1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/tools/toolutil/swapimpl.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,828 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 2005-2012, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: swapimpl.cpp
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 2005may05
1.17 +* created by: Markus W. Scherer
1.18 +*
1.19 +* Data file swapping functions moved here from the common library
1.20 +* because some data is hardcoded in ICU4C and needs not be swapped any more.
1.21 +* Moving the functions here simplifies testing (for code coverage) because
1.22 +* we need not jump through hoops (like adding snapshots of these files
1.23 +* to testdata).
1.24 +*
1.25 +* The declarations for these functions remain in the internal header files
1.26 +* in icu/source/common/
1.27 +*/
1.28 +
1.29 +#include "unicode/utypes.h"
1.30 +#include "unicode/putil.h"
1.31 +#include "unicode/udata.h"
1.32 +
1.33 +/* Explicit include statement for std_string.h is needed
1.34 + * for compilation on certain platforms. (e.g. AIX/VACPP)
1.35 + */
1.36 +#include "unicode/std_string.h"
1.37 +
1.38 +#include "cmemory.h"
1.39 +#include "cstring.h"
1.40 +#include "uinvchar.h"
1.41 +#include "uassert.h"
1.42 +#include "uarrsort.h"
1.43 +#include "ucmndata.h"
1.44 +#include "udataswp.h"
1.45 +
1.46 +/* swapping implementations in common */
1.47 +
1.48 +#include "uresdata.h"
1.49 +#include "ucnv_io.h"
1.50 +#include "uprops.h"
1.51 +#include "ucase.h"
1.52 +#include "ubidi_props.h"
1.53 +#include "ucol_swp.h"
1.54 +#include "ucnv_bld.h"
1.55 +#include "unormimp.h"
1.56 +#include "normalizer2impl.h"
1.57 +#include "sprpimpl.h"
1.58 +#include "propname.h"
1.59 +#include "rbbidata.h"
1.60 +#include "utrie2.h"
1.61 +#include "dictionarydata.h"
1.62 +
1.63 +/* swapping implementations in i18n */
1.64 +
1.65 +#if !UCONFIG_NO_NORMALIZATION
1.66 +#include "uspoof_impl.h"
1.67 +#endif
1.68 +
1.69 +U_NAMESPACE_USE
1.70 +
1.71 +/* definitions */
1.72 +
1.73 +#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
1.74 +
1.75 +/* Unicode property (value) aliases data swapping --------------------------- */
1.76 +
1.77 +static int32_t U_CALLCONV
1.78 +upname_swap(const UDataSwapper *ds,
1.79 + const void *inData, int32_t length, void *outData,
1.80 + UErrorCode *pErrorCode) {
1.81 + /* udata_swapDataHeader checks the arguments */
1.82 + int32_t headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1.83 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.84 + return 0;
1.85 + }
1.86 +
1.87 + /* check data format and format version */
1.88 + const UDataInfo *pInfo=
1.89 + reinterpret_cast<const UDataInfo *>(
1.90 + static_cast<const char *>(inData)+4);
1.91 + if(!(
1.92 + pInfo->dataFormat[0]==0x70 && /* dataFormat="pnam" */
1.93 + pInfo->dataFormat[1]==0x6e &&
1.94 + pInfo->dataFormat[2]==0x61 &&
1.95 + pInfo->dataFormat[3]==0x6d &&
1.96 + pInfo->formatVersion[0]==2
1.97 + )) {
1.98 + udata_printError(ds, "upname_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as pnames.icu\n",
1.99 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.100 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.101 + pInfo->formatVersion[0]);
1.102 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.103 + return 0;
1.104 + }
1.105 +
1.106 + const uint8_t *inBytes=static_cast<const uint8_t *>(inData)+headerSize;
1.107 + uint8_t *outBytes=static_cast<uint8_t *>(outData)+headerSize;
1.108 +
1.109 + if(length>=0) {
1.110 + length-=headerSize;
1.111 + // formatVersion 2 initially has indexes[8], 32 bytes.
1.112 + if(length<32) {
1.113 + udata_printError(ds, "upname_swap(): too few bytes (%d after header) for pnames.icu\n",
1.114 + (int)length);
1.115 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.116 + return 0;
1.117 + }
1.118 + }
1.119 +
1.120 + const int32_t *inIndexes=reinterpret_cast<const int32_t *>(inBytes);
1.121 + int32_t totalSize=udata_readInt32(ds, inIndexes[PropNameData::IX_TOTAL_SIZE]);
1.122 + if(length>=0) {
1.123 + if(length<totalSize) {
1.124 + udata_printError(ds, "upname_swap(): too few bytes (%d after header, should be %d) "
1.125 + "for pnames.icu\n",
1.126 + (int)length, (int)totalSize);
1.127 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.128 + return 0;
1.129 + }
1.130 +
1.131 + int32_t numBytesIndexesAndValueMaps=
1.132 + udata_readInt32(ds, inIndexes[PropNameData::IX_BYTE_TRIES_OFFSET]);
1.133 +
1.134 + // Swap the indexes[] and the valueMaps[].
1.135 + ds->swapArray32(ds, inBytes, numBytesIndexesAndValueMaps, outBytes, pErrorCode);
1.136 +
1.137 + // Copy the rest of the data.
1.138 + if(inBytes!=outBytes) {
1.139 + uprv_memcpy(outBytes+numBytesIndexesAndValueMaps,
1.140 + inBytes+numBytesIndexesAndValueMaps,
1.141 + totalSize-numBytesIndexesAndValueMaps);
1.142 + }
1.143 +
1.144 + // We need not swap anything else:
1.145 + //
1.146 + // The ByteTries are already byte-serialized, and are fixed on ASCII.
1.147 + // (On an EBCDIC machine, the input string is converted to lowercase ASCII
1.148 + // while matching.)
1.149 + //
1.150 + // The name groups are mostly invariant characters, but since we only
1.151 + // generate, and keep in subversion, ASCII versions of pnames.icu,
1.152 + // and since only ICU4J uses the pnames.icu data file
1.153 + // (the data is hardcoded in ICU4C) and ICU4J uses ASCII data files,
1.154 + // we just copy those bytes too.
1.155 + }
1.156 +
1.157 + return headerSize+totalSize;
1.158 +}
1.159 +
1.160 +/* Unicode properties data swapping ----------------------------------------- */
1.161 +
1.162 +static int32_t U_CALLCONV
1.163 +uprops_swap(const UDataSwapper *ds,
1.164 + const void *inData, int32_t length, void *outData,
1.165 + UErrorCode *pErrorCode) {
1.166 + const UDataInfo *pInfo;
1.167 + int32_t headerSize, i;
1.168 +
1.169 + int32_t dataIndexes[UPROPS_INDEX_COUNT];
1.170 + const int32_t *inData32;
1.171 +
1.172 + /* udata_swapDataHeader checks the arguments */
1.173 + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1.174 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.175 + return 0;
1.176 + }
1.177 +
1.178 + /* check data format and format version */
1.179 + pInfo=(const UDataInfo *)((const char *)inData+4);
1.180 + if(!(
1.181 + pInfo->dataFormat[0]==0x55 && /* dataFormat="UPro" */
1.182 + pInfo->dataFormat[1]==0x50 &&
1.183 + pInfo->dataFormat[2]==0x72 &&
1.184 + pInfo->dataFormat[3]==0x6f &&
1.185 + (3<=pInfo->formatVersion[0] && pInfo->formatVersion[0]<=7) &&
1.186 + (pInfo->formatVersion[0]>=7 ||
1.187 + (pInfo->formatVersion[2]==UTRIE_SHIFT &&
1.188 + pInfo->formatVersion[3]==UTRIE_INDEX_SHIFT))
1.189 + )) {
1.190 + udata_printError(ds, "uprops_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not a Unicode properties file\n",
1.191 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.192 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.193 + pInfo->formatVersion[0]);
1.194 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.195 + return 0;
1.196 + }
1.197 +
1.198 + /* the properties file must contain at least the indexes array */
1.199 + if(length>=0 && (length-headerSize)<(int32_t)sizeof(dataIndexes)) {
1.200 + udata_printError(ds, "uprops_swap(): too few bytes (%d after header) for a Unicode properties file\n",
1.201 + length-headerSize);
1.202 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.203 + return 0;
1.204 + }
1.205 +
1.206 + /* read the indexes */
1.207 + inData32=(const int32_t *)((const char *)inData+headerSize);
1.208 + for(i=0; i<UPROPS_INDEX_COUNT; ++i) {
1.209 + dataIndexes[i]=udata_readInt32(ds, inData32[i]);
1.210 + }
1.211 +
1.212 + /*
1.213 + * comments are copied from the data format description in genprops/store.c
1.214 + * indexes[] constants are in uprops.h
1.215 + */
1.216 + int32_t dataTop;
1.217 + if(length>=0) {
1.218 + int32_t *outData32;
1.219 +
1.220 + /*
1.221 + * In formatVersion 7, UPROPS_DATA_TOP_INDEX has the post-header data size.
1.222 + * In earlier formatVersions, it is 0 and a lower dataIndexes entry
1.223 + * has the top of the last item.
1.224 + */
1.225 + for(i=UPROPS_DATA_TOP_INDEX; i>0 && (dataTop=dataIndexes[i])==0; --i) {}
1.226 +
1.227 + if((length-headerSize)<(4*dataTop)) {
1.228 + udata_printError(ds, "uprops_swap(): too few bytes (%d after header) for a Unicode properties file\n",
1.229 + length-headerSize);
1.230 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.231 + return 0;
1.232 + }
1.233 +
1.234 + outData32=(int32_t *)((char *)outData+headerSize);
1.235 +
1.236 + /* copy everything for inaccessible data (padding) */
1.237 + if(inData32!=outData32) {
1.238 + uprv_memcpy(outData32, inData32, 4*dataTop);
1.239 + }
1.240 +
1.241 + /* swap the indexes[16] */
1.242 + ds->swapArray32(ds, inData32, 4*UPROPS_INDEX_COUNT, outData32, pErrorCode);
1.243 +
1.244 + /*
1.245 + * swap the main properties UTrie
1.246 + * PT serialized properties trie, see utrie.h (byte size: 4*(i0-16))
1.247 + */
1.248 + utrie2_swapAnyVersion(ds,
1.249 + inData32+UPROPS_INDEX_COUNT,
1.250 + 4*(dataIndexes[UPROPS_PROPS32_INDEX]-UPROPS_INDEX_COUNT),
1.251 + outData32+UPROPS_INDEX_COUNT,
1.252 + pErrorCode);
1.253 +
1.254 + /*
1.255 + * swap the properties and exceptions words
1.256 + * P const uint32_t props32[i1-i0];
1.257 + * E const uint32_t exceptions[i2-i1];
1.258 + */
1.259 + ds->swapArray32(ds,
1.260 + inData32+dataIndexes[UPROPS_PROPS32_INDEX],
1.261 + 4*(dataIndexes[UPROPS_EXCEPTIONS_TOP_INDEX]-dataIndexes[UPROPS_PROPS32_INDEX]),
1.262 + outData32+dataIndexes[UPROPS_PROPS32_INDEX],
1.263 + pErrorCode);
1.264 +
1.265 + /*
1.266 + * swap the UChars
1.267 + * U const UChar uchars[2*(i3-i2)];
1.268 + */
1.269 + ds->swapArray16(ds,
1.270 + inData32+dataIndexes[UPROPS_EXCEPTIONS_TOP_INDEX],
1.271 + 4*(dataIndexes[UPROPS_ADDITIONAL_TRIE_INDEX]-dataIndexes[UPROPS_EXCEPTIONS_TOP_INDEX]),
1.272 + outData32+dataIndexes[UPROPS_EXCEPTIONS_TOP_INDEX],
1.273 + pErrorCode);
1.274 +
1.275 + /*
1.276 + * swap the additional UTrie
1.277 + * i3 additionalTrieIndex; -- 32-bit unit index to the additional trie for more properties
1.278 + */
1.279 + utrie2_swapAnyVersion(ds,
1.280 + inData32+dataIndexes[UPROPS_ADDITIONAL_TRIE_INDEX],
1.281 + 4*(dataIndexes[UPROPS_ADDITIONAL_VECTORS_INDEX]-dataIndexes[UPROPS_ADDITIONAL_TRIE_INDEX]),
1.282 + outData32+dataIndexes[UPROPS_ADDITIONAL_TRIE_INDEX],
1.283 + pErrorCode);
1.284 +
1.285 + /*
1.286 + * swap the properties vectors
1.287 + * PV const uint32_t propsVectors[(i6-i4)/i5][i5]==uint32_t propsVectors[i6-i4];
1.288 + */
1.289 + ds->swapArray32(ds,
1.290 + inData32+dataIndexes[UPROPS_ADDITIONAL_VECTORS_INDEX],
1.291 + 4*(dataIndexes[UPROPS_SCRIPT_EXTENSIONS_INDEX]-dataIndexes[UPROPS_ADDITIONAL_VECTORS_INDEX]),
1.292 + outData32+dataIndexes[UPROPS_ADDITIONAL_VECTORS_INDEX],
1.293 + pErrorCode);
1.294 +
1.295 + // swap the Script_Extensions data
1.296 + // SCX const uint16_t scriptExtensions[2*(i7-i6)];
1.297 + ds->swapArray16(ds,
1.298 + inData32+dataIndexes[UPROPS_SCRIPT_EXTENSIONS_INDEX],
1.299 + 4*(dataIndexes[UPROPS_RESERVED_INDEX_7]-dataIndexes[UPROPS_SCRIPT_EXTENSIONS_INDEX]),
1.300 + outData32+dataIndexes[UPROPS_SCRIPT_EXTENSIONS_INDEX],
1.301 + pErrorCode);
1.302 + }
1.303 +
1.304 + /* i7 reservedIndex7; -- 32-bit unit index to the top of the Script_Extensions data */
1.305 + return headerSize+4*dataIndexes[UPROPS_RESERVED_INDEX_7];
1.306 +}
1.307 +
1.308 +/* Unicode case mapping data swapping --------------------------------------- */
1.309 +
1.310 +static int32_t U_CALLCONV
1.311 +ucase_swap(const UDataSwapper *ds,
1.312 + const void *inData, int32_t length, void *outData,
1.313 + UErrorCode *pErrorCode) {
1.314 + const UDataInfo *pInfo;
1.315 + int32_t headerSize;
1.316 +
1.317 + const uint8_t *inBytes;
1.318 + uint8_t *outBytes;
1.319 +
1.320 + const int32_t *inIndexes;
1.321 + int32_t indexes[16];
1.322 +
1.323 + int32_t i, offset, count, size;
1.324 +
1.325 + /* udata_swapDataHeader checks the arguments */
1.326 + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1.327 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.328 + return 0;
1.329 + }
1.330 +
1.331 + /* check data format and format version */
1.332 + pInfo=(const UDataInfo *)((const char *)inData+4);
1.333 + if(!(
1.334 + pInfo->dataFormat[0]==UCASE_FMT_0 && /* dataFormat="cAsE" */
1.335 + pInfo->dataFormat[1]==UCASE_FMT_1 &&
1.336 + pInfo->dataFormat[2]==UCASE_FMT_2 &&
1.337 + pInfo->dataFormat[3]==UCASE_FMT_3 &&
1.338 + ((pInfo->formatVersion[0]==1 &&
1.339 + pInfo->formatVersion[2]==UTRIE_SHIFT &&
1.340 + pInfo->formatVersion[3]==UTRIE_INDEX_SHIFT) ||
1.341 + pInfo->formatVersion[0]==2 || pInfo->formatVersion[0]==3)
1.342 + )) {
1.343 + udata_printError(ds, "ucase_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as case mapping data\n",
1.344 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.345 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.346 + pInfo->formatVersion[0]);
1.347 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.348 + return 0;
1.349 + }
1.350 +
1.351 + inBytes=(const uint8_t *)inData+headerSize;
1.352 + outBytes=(uint8_t *)outData+headerSize;
1.353 +
1.354 + inIndexes=(const int32_t *)inBytes;
1.355 +
1.356 + if(length>=0) {
1.357 + length-=headerSize;
1.358 + if(length<16*4) {
1.359 + udata_printError(ds, "ucase_swap(): too few bytes (%d after header) for case mapping data\n",
1.360 + length);
1.361 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.362 + return 0;
1.363 + }
1.364 + }
1.365 +
1.366 + /* read the first 16 indexes (ICU 3.2/format version 1: UCASE_IX_TOP==16, might grow) */
1.367 + for(i=0; i<16; ++i) {
1.368 + indexes[i]=udata_readInt32(ds, inIndexes[i]);
1.369 + }
1.370 +
1.371 + /* get the total length of the data */
1.372 + size=indexes[UCASE_IX_LENGTH];
1.373 +
1.374 + if(length>=0) {
1.375 + if(length<size) {
1.376 + udata_printError(ds, "ucase_swap(): too few bytes (%d after header) for all of case mapping data\n",
1.377 + length);
1.378 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.379 + return 0;
1.380 + }
1.381 +
1.382 + /* copy the data for inaccessible bytes */
1.383 + if(inBytes!=outBytes) {
1.384 + uprv_memcpy(outBytes, inBytes, size);
1.385 + }
1.386 +
1.387 + offset=0;
1.388 +
1.389 + /* swap the int32_t indexes[] */
1.390 + count=indexes[UCASE_IX_INDEX_TOP]*4;
1.391 + ds->swapArray32(ds, inBytes, count, outBytes, pErrorCode);
1.392 + offset+=count;
1.393 +
1.394 + /* swap the UTrie */
1.395 + count=indexes[UCASE_IX_TRIE_SIZE];
1.396 + utrie2_swapAnyVersion(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.397 + offset+=count;
1.398 +
1.399 + /* swap the uint16_t exceptions[] and unfold[] */
1.400 + count=(indexes[UCASE_IX_EXC_LENGTH]+indexes[UCASE_IX_UNFOLD_LENGTH])*2;
1.401 + ds->swapArray16(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.402 + offset+=count;
1.403 +
1.404 + U_ASSERT(offset==size);
1.405 + }
1.406 +
1.407 + return headerSize+size;
1.408 +}
1.409 +
1.410 +/* Unicode bidi/shaping data swapping --------------------------------------- */
1.411 +
1.412 +static int32_t U_CALLCONV
1.413 +ubidi_swap(const UDataSwapper *ds,
1.414 + const void *inData, int32_t length, void *outData,
1.415 + UErrorCode *pErrorCode) {
1.416 + const UDataInfo *pInfo;
1.417 + int32_t headerSize;
1.418 +
1.419 + const uint8_t *inBytes;
1.420 + uint8_t *outBytes;
1.421 +
1.422 + const int32_t *inIndexes;
1.423 + int32_t indexes[16];
1.424 +
1.425 + int32_t i, offset, count, size;
1.426 +
1.427 + /* udata_swapDataHeader checks the arguments */
1.428 + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1.429 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.430 + return 0;
1.431 + }
1.432 +
1.433 + /* check data format and format version */
1.434 + pInfo=(const UDataInfo *)((const char *)inData+4);
1.435 + if(!(
1.436 + pInfo->dataFormat[0]==UBIDI_FMT_0 && /* dataFormat="BiDi" */
1.437 + pInfo->dataFormat[1]==UBIDI_FMT_1 &&
1.438 + pInfo->dataFormat[2]==UBIDI_FMT_2 &&
1.439 + pInfo->dataFormat[3]==UBIDI_FMT_3 &&
1.440 + ((pInfo->formatVersion[0]==1 &&
1.441 + pInfo->formatVersion[2]==UTRIE_SHIFT &&
1.442 + pInfo->formatVersion[3]==UTRIE_INDEX_SHIFT) ||
1.443 + pInfo->formatVersion[0]==2)
1.444 + )) {
1.445 + udata_printError(ds, "ubidi_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as bidi/shaping data\n",
1.446 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.447 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.448 + pInfo->formatVersion[0]);
1.449 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.450 + return 0;
1.451 + }
1.452 +
1.453 + inBytes=(const uint8_t *)inData+headerSize;
1.454 + outBytes=(uint8_t *)outData+headerSize;
1.455 +
1.456 + inIndexes=(const int32_t *)inBytes;
1.457 +
1.458 + if(length>=0) {
1.459 + length-=headerSize;
1.460 + if(length<16*4) {
1.461 + udata_printError(ds, "ubidi_swap(): too few bytes (%d after header) for bidi/shaping data\n",
1.462 + length);
1.463 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.464 + return 0;
1.465 + }
1.466 + }
1.467 +
1.468 + /* read the first 16 indexes (ICU 3.4/format version 1: UBIDI_IX_TOP==16, might grow) */
1.469 + for(i=0; i<16; ++i) {
1.470 + indexes[i]=udata_readInt32(ds, inIndexes[i]);
1.471 + }
1.472 +
1.473 + /* get the total length of the data */
1.474 + size=indexes[UBIDI_IX_LENGTH];
1.475 +
1.476 + if(length>=0) {
1.477 + if(length<size) {
1.478 + udata_printError(ds, "ubidi_swap(): too few bytes (%d after header) for all of bidi/shaping data\n",
1.479 + length);
1.480 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.481 + return 0;
1.482 + }
1.483 +
1.484 + /* copy the data for inaccessible bytes */
1.485 + if(inBytes!=outBytes) {
1.486 + uprv_memcpy(outBytes, inBytes, size);
1.487 + }
1.488 +
1.489 + offset=0;
1.490 +
1.491 + /* swap the int32_t indexes[] */
1.492 + count=indexes[UBIDI_IX_INDEX_TOP]*4;
1.493 + ds->swapArray32(ds, inBytes, count, outBytes, pErrorCode);
1.494 + offset+=count;
1.495 +
1.496 + /* swap the UTrie */
1.497 + count=indexes[UBIDI_IX_TRIE_SIZE];
1.498 + utrie2_swapAnyVersion(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.499 + offset+=count;
1.500 +
1.501 + /* swap the uint32_t mirrors[] */
1.502 + count=indexes[UBIDI_IX_MIRROR_LENGTH]*4;
1.503 + ds->swapArray32(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.504 + offset+=count;
1.505 +
1.506 + /* just skip the uint8_t jgArray[] */
1.507 + count=indexes[UBIDI_IX_JG_LIMIT]-indexes[UBIDI_IX_JG_START];
1.508 + offset+=count;
1.509 +
1.510 + U_ASSERT(offset==size);
1.511 + }
1.512 +
1.513 + return headerSize+size;
1.514 +}
1.515 +
1.516 +/* Unicode normalization data swapping -------------------------------------- */
1.517 +
1.518 +#if !UCONFIG_NO_NORMALIZATION
1.519 +
1.520 +static int32_t U_CALLCONV
1.521 +unorm_swap(const UDataSwapper *ds,
1.522 + const void *inData, int32_t length, void *outData,
1.523 + UErrorCode *pErrorCode) {
1.524 + const UDataInfo *pInfo;
1.525 + int32_t headerSize;
1.526 +
1.527 + const uint8_t *inBytes;
1.528 + uint8_t *outBytes;
1.529 +
1.530 + const int32_t *inIndexes;
1.531 + int32_t indexes[32];
1.532 +
1.533 + int32_t i, offset, count, size;
1.534 +
1.535 + /* udata_swapDataHeader checks the arguments */
1.536 + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1.537 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.538 + return 0;
1.539 + }
1.540 +
1.541 + /* check data format and format version */
1.542 + pInfo=(const UDataInfo *)((const char *)inData+4);
1.543 + if(!(
1.544 + pInfo->dataFormat[0]==0x4e && /* dataFormat="Norm" */
1.545 + pInfo->dataFormat[1]==0x6f &&
1.546 + pInfo->dataFormat[2]==0x72 &&
1.547 + pInfo->dataFormat[3]==0x6d &&
1.548 + pInfo->formatVersion[0]==2
1.549 + )) {
1.550 + udata_printError(ds, "unorm_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as unorm.icu\n",
1.551 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.552 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.553 + pInfo->formatVersion[0]);
1.554 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.555 + return 0;
1.556 + }
1.557 +
1.558 + inBytes=(const uint8_t *)inData+headerSize;
1.559 + outBytes=(uint8_t *)outData+headerSize;
1.560 +
1.561 + inIndexes=(const int32_t *)inBytes;
1.562 +
1.563 + if(length>=0) {
1.564 + length-=headerSize;
1.565 + if(length<32*4) {
1.566 + udata_printError(ds, "unorm_swap(): too few bytes (%d after header) for unorm.icu\n",
1.567 + length);
1.568 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.569 + return 0;
1.570 + }
1.571 + }
1.572 +
1.573 + /* read the first 32 indexes (ICU 2.8/format version 2.2: _NORM_INDEX_TOP==32, might grow) */
1.574 + for(i=0; i<32; ++i) {
1.575 + indexes[i]=udata_readInt32(ds, inIndexes[i]);
1.576 + }
1.577 +
1.578 + /* calculate the total length of the data */
1.579 + size=
1.580 + 32*4+ /* size of indexes[] */
1.581 + indexes[_NORM_INDEX_TRIE_SIZE]+
1.582 + indexes[_NORM_INDEX_UCHAR_COUNT]*2+
1.583 + indexes[_NORM_INDEX_COMBINE_DATA_COUNT]*2+
1.584 + indexes[_NORM_INDEX_FCD_TRIE_SIZE]+
1.585 + indexes[_NORM_INDEX_AUX_TRIE_SIZE]+
1.586 + indexes[_NORM_INDEX_CANON_SET_COUNT]*2;
1.587 +
1.588 + if(length>=0) {
1.589 + if(length<size) {
1.590 + udata_printError(ds, "unorm_swap(): too few bytes (%d after header) for all of unorm.icu\n",
1.591 + length);
1.592 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.593 + return 0;
1.594 + }
1.595 +
1.596 + /* copy the data for inaccessible bytes */
1.597 + if(inBytes!=outBytes) {
1.598 + uprv_memcpy(outBytes, inBytes, size);
1.599 + }
1.600 +
1.601 + offset=0;
1.602 +
1.603 + /* swap the indexes[] */
1.604 + count=32*4;
1.605 + ds->swapArray32(ds, inBytes, count, outBytes, pErrorCode);
1.606 + offset+=count;
1.607 +
1.608 + /* swap the main UTrie */
1.609 + count=indexes[_NORM_INDEX_TRIE_SIZE];
1.610 + utrie_swap(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.611 + offset+=count;
1.612 +
1.613 + /* swap the uint16_t extraData[] and the uint16_t combiningTable[] */
1.614 + count=(indexes[_NORM_INDEX_UCHAR_COUNT]+indexes[_NORM_INDEX_COMBINE_DATA_COUNT])*2;
1.615 + ds->swapArray16(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.616 + offset+=count;
1.617 +
1.618 + /* swap the FCD UTrie */
1.619 + count=indexes[_NORM_INDEX_FCD_TRIE_SIZE];
1.620 + if(count!=0) {
1.621 + utrie_swap(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.622 + offset+=count;
1.623 + }
1.624 +
1.625 + /* swap the aux UTrie */
1.626 + count=indexes[_NORM_INDEX_AUX_TRIE_SIZE];
1.627 + if(count!=0) {
1.628 + utrie_swap(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.629 + offset+=count;
1.630 + }
1.631 +
1.632 + /* swap the uint16_t combiningTable[] */
1.633 + count=indexes[_NORM_INDEX_CANON_SET_COUNT]*2;
1.634 + ds->swapArray16(ds, inBytes+offset, count, outBytes+offset, pErrorCode);
1.635 + offset+=count;
1.636 + }
1.637 +
1.638 + return headerSize+size;
1.639 +}
1.640 +
1.641 +#endif
1.642 +
1.643 +/* Swap 'Test' data from gentest */
1.644 +static int32_t U_CALLCONV
1.645 +test_swap(const UDataSwapper *ds,
1.646 + const void *inData, int32_t length, void *outData,
1.647 + UErrorCode *pErrorCode) {
1.648 + const UDataInfo *pInfo;
1.649 + int32_t headerSize;
1.650 +
1.651 + const uint8_t *inBytes;
1.652 + uint8_t *outBytes;
1.653 +
1.654 + int32_t offset;
1.655 +
1.656 + /* udata_swapDataHeader checks the arguments */
1.657 + headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode);
1.658 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.659 + udata_printError(ds, "test_swap(): data header swap failed %s\n", pErrorCode != NULL ? u_errorName(*pErrorCode) : "pErrorCode is NULL");
1.660 + return 0;
1.661 + }
1.662 +
1.663 + /* check data format and format version */
1.664 + pInfo=(const UDataInfo *)((const char *)inData+4);
1.665 + if(!(
1.666 + pInfo->dataFormat[0]==0x54 && /* dataFormat="Norm" */
1.667 + pInfo->dataFormat[1]==0x65 &&
1.668 + pInfo->dataFormat[2]==0x73 &&
1.669 + pInfo->dataFormat[3]==0x74 &&
1.670 + pInfo->formatVersion[0]==1
1.671 + )) {
1.672 + udata_printError(ds, "test_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as testdata\n",
1.673 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.674 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.675 + pInfo->formatVersion[0]);
1.676 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.677 + return 0;
1.678 + }
1.679 +
1.680 + inBytes=(const uint8_t *)inData+headerSize;
1.681 + outBytes=(uint8_t *)outData+headerSize;
1.682 +
1.683 + int32_t size16 = 2; // 16bit plus padding
1.684 + int32_t sizeStr = 5; // 4 char inv-str plus null
1.685 + int32_t size = size16 + sizeStr;
1.686 +
1.687 + if(length>=0) {
1.688 + if(length<size) {
1.689 + udata_printError(ds, "test_swap(): too few bytes (%d after header, wanted %d) for all of testdata\n",
1.690 + length, size);
1.691 + *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1.692 + return 0;
1.693 + }
1.694 +
1.695 + offset =0;
1.696 + /* swap a 1 entry array */
1.697 + ds->swapArray16(ds, inBytes+offset, size16, outBytes+offset, pErrorCode);
1.698 + offset+=size16;
1.699 + ds->swapInvChars(ds, inBytes+offset, sizeStr, outBytes+offset, pErrorCode);
1.700 + }
1.701 +
1.702 + return headerSize+size;
1.703 +}
1.704 +
1.705 +/* swap any data (except a .dat package) ------------------------------------ */
1.706 +
1.707 +static const struct {
1.708 + uint8_t dataFormat[4];
1.709 + UDataSwapFn *swapFn;
1.710 +} swapFns[]={
1.711 + { { 0x52, 0x65, 0x73, 0x42 }, ures_swap }, /* dataFormat="ResB" */
1.712 +#if !UCONFIG_NO_LEGACY_CONVERSION
1.713 + { { 0x63, 0x6e, 0x76, 0x74 }, ucnv_swap }, /* dataFormat="cnvt" */
1.714 +#endif
1.715 +#if !UCONFIG_NO_CONVERSION
1.716 + { { 0x43, 0x76, 0x41, 0x6c }, ucnv_swapAliases }, /* dataFormat="CvAl" */
1.717 +#endif
1.718 +#if !UCONFIG_NO_IDNA
1.719 + { { 0x53, 0x50, 0x52, 0x50 }, usprep_swap }, /* dataFormat="SPRP" */
1.720 +#endif
1.721 + /* insert data formats here, descending by expected frequency of occurrence */
1.722 + { { 0x55, 0x50, 0x72, 0x6f }, uprops_swap }, /* dataFormat="UPro" */
1.723 +
1.724 + { { UCASE_FMT_0, UCASE_FMT_1, UCASE_FMT_2, UCASE_FMT_3 },
1.725 + ucase_swap }, /* dataFormat="cAsE" */
1.726 +
1.727 + { { UBIDI_FMT_0, UBIDI_FMT_1, UBIDI_FMT_2, UBIDI_FMT_3 },
1.728 + ubidi_swap }, /* dataFormat="BiDi" */
1.729 +
1.730 +#if !UCONFIG_NO_NORMALIZATION
1.731 + { { 0x4e, 0x6f, 0x72, 0x6d }, unorm_swap }, /* dataFormat="Norm" */
1.732 + { { 0x4e, 0x72, 0x6d, 0x32 }, unorm2_swap }, /* dataFormat="Nrm2" */
1.733 +#endif
1.734 +#if !UCONFIG_NO_COLLATION
1.735 + { { 0x55, 0x43, 0x6f, 0x6c }, ucol_swap }, /* dataFormat="UCol" */
1.736 + { { 0x49, 0x6e, 0x76, 0x43 }, ucol_swapInverseUCA },/* dataFormat="InvC" */
1.737 +#endif
1.738 +#if !UCONFIG_NO_BREAK_ITERATION
1.739 + { { 0x42, 0x72, 0x6b, 0x20 }, ubrk_swap }, /* dataFormat="Brk " */
1.740 + { { 0x44, 0x69, 0x63, 0x74 }, udict_swap }, /* dataFormat="Dict" */
1.741 +#endif
1.742 + { { 0x70, 0x6e, 0x61, 0x6d }, upname_swap }, /* dataFormat="pnam" */
1.743 + { { 0x75, 0x6e, 0x61, 0x6d }, uchar_swapNames }, /* dataFormat="unam" */
1.744 +#if !UCONFIG_NO_NORMALIZATION
1.745 + { { 0x43, 0x66, 0x75, 0x20 }, uspoof_swap }, /* dataFormat="Cfu " */
1.746 +#endif
1.747 + { { 0x54, 0x65, 0x73, 0x74 }, test_swap } /* dataFormat="Test" */
1.748 +};
1.749 +
1.750 +U_CAPI int32_t U_EXPORT2
1.751 +udata_swap(const UDataSwapper *ds,
1.752 + const void *inData, int32_t length, void *outData,
1.753 + UErrorCode *pErrorCode) {
1.754 + char dataFormatChars[4];
1.755 + const UDataInfo *pInfo;
1.756 + int32_t i, swappedLength;
1.757 +
1.758 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.759 + return 0;
1.760 + }
1.761 +
1.762 + /*
1.763 + * Preflight the header first; checks for illegal arguments, too.
1.764 + * Do not swap the header right away because the format-specific swapper
1.765 + * will swap it, get the headerSize again, and also use the header
1.766 + * information. Otherwise we would have to pass some of the information
1.767 + * and not be able to use the UDataSwapFn signature.
1.768 + */
1.769 + udata_swapDataHeader(ds, inData, -1, NULL, pErrorCode);
1.770 +
1.771 + /*
1.772 + * If we wanted udata_swap() to also handle non-loadable data like a UTrie,
1.773 + * then we could check here for further known magic values and structures.
1.774 + */
1.775 + if(U_FAILURE(*pErrorCode)) {
1.776 + return 0; /* the data format was not recognized */
1.777 + }
1.778 +
1.779 + pInfo=(const UDataInfo *)((const char *)inData+4);
1.780 +
1.781 + {
1.782 + /* convert the data format from ASCII to Unicode to the system charset */
1.783 + UChar u[4]={
1.784 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.785 + pInfo->dataFormat[2], pInfo->dataFormat[3]
1.786 + };
1.787 +
1.788 + if(uprv_isInvariantUString(u, 4)) {
1.789 + u_UCharsToChars(u, dataFormatChars, 4);
1.790 + } else {
1.791 + dataFormatChars[0]=dataFormatChars[1]=dataFormatChars[2]=dataFormatChars[3]='?';
1.792 + }
1.793 + }
1.794 +
1.795 + /* dispatch to the swap function for the dataFormat */
1.796 + for(i=0; i<LENGTHOF(swapFns); ++i) {
1.797 + if(0==memcmp(swapFns[i].dataFormat, pInfo->dataFormat, 4)) {
1.798 + swappedLength=swapFns[i].swapFn(ds, inData, length, outData, pErrorCode);
1.799 +
1.800 + if(U_FAILURE(*pErrorCode)) {
1.801 + udata_printError(ds, "udata_swap(): failure swapping data format %02x.%02x.%02x.%02x (\"%c%c%c%c\") - %s\n",
1.802 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.803 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.804 + dataFormatChars[0], dataFormatChars[1],
1.805 + dataFormatChars[2], dataFormatChars[3],
1.806 + u_errorName(*pErrorCode));
1.807 + } else if(swappedLength<(length-15)) {
1.808 + /* swapped less than expected */
1.809 + udata_printError(ds, "udata_swap() warning: swapped only %d out of %d bytes - data format %02x.%02x.%02x.%02x (\"%c%c%c%c\")\n",
1.810 + swappedLength, length,
1.811 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.812 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.813 + dataFormatChars[0], dataFormatChars[1],
1.814 + dataFormatChars[2], dataFormatChars[3],
1.815 + u_errorName(*pErrorCode));
1.816 + }
1.817 +
1.818 + return swappedLength;
1.819 + }
1.820 + }
1.821 +
1.822 + /* the dataFormat was not recognized */
1.823 + udata_printError(ds, "udata_swap(): unknown data format %02x.%02x.%02x.%02x (\"%c%c%c%c\")\n",
1.824 + pInfo->dataFormat[0], pInfo->dataFormat[1],
1.825 + pInfo->dataFormat[2], pInfo->dataFormat[3],
1.826 + dataFormatChars[0], dataFormatChars[1],
1.827 + dataFormatChars[2], dataFormatChars[3]);
1.828 +
1.829 + *pErrorCode=U_UNSUPPORTED_ERROR;
1.830 + return 0;
1.831 +}
