Thu, 22 Jan 2015 13:21:57 +0100
Incorporate requested changes from Mozilla in review:
https://bugzilla.mozilla.org/show_bug.cgi?id=1123480#c6
michael@0 | 1 | /* |
michael@0 | 2 | *************************************************************************** |
michael@0 | 3 | * Copyright (C) 1999-2010 International Business Machines Corporation * |
michael@0 | 4 | * and others. All rights reserved. * |
michael@0 | 5 | *************************************************************************** |
michael@0 | 6 | */ |
michael@0 | 7 | |
michael@0 | 8 | #include "unicode/utypes.h" |
michael@0 | 9 | |
michael@0 | 10 | #if !UCONFIG_NO_BREAK_ITERATION |
michael@0 | 11 | |
michael@0 | 12 | #include "unicode/utypes.h" |
michael@0 | 13 | #include "rbbidata.h" |
michael@0 | 14 | #include "rbbirb.h" |
michael@0 | 15 | #include "utrie.h" |
michael@0 | 16 | #include "udatamem.h" |
michael@0 | 17 | #include "cmemory.h" |
michael@0 | 18 | #include "cstring.h" |
michael@0 | 19 | #include "umutex.h" |
michael@0 | 20 | |
michael@0 | 21 | #include "uassert.h" |
michael@0 | 22 | |
michael@0 | 23 | |
michael@0 | 24 | //----------------------------------------------------------------------------------- |
michael@0 | 25 | // |
michael@0 | 26 | // Trie access folding function. Copied as-is from properties code in uchar.c |
michael@0 | 27 | // |
michael@0 | 28 | //----------------------------------------------------------------------------------- |
michael@0 | 29 | U_CDECL_BEGIN |
michael@0 | 30 | static int32_t U_CALLCONV |
michael@0 | 31 | getFoldingOffset(uint32_t data) { |
michael@0 | 32 | /* if bit 15 is set, then the folding offset is in bits 14..0 of the 16-bit trie result */ |
michael@0 | 33 | if(data&0x8000) { |
michael@0 | 34 | return (int32_t)(data&0x7fff); |
michael@0 | 35 | } else { |
michael@0 | 36 | return 0; |
michael@0 | 37 | } |
michael@0 | 38 | } |
michael@0 | 39 | U_CDECL_END |
michael@0 | 40 | |
michael@0 | 41 | U_NAMESPACE_BEGIN |
michael@0 | 42 | |
michael@0 | 43 | //----------------------------------------------------------------------------- |
michael@0 | 44 | // |
michael@0 | 45 | // Constructors. |
michael@0 | 46 | // |
michael@0 | 47 | //----------------------------------------------------------------------------- |
michael@0 | 48 | RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status) { |
michael@0 | 49 | init(data, status); |
michael@0 | 50 | } |
michael@0 | 51 | |
michael@0 | 52 | RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt, UErrorCode &status) { |
michael@0 | 53 | init(data, status); |
michael@0 | 54 | fDontFreeData = TRUE; |
michael@0 | 55 | } |
michael@0 | 56 | |
michael@0 | 57 | RBBIDataWrapper::RBBIDataWrapper(UDataMemory* udm, UErrorCode &status) { |
michael@0 | 58 | const RBBIDataHeader *d = (const RBBIDataHeader *) |
michael@0 | 59 | // ((char *)&(udm->pHeader->info) + udm->pHeader->info.size); |
michael@0 | 60 | // taking into consideration the padding added in by udata_write |
michael@0 | 61 | ((char *)(udm->pHeader) + udm->pHeader->dataHeader.headerSize); |
michael@0 | 62 | init(d, status); |
michael@0 | 63 | fUDataMem = udm; |
michael@0 | 64 | } |
michael@0 | 65 | |
michael@0 | 66 | //----------------------------------------------------------------------------- |
michael@0 | 67 | // |
michael@0 | 68 | // init(). Does most of the work of construction, shared between the |
michael@0 | 69 | // constructors. |
michael@0 | 70 | // |
michael@0 | 71 | //----------------------------------------------------------------------------- |
michael@0 | 72 | void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) { |
michael@0 | 73 | if (U_FAILURE(status)) { |
michael@0 | 74 | return; |
michael@0 | 75 | } |
michael@0 | 76 | fHeader = data; |
michael@0 | 77 | if (fHeader->fMagic != 0xb1a0 || fHeader->fFormatVersion[0] != 3) |
michael@0 | 78 | { |
michael@0 | 79 | status = U_INVALID_FORMAT_ERROR; |
michael@0 | 80 | return; |
michael@0 | 81 | } |
michael@0 | 82 | // Note: in ICU version 3.2 and earlier, there was a formatVersion 1 |
michael@0 | 83 | // that is no longer supported. At that time fFormatVersion was |
michael@0 | 84 | // an int32_t field, rather than an array of 4 bytes. |
michael@0 | 85 | |
michael@0 | 86 | fDontFreeData = FALSE; |
michael@0 | 87 | fUDataMem = NULL; |
michael@0 | 88 | fReverseTable = NULL; |
michael@0 | 89 | fSafeFwdTable = NULL; |
michael@0 | 90 | fSafeRevTable = NULL; |
michael@0 | 91 | if (data->fFTableLen != 0) { |
michael@0 | 92 | fForwardTable = (RBBIStateTable *)((char *)data + fHeader->fFTable); |
michael@0 | 93 | } |
michael@0 | 94 | if (data->fRTableLen != 0) { |
michael@0 | 95 | fReverseTable = (RBBIStateTable *)((char *)data + fHeader->fRTable); |
michael@0 | 96 | } |
michael@0 | 97 | if (data->fSFTableLen != 0) { |
michael@0 | 98 | fSafeFwdTable = (RBBIStateTable *)((char *)data + fHeader->fSFTable); |
michael@0 | 99 | } |
michael@0 | 100 | if (data->fSRTableLen != 0) { |
michael@0 | 101 | fSafeRevTable = (RBBIStateTable *)((char *)data + fHeader->fSRTable); |
michael@0 | 102 | } |
michael@0 | 103 | |
michael@0 | 104 | |
michael@0 | 105 | utrie_unserialize(&fTrie, |
michael@0 | 106 | (uint8_t *)data + fHeader->fTrie, |
michael@0 | 107 | fHeader->fTrieLen, |
michael@0 | 108 | &status); |
michael@0 | 109 | if (U_FAILURE(status)) { |
michael@0 | 110 | return; |
michael@0 | 111 | } |
michael@0 | 112 | fTrie.getFoldingOffset=getFoldingOffset; |
michael@0 | 113 | |
michael@0 | 114 | |
michael@0 | 115 | fRuleSource = (UChar *)((char *)data + fHeader->fRuleSource); |
michael@0 | 116 | fRuleString.setTo(TRUE, fRuleSource, -1); |
michael@0 | 117 | U_ASSERT(data->fRuleSourceLen > 0); |
michael@0 | 118 | |
michael@0 | 119 | fRuleStatusTable = (int32_t *)((char *)data + fHeader->fStatusTable); |
michael@0 | 120 | fStatusMaxIdx = data->fStatusTableLen / sizeof(int32_t); |
michael@0 | 121 | |
michael@0 | 122 | fRefCount = 1; |
michael@0 | 123 | |
michael@0 | 124 | #ifdef RBBI_DEBUG |
michael@0 | 125 | char *debugEnv = getenv("U_RBBIDEBUG"); |
michael@0 | 126 | if (debugEnv && uprv_strstr(debugEnv, "data")) {this->printData();} |
michael@0 | 127 | #endif |
michael@0 | 128 | } |
michael@0 | 129 | |
michael@0 | 130 | |
michael@0 | 131 | //----------------------------------------------------------------------------- |
michael@0 | 132 | // |
michael@0 | 133 | // Destructor. Don't call this - use removeReference() instead. |
michael@0 | 134 | // |
michael@0 | 135 | //----------------------------------------------------------------------------- |
michael@0 | 136 | RBBIDataWrapper::~RBBIDataWrapper() { |
michael@0 | 137 | U_ASSERT(fRefCount == 0); |
michael@0 | 138 | if (fUDataMem) { |
michael@0 | 139 | udata_close(fUDataMem); |
michael@0 | 140 | } else if (!fDontFreeData) { |
michael@0 | 141 | uprv_free((void *)fHeader); |
michael@0 | 142 | } |
michael@0 | 143 | } |
michael@0 | 144 | |
michael@0 | 145 | |
michael@0 | 146 | |
michael@0 | 147 | //----------------------------------------------------------------------------- |
michael@0 | 148 | // |
michael@0 | 149 | // Operator == Consider two RBBIDataWrappers to be equal if they |
michael@0 | 150 | // refer to the same underlying data. Although |
michael@0 | 151 | // the data wrappers are normally shared between |
michael@0 | 152 | // iterator instances, it's possible to independently |
michael@0 | 153 | // open the same data twice, and get two instances, which |
michael@0 | 154 | // should still be ==. |
michael@0 | 155 | // |
michael@0 | 156 | //----------------------------------------------------------------------------- |
michael@0 | 157 | UBool RBBIDataWrapper::operator ==(const RBBIDataWrapper &other) const { |
michael@0 | 158 | if (fHeader == other.fHeader) { |
michael@0 | 159 | return TRUE; |
michael@0 | 160 | } |
michael@0 | 161 | if (fHeader->fLength != other.fHeader->fLength) { |
michael@0 | 162 | return FALSE; |
michael@0 | 163 | } |
michael@0 | 164 | if (uprv_memcmp(fHeader, other.fHeader, fHeader->fLength) == 0) { |
michael@0 | 165 | return TRUE; |
michael@0 | 166 | } |
michael@0 | 167 | return FALSE; |
michael@0 | 168 | } |
michael@0 | 169 | |
michael@0 | 170 | int32_t RBBIDataWrapper::hashCode() { |
michael@0 | 171 | return fHeader->fFTableLen; |
michael@0 | 172 | } |
michael@0 | 173 | |
michael@0 | 174 | |
michael@0 | 175 | |
michael@0 | 176 | //----------------------------------------------------------------------------- |
michael@0 | 177 | // |
michael@0 | 178 | // Reference Counting. A single RBBIDataWrapper object is shared among |
michael@0 | 179 | // however many RulesBasedBreakIterator instances are |
michael@0 | 180 | // referencing the same data. |
michael@0 | 181 | // |
michael@0 | 182 | //----------------------------------------------------------------------------- |
michael@0 | 183 | void RBBIDataWrapper::removeReference() { |
michael@0 | 184 | if (umtx_atomic_dec(&fRefCount) == 0) { |
michael@0 | 185 | delete this; |
michael@0 | 186 | } |
michael@0 | 187 | } |
michael@0 | 188 | |
michael@0 | 189 | |
michael@0 | 190 | RBBIDataWrapper *RBBIDataWrapper::addReference() { |
michael@0 | 191 | umtx_atomic_inc(&fRefCount); |
michael@0 | 192 | return this; |
michael@0 | 193 | } |
michael@0 | 194 | |
michael@0 | 195 | |
michael@0 | 196 | |
michael@0 | 197 | //----------------------------------------------------------------------------- |
michael@0 | 198 | // |
michael@0 | 199 | // getRuleSourceString |
michael@0 | 200 | // |
michael@0 | 201 | //----------------------------------------------------------------------------- |
michael@0 | 202 | const UnicodeString &RBBIDataWrapper::getRuleSourceString() const { |
michael@0 | 203 | return fRuleString; |
michael@0 | 204 | } |
michael@0 | 205 | |
michael@0 | 206 | |
michael@0 | 207 | //----------------------------------------------------------------------------- |
michael@0 | 208 | // |
michael@0 | 209 | // print - debugging function to dump the runtime data tables. |
michael@0 | 210 | // |
michael@0 | 211 | //----------------------------------------------------------------------------- |
michael@0 | 212 | #ifdef RBBI_DEBUG |
michael@0 | 213 | void RBBIDataWrapper::printTable(const char *heading, const RBBIStateTable *table) { |
michael@0 | 214 | uint32_t c; |
michael@0 | 215 | uint32_t s; |
michael@0 | 216 | |
michael@0 | 217 | RBBIDebugPrintf(" %s\n", heading); |
michael@0 | 218 | |
michael@0 | 219 | RBBIDebugPrintf("State | Acc LA TagIx"); |
michael@0 | 220 | for (c=0; c<fHeader->fCatCount; c++) {RBBIDebugPrintf("%3d ", c);} |
michael@0 | 221 | RBBIDebugPrintf("\n------|---------------"); for (c=0;c<fHeader->fCatCount; c++) { |
michael@0 | 222 | RBBIDebugPrintf("----"); |
michael@0 | 223 | } |
michael@0 | 224 | RBBIDebugPrintf("\n"); |
michael@0 | 225 | |
michael@0 | 226 | if (table == NULL) { |
michael@0 | 227 | RBBIDebugPrintf(" N U L L T A B L E\n\n"); |
michael@0 | 228 | return; |
michael@0 | 229 | } |
michael@0 | 230 | for (s=0; s<table->fNumStates; s++) { |
michael@0 | 231 | RBBIStateTableRow *row = (RBBIStateTableRow *) |
michael@0 | 232 | (table->fTableData + (table->fRowLen * s)); |
michael@0 | 233 | RBBIDebugPrintf("%4d | %3d %3d %3d ", s, row->fAccepting, row->fLookAhead, row->fTagIdx); |
michael@0 | 234 | for (c=0; c<fHeader->fCatCount; c++) { |
michael@0 | 235 | RBBIDebugPrintf("%3d ", row->fNextState[c]); |
michael@0 | 236 | } |
michael@0 | 237 | RBBIDebugPrintf("\n"); |
michael@0 | 238 | } |
michael@0 | 239 | RBBIDebugPrintf("\n"); |
michael@0 | 240 | } |
michael@0 | 241 | #endif |
michael@0 | 242 | |
michael@0 | 243 | |
michael@0 | 244 | #ifdef RBBI_DEBUG |
michael@0 | 245 | void RBBIDataWrapper::printData() { |
michael@0 | 246 | RBBIDebugPrintf("RBBI Data at %p\n", (void *)fHeader); |
michael@0 | 247 | RBBIDebugPrintf(" Version = {%d %d %d %d}\n", fHeader->fFormatVersion[0], fHeader->fFormatVersion[1], |
michael@0 | 248 | fHeader->fFormatVersion[2], fHeader->fFormatVersion[3]); |
michael@0 | 249 | RBBIDebugPrintf(" total length of data = %d\n", fHeader->fLength); |
michael@0 | 250 | RBBIDebugPrintf(" number of character categories = %d\n\n", fHeader->fCatCount); |
michael@0 | 251 | |
michael@0 | 252 | printTable("Forward State Transition Table", fForwardTable); |
michael@0 | 253 | printTable("Reverse State Transition Table", fReverseTable); |
michael@0 | 254 | printTable("Safe Forward State Transition Table", fSafeFwdTable); |
michael@0 | 255 | printTable("Safe Reverse State Transition Table", fSafeRevTable); |
michael@0 | 256 | |
michael@0 | 257 | RBBIDebugPrintf("\nOrignal Rules source:\n"); |
michael@0 | 258 | for (int32_t c=0; fRuleSource[c] != 0; c++) { |
michael@0 | 259 | RBBIDebugPrintf("%c", fRuleSource[c]); |
michael@0 | 260 | } |
michael@0 | 261 | RBBIDebugPrintf("\n\n"); |
michael@0 | 262 | } |
michael@0 | 263 | #endif |
michael@0 | 264 | |
michael@0 | 265 | |
michael@0 | 266 | U_NAMESPACE_END |
michael@0 | 267 | U_NAMESPACE_USE |
michael@0 | 268 | |
michael@0 | 269 | //----------------------------------------------------------------------------- |
michael@0 | 270 | // |
michael@0 | 271 | // ubrk_swap - byte swap and char encoding swap of RBBI data |
michael@0 | 272 | // |
michael@0 | 273 | //----------------------------------------------------------------------------- |
michael@0 | 274 | |
michael@0 | 275 | U_CAPI int32_t U_EXPORT2 |
michael@0 | 276 | ubrk_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData, |
michael@0 | 277 | UErrorCode *status) { |
michael@0 | 278 | |
michael@0 | 279 | if (status == NULL || U_FAILURE(*status)) { |
michael@0 | 280 | return 0; |
michael@0 | 281 | } |
michael@0 | 282 | if(ds==NULL || inData==NULL || length<-1 || (length>0 && outData==NULL)) { |
michael@0 | 283 | *status=U_ILLEGAL_ARGUMENT_ERROR; |
michael@0 | 284 | return 0; |
michael@0 | 285 | } |
michael@0 | 286 | |
michael@0 | 287 | // |
michael@0 | 288 | // Check that the data header is for for break data. |
michael@0 | 289 | // (Header contents are defined in genbrk.cpp) |
michael@0 | 290 | // |
michael@0 | 291 | const UDataInfo *pInfo = (const UDataInfo *)((const char *)inData+4); |
michael@0 | 292 | if(!( pInfo->dataFormat[0]==0x42 && /* dataFormat="Brk " */ |
michael@0 | 293 | pInfo->dataFormat[1]==0x72 && |
michael@0 | 294 | pInfo->dataFormat[2]==0x6b && |
michael@0 | 295 | pInfo->dataFormat[3]==0x20 && |
michael@0 | 296 | pInfo->formatVersion[0]==3 )) { |
michael@0 | 297 | udata_printError(ds, "ubrk_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized\n", |
michael@0 | 298 | pInfo->dataFormat[0], pInfo->dataFormat[1], |
michael@0 | 299 | pInfo->dataFormat[2], pInfo->dataFormat[3], |
michael@0 | 300 | pInfo->formatVersion[0]); |
michael@0 | 301 | *status=U_UNSUPPORTED_ERROR; |
michael@0 | 302 | return 0; |
michael@0 | 303 | } |
michael@0 | 304 | |
michael@0 | 305 | // |
michael@0 | 306 | // Swap the data header. (This is the generic ICU Data Header, not the RBBI Specific |
michael@0 | 307 | // RBBIDataHeader). This swap also conveniently gets us |
michael@0 | 308 | // the size of the ICU d.h., which lets us locate the start |
michael@0 | 309 | // of the RBBI specific data. |
michael@0 | 310 | // |
michael@0 | 311 | int32_t headerSize=udata_swapDataHeader(ds, inData, length, outData, status); |
michael@0 | 312 | |
michael@0 | 313 | |
michael@0 | 314 | // |
michael@0 | 315 | // Get the RRBI Data Header, and check that it appears to be OK. |
michael@0 | 316 | // |
michael@0 | 317 | // Note: ICU 3.2 and earlier, RBBIDataHeader::fDataFormat was actually |
michael@0 | 318 | // an int32_t with a value of 1. Starting with ICU 3.4, |
michael@0 | 319 | // RBBI's fDataFormat matches the dataFormat field from the |
michael@0 | 320 | // UDataInfo header, four int8_t bytes. The value is {3,1,0,0} |
michael@0 | 321 | // |
michael@0 | 322 | const uint8_t *inBytes =(const uint8_t *)inData+headerSize; |
michael@0 | 323 | RBBIDataHeader *rbbiDH = (RBBIDataHeader *)inBytes; |
michael@0 | 324 | if (ds->readUInt32(rbbiDH->fMagic) != 0xb1a0 || |
michael@0 | 325 | rbbiDH->fFormatVersion[0] != 3 || |
michael@0 | 326 | ds->readUInt32(rbbiDH->fLength) < sizeof(RBBIDataHeader)) |
michael@0 | 327 | { |
michael@0 | 328 | udata_printError(ds, "ubrk_swap(): RBBI Data header is invalid.\n"); |
michael@0 | 329 | *status=U_UNSUPPORTED_ERROR; |
michael@0 | 330 | return 0; |
michael@0 | 331 | } |
michael@0 | 332 | |
michael@0 | 333 | // |
michael@0 | 334 | // Prefight operation? Just return the size |
michael@0 | 335 | // |
michael@0 | 336 | int32_t breakDataLength = ds->readUInt32(rbbiDH->fLength); |
michael@0 | 337 | int32_t totalSize = headerSize + breakDataLength; |
michael@0 | 338 | if (length < 0) { |
michael@0 | 339 | return totalSize; |
michael@0 | 340 | } |
michael@0 | 341 | |
michael@0 | 342 | // |
michael@0 | 343 | // Check that length passed in is consistent with length from RBBI data header. |
michael@0 | 344 | // |
michael@0 | 345 | if (length < totalSize) { |
michael@0 | 346 | udata_printError(ds, "ubrk_swap(): too few bytes (%d after ICU Data header) for break data.\n", |
michael@0 | 347 | breakDataLength); |
michael@0 | 348 | *status=U_INDEX_OUTOFBOUNDS_ERROR; |
michael@0 | 349 | return 0; |
michael@0 | 350 | } |
michael@0 | 351 | |
michael@0 | 352 | |
michael@0 | 353 | // |
michael@0 | 354 | // Swap the Data. Do the data itself first, then the RBBI Data Header, because |
michael@0 | 355 | // we need to reference the header to locate the data, and an |
michael@0 | 356 | // inplace swap of the header leaves it unusable. |
michael@0 | 357 | // |
michael@0 | 358 | uint8_t *outBytes = (uint8_t *)outData + headerSize; |
michael@0 | 359 | RBBIDataHeader *outputDH = (RBBIDataHeader *)outBytes; |
michael@0 | 360 | |
michael@0 | 361 | int32_t tableStartOffset; |
michael@0 | 362 | int32_t tableLength; |
michael@0 | 363 | |
michael@0 | 364 | // |
michael@0 | 365 | // If not swapping in place, zero out the output buffer before starting. |
michael@0 | 366 | // Individual tables and other data items within are aligned to 8 byte boundaries |
michael@0 | 367 | // when originally created. Any unused space between items needs to be zero. |
michael@0 | 368 | // |
michael@0 | 369 | if (inBytes != outBytes) { |
michael@0 | 370 | uprv_memset(outBytes, 0, breakDataLength); |
michael@0 | 371 | } |
michael@0 | 372 | |
michael@0 | 373 | // |
michael@0 | 374 | // Each state table begins with several 32 bit fields. Calculate the size |
michael@0 | 375 | // in bytes of these. |
michael@0 | 376 | // |
michael@0 | 377 | int32_t topSize = offsetof(RBBIStateTable, fTableData); |
michael@0 | 378 | |
michael@0 | 379 | // Forward state table. |
michael@0 | 380 | tableStartOffset = ds->readUInt32(rbbiDH->fFTable); |
michael@0 | 381 | tableLength = ds->readUInt32(rbbiDH->fFTableLen); |
michael@0 | 382 | |
michael@0 | 383 | if (tableLength > 0) { |
michael@0 | 384 | ds->swapArray32(ds, inBytes+tableStartOffset, topSize, |
michael@0 | 385 | outBytes+tableStartOffset, status); |
michael@0 | 386 | ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize, |
michael@0 | 387 | outBytes+tableStartOffset+topSize, status); |
michael@0 | 388 | } |
michael@0 | 389 | |
michael@0 | 390 | // Reverse state table. Same layout as forward table, above. |
michael@0 | 391 | tableStartOffset = ds->readUInt32(rbbiDH->fRTable); |
michael@0 | 392 | tableLength = ds->readUInt32(rbbiDH->fRTableLen); |
michael@0 | 393 | |
michael@0 | 394 | if (tableLength > 0) { |
michael@0 | 395 | ds->swapArray32(ds, inBytes+tableStartOffset, topSize, |
michael@0 | 396 | outBytes+tableStartOffset, status); |
michael@0 | 397 | ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize, |
michael@0 | 398 | outBytes+tableStartOffset+topSize, status); |
michael@0 | 399 | } |
michael@0 | 400 | |
michael@0 | 401 | // Safe Forward state table. Same layout as forward table, above. |
michael@0 | 402 | tableStartOffset = ds->readUInt32(rbbiDH->fSFTable); |
michael@0 | 403 | tableLength = ds->readUInt32(rbbiDH->fSFTableLen); |
michael@0 | 404 | |
michael@0 | 405 | if (tableLength > 0) { |
michael@0 | 406 | ds->swapArray32(ds, inBytes+tableStartOffset, topSize, |
michael@0 | 407 | outBytes+tableStartOffset, status); |
michael@0 | 408 | ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize, |
michael@0 | 409 | outBytes+tableStartOffset+topSize, status); |
michael@0 | 410 | } |
michael@0 | 411 | |
michael@0 | 412 | // Safe Reverse state table. Same layout as forward table, above. |
michael@0 | 413 | tableStartOffset = ds->readUInt32(rbbiDH->fSRTable); |
michael@0 | 414 | tableLength = ds->readUInt32(rbbiDH->fSRTableLen); |
michael@0 | 415 | |
michael@0 | 416 | if (tableLength > 0) { |
michael@0 | 417 | ds->swapArray32(ds, inBytes+tableStartOffset, topSize, |
michael@0 | 418 | outBytes+tableStartOffset, status); |
michael@0 | 419 | ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize, |
michael@0 | 420 | outBytes+tableStartOffset+topSize, status); |
michael@0 | 421 | } |
michael@0 | 422 | |
michael@0 | 423 | // Trie table for character categories |
michael@0 | 424 | utrie_swap(ds, inBytes+ds->readUInt32(rbbiDH->fTrie), ds->readUInt32(rbbiDH->fTrieLen), |
michael@0 | 425 | outBytes+ds->readUInt32(rbbiDH->fTrie), status); |
michael@0 | 426 | |
michael@0 | 427 | // Source Rules Text. It's UChar data |
michael@0 | 428 | ds->swapArray16(ds, inBytes+ds->readUInt32(rbbiDH->fRuleSource), ds->readUInt32(rbbiDH->fRuleSourceLen), |
michael@0 | 429 | outBytes+ds->readUInt32(rbbiDH->fRuleSource), status); |
michael@0 | 430 | |
michael@0 | 431 | // Table of rule status values. It's all int_32 values |
michael@0 | 432 | ds->swapArray32(ds, inBytes+ds->readUInt32(rbbiDH->fStatusTable), ds->readUInt32(rbbiDH->fStatusTableLen), |
michael@0 | 433 | outBytes+ds->readUInt32(rbbiDH->fStatusTable), status); |
michael@0 | 434 | |
michael@0 | 435 | // And, last, the header. |
michael@0 | 436 | // It is all int32_t values except for fFormataVersion, which is an array of four bytes. |
michael@0 | 437 | // Swap the whole thing as int32_t, then re-swap the one field. |
michael@0 | 438 | // |
michael@0 | 439 | ds->swapArray32(ds, inBytes, sizeof(RBBIDataHeader), outBytes, status); |
michael@0 | 440 | ds->swapArray32(ds, outputDH->fFormatVersion, 4, outputDH->fFormatVersion, status); |
michael@0 | 441 | |
michael@0 | 442 | return totalSize; |
michael@0 | 443 | } |
michael@0 | 444 | |
michael@0 | 445 | |
michael@0 | 446 | #endif /* #if !UCONFIG_NO_BREAK_ITERATION */ |