUCATableHeader (the byteSize field was read above) */
1.450 + header.tableSize= ds->readUInt32(inHeader->tableSize);
1.451 + header.contsSize= ds->readUInt32(inHeader->contsSize);
1.452 + header.table= ds->readUInt32(inHeader->table);
1.453 + header.conts= ds->readUInt32(inHeader->conts);
1.454 +
1.455 + /* swap the 32-bit integers in the header */
1.456 + ds->swapArray32(ds, inHeader, 5*4, outHeader, pErrorCode);
1.457 +
1.458 + /* swap the inverse table; tableSize counts uint32_t[3] rows */
1.459 + ds->swapArray32(ds, inBytes+header.table, header.tableSize*3*4,
1.460 + outBytes+header.table, pErrorCode);
1.461 +
1.462 + /* swap the continuation table; contsSize counts UChars */
1.463 + ds->swapArray16(ds, inBytes+header.conts, header.contsSize*U_SIZEOF_UCHAR,
1.464 + outBytes+header.conts, pErrorCode);
1.465 + }
1.466 +
1.467 + return headerSize+header.byteSize;
1.468 +}
1.469 +
1.470 +#endif /* #if !UCONFIG_NO_COLLATION */
