/src/skia/third_party/externals/icu/source/common/ucol_swp.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // © 2016 and later: Unicode, Inc. and others. |
2 | | // License & terms of use: http://www.unicode.org/copyright.html |
3 | | /* |
4 | | ******************************************************************************* |
5 | | * |
6 | | * Copyright (C) 2003-2015, International Business Machines |
7 | | * Corporation and others. All Rights Reserved. |
8 | | * |
9 | | ******************************************************************************* |
10 | | * file name: ucol_swp.cpp |
11 | | * encoding: UTF-8 |
12 | | * tab size: 8 (not used) |
13 | | * indentation:4 |
14 | | * |
15 | | * created on: 2003sep10 |
16 | | * created by: Markus W. Scherer |
17 | | * |
18 | | * Swap collation binaries. |
19 | | */ |
20 | | |
21 | | #include "unicode/udata.h" /* UDataInfo */ |
22 | | #include "utrie.h" |
23 | | #include "utrie2.h" |
24 | | #include "udataswp.h" |
25 | | #include "cmemory.h" |
26 | | #include "ucol_data.h" |
27 | | #include "ucol_swp.h" |
28 | | |
29 | | /* swapping ----------------------------------------------------------------- */ |
30 | | |
31 | | #if !UCONFIG_NO_COLLATION |
32 | | |
33 | | U_CAPI UBool U_EXPORT2 |
34 | | ucol_looksLikeCollationBinary(const UDataSwapper *ds, |
35 | 0 | const void *inData, int32_t length) { |
36 | 0 | if(ds==NULL || inData==NULL || length<-1) { |
37 | 0 | return FALSE; |
38 | 0 | } |
39 | | |
40 | | // First check for format version 4+ which has a standard data header. |
41 | 0 | UErrorCode errorCode=U_ZERO_ERROR; |
42 | 0 | (void)udata_swapDataHeader(ds, inData, -1, NULL, &errorCode); |
43 | 0 | if(U_SUCCESS(errorCode)) { |
44 | 0 | const UDataInfo &info=*(const UDataInfo *)((const char *)inData+4); |
45 | 0 | if(info.dataFormat[0]==0x55 && // dataFormat="UCol" |
46 | 0 | info.dataFormat[1]==0x43 && |
47 | 0 | info.dataFormat[2]==0x6f && |
48 | 0 | info.dataFormat[3]==0x6c) { |
49 | 0 | return TRUE; |
50 | 0 | } |
51 | 0 | } |
52 | | |
53 | | // Else check for format version 3. |
54 | 0 | const UCATableHeader *inHeader=(const UCATableHeader *)inData; |
55 | | |
56 | | /* |
57 | | * The collation binary must contain at least the UCATableHeader, |
58 | | * starting with its size field. |
59 | | * sizeof(UCATableHeader)==42*4 in ICU 2.8 |
60 | | * check the length against the header size before reading the size field |
61 | | */ |
62 | 0 | UCATableHeader header; |
63 | 0 | uprv_memset(&header, 0, sizeof(header)); |
64 | 0 | if(length<0) { |
65 | 0 | header.size=udata_readInt32(ds, inHeader->size); |
66 | 0 | } else if((length<(42*4) || length<(header.size=udata_readInt32(ds, inHeader->size)))) { |
67 | 0 | return FALSE; |
68 | 0 | } |
69 | | |
70 | 0 | header.magic=ds->readUInt32(inHeader->magic); |
71 | 0 | if(!( |
72 | 0 | header.magic==UCOL_HEADER_MAGIC && |
73 | 0 | inHeader->formatVersion[0]==3 /*&& |
74 | | inHeader->formatVersion[1]>=0*/ |
75 | 0 | )) { |
76 | 0 | return FALSE; |
77 | 0 | } |
78 | | |
79 | 0 | if(inHeader->isBigEndian!=ds->inIsBigEndian || inHeader->charSetFamily!=ds->inCharset) { |
80 | 0 | return FALSE; |
81 | 0 | } |
82 | | |
83 | 0 | return TRUE; |
84 | 0 | } |
85 | | |
86 | | namespace { |
87 | | |
88 | | /* swap a header-less collation formatVersion=3 binary, inside a resource bundle or ucadata.icu */ |
89 | | int32_t |
90 | | swapFormatVersion3(const UDataSwapper *ds, |
91 | | const void *inData, int32_t length, void *outData, |
92 | 0 | UErrorCode *pErrorCode) { |
93 | 0 | const uint8_t *inBytes; |
94 | 0 | uint8_t *outBytes; |
95 | |
|
96 | 0 | const UCATableHeader *inHeader; |
97 | 0 | UCATableHeader *outHeader; |
98 | 0 | UCATableHeader header; |
99 | |
|
100 | 0 | uint32_t count; |
101 | | |
102 | | /* argument checking in case we were not called from ucol_swap() */ |
103 | 0 | if(U_FAILURE(*pErrorCode)) { |
104 | 0 | return 0; |
105 | 0 | } |
106 | 0 | if(ds==NULL || inData==NULL || length<-1 || (length>0 && outData==NULL)) { |
107 | 0 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
108 | 0 | return 0; |
109 | 0 | } |
110 | | |
111 | 0 | inBytes=(const uint8_t *)inData; |
112 | 0 | outBytes=(uint8_t *)outData; |
113 | |
|
114 | 0 | inHeader=(const UCATableHeader *)inData; |
115 | 0 | outHeader=(UCATableHeader *)outData; |
116 | | |
117 | | /* |
118 | | * The collation binary must contain at least the UCATableHeader, |
119 | | * starting with its size field. |
120 | | * sizeof(UCATableHeader)==42*4 in ICU 2.8 |
121 | | * check the length against the header size before reading the size field |
122 | | */ |
123 | 0 | uprv_memset(&header, 0, sizeof(header)); |
124 | 0 | if(length<0) { |
125 | 0 | header.size=udata_readInt32(ds, inHeader->size); |
126 | 0 | } else if((length<(42*4) || length<(header.size=udata_readInt32(ds, inHeader->size)))) { |
127 | 0 | udata_printError(ds, "ucol_swap(formatVersion=3): too few bytes (%d after header) for collation data\n", |
128 | 0 | length); |
129 | 0 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
130 | 0 | return 0; |
131 | 0 | } |
132 | | |
133 | 0 | header.magic=ds->readUInt32(inHeader->magic); |
134 | 0 | if(!( |
135 | 0 | header.magic==UCOL_HEADER_MAGIC && |
136 | 0 | inHeader->formatVersion[0]==3 /*&& |
137 | | inHeader->formatVersion[1]>=0*/ |
138 | 0 | )) { |
139 | 0 | udata_printError(ds, "ucol_swap(formatVersion=3): magic 0x%08x or format version %02x.%02x is not a collation binary\n", |
140 | 0 | header.magic, |
141 | 0 | inHeader->formatVersion[0], inHeader->formatVersion[1]); |
142 | 0 | *pErrorCode=U_UNSUPPORTED_ERROR; |
143 | 0 | return 0; |
144 | 0 | } |
145 | | |
146 | 0 | if(inHeader->isBigEndian!=ds->inIsBigEndian || inHeader->charSetFamily!=ds->inCharset) { |
147 | 0 | udata_printError(ds, "ucol_swap(formatVersion=3): endianness %d or charset %d does not match the swapper\n", |
148 | 0 | inHeader->isBigEndian, inHeader->charSetFamily); |
149 | 0 | *pErrorCode=U_INVALID_FORMAT_ERROR; |
150 | 0 | return 0; |
151 | 0 | } |
152 | | |
153 | 0 | if(length>=0) { |
154 | | /* copy everything, takes care of data that needs no swapping */ |
155 | 0 | if(inBytes!=outBytes) { |
156 | 0 | uprv_memcpy(outBytes, inBytes, header.size); |
157 | 0 | } |
158 | | |
159 | | /* swap the necessary pieces in the order of their occurrence in the data */ |
160 | | |
161 | | /* read more of the UCATableHeader (the size field was read above) */ |
162 | 0 | header.options= ds->readUInt32(inHeader->options); |
163 | 0 | header.UCAConsts= ds->readUInt32(inHeader->UCAConsts); |
164 | 0 | header.contractionUCACombos= ds->readUInt32(inHeader->contractionUCACombos); |
165 | 0 | header.mappingPosition= ds->readUInt32(inHeader->mappingPosition); |
166 | 0 | header.expansion= ds->readUInt32(inHeader->expansion); |
167 | 0 | header.contractionIndex= ds->readUInt32(inHeader->contractionIndex); |
168 | 0 | header.contractionCEs= ds->readUInt32(inHeader->contractionCEs); |
169 | 0 | header.contractionSize= ds->readUInt32(inHeader->contractionSize); |
170 | 0 | header.endExpansionCE= ds->readUInt32(inHeader->endExpansionCE); |
171 | 0 | header.expansionCESize= ds->readUInt32(inHeader->expansionCESize); |
172 | 0 | header.endExpansionCECount= udata_readInt32(ds, inHeader->endExpansionCECount); |
173 | 0 | header.contractionUCACombosSize=udata_readInt32(ds, inHeader->contractionUCACombosSize); |
174 | 0 | header.scriptToLeadByte= ds->readUInt32(inHeader->scriptToLeadByte); |
175 | 0 | header.leadByteToScript= ds->readUInt32(inHeader->leadByteToScript); |
176 | | |
177 | | /* swap the 32-bit integers in the header */ |
178 | 0 | ds->swapArray32(ds, inHeader, (int32_t)((const char *)&inHeader->jamoSpecial-(const char *)inHeader), |
179 | 0 | outHeader, pErrorCode); |
180 | 0 | ds->swapArray32(ds, &(inHeader->scriptToLeadByte), sizeof(header.scriptToLeadByte) + sizeof(header.leadByteToScript), |
181 | 0 | &(outHeader->scriptToLeadByte), pErrorCode); |
182 | | /* set the output platform properties */ |
183 | 0 | outHeader->isBigEndian=ds->outIsBigEndian; |
184 | 0 | outHeader->charSetFamily=ds->outCharset; |
185 | | |
186 | | /* swap the options */ |
187 | 0 | if(header.options!=0) { |
188 | 0 | ds->swapArray32(ds, inBytes+header.options, header.expansion-header.options, |
189 | 0 | outBytes+header.options, pErrorCode); |
190 | 0 | } |
191 | | |
192 | | /* swap the expansions */ |
193 | 0 | if(header.mappingPosition!=0 && header.expansion!=0) { |
194 | 0 | if(header.contractionIndex!=0) { |
195 | | /* expansions bounded by contractions */ |
196 | 0 | count=header.contractionIndex-header.expansion; |
197 | 0 | } else { |
198 | | /* no contractions: expansions bounded by the main trie */ |
199 | 0 | count=header.mappingPosition-header.expansion; |
200 | 0 | } |
201 | 0 | ds->swapArray32(ds, inBytes+header.expansion, (int32_t)count, |
202 | 0 | outBytes+header.expansion, pErrorCode); |
203 | 0 | } |
204 | | |
205 | | /* swap the contractions */ |
206 | 0 | if(header.contractionSize!=0) { |
207 | | /* contractionIndex: UChar[] */ |
208 | 0 | ds->swapArray16(ds, inBytes+header.contractionIndex, header.contractionSize*2, |
209 | 0 | outBytes+header.contractionIndex, pErrorCode); |
210 | | |
211 | | /* contractionCEs: CEs[] */ |
212 | 0 | ds->swapArray32(ds, inBytes+header.contractionCEs, header.contractionSize*4, |
213 | 0 | outBytes+header.contractionCEs, pErrorCode); |
214 | 0 | } |
215 | | |
216 | | /* swap the main trie */ |
217 | 0 | if(header.mappingPosition!=0) { |
218 | 0 | count=header.endExpansionCE-header.mappingPosition; |
219 | 0 | utrie_swap(ds, inBytes+header.mappingPosition, (int32_t)count, |
220 | 0 | outBytes+header.mappingPosition, pErrorCode); |
221 | 0 | } |
222 | | |
223 | | /* swap the max expansion table */ |
224 | 0 | if(header.endExpansionCECount!=0) { |
225 | 0 | ds->swapArray32(ds, inBytes+header.endExpansionCE, header.endExpansionCECount*4, |
226 | 0 | outBytes+header.endExpansionCE, pErrorCode); |
227 | 0 | } |
228 | | |
229 | | /* expansionCESize, unsafeCP, contrEndCP: uint8_t[], no need to swap */ |
230 | | |
231 | | /* swap UCA constants */ |
232 | 0 | if(header.UCAConsts!=0) { |
233 | | /* |
234 | | * if UCAConsts!=0 then contractionUCACombos because we are swapping |
235 | | * the UCA data file, and we know that the UCA contains contractions |
236 | | */ |
237 | 0 | ds->swapArray32(ds, inBytes+header.UCAConsts, header.contractionUCACombos-header.UCAConsts, |
238 | 0 | outBytes+header.UCAConsts, pErrorCode); |
239 | 0 | } |
240 | | |
241 | | /* swap UCA contractions */ |
242 | 0 | if(header.contractionUCACombosSize!=0) { |
243 | 0 | count=header.contractionUCACombosSize*inHeader->contractionUCACombosWidth*U_SIZEOF_UCHAR; |
244 | 0 | ds->swapArray16(ds, inBytes+header.contractionUCACombos, (int32_t)count, |
245 | 0 | outBytes+header.contractionUCACombos, pErrorCode); |
246 | 0 | } |
247 | | |
248 | | /* swap the script to lead bytes */ |
249 | 0 | if(header.scriptToLeadByte!=0) { |
250 | 0 | int indexCount = ds->readUInt16(*((uint16_t*)(inBytes+header.scriptToLeadByte))); // each entry = 2 * uint16 |
251 | 0 | int dataCount = ds->readUInt16(*((uint16_t*)(inBytes+header.scriptToLeadByte + 2))); // each entry = uint16 |
252 | 0 | ds->swapArray16(ds, inBytes+header.scriptToLeadByte, |
253 | 0 | 4 + (4 * indexCount) + (2 * dataCount), |
254 | 0 | outBytes+header.scriptToLeadByte, pErrorCode); |
255 | 0 | } |
256 | | |
257 | | /* swap the lead byte to scripts */ |
258 | 0 | if(header.leadByteToScript!=0) { |
259 | 0 | int indexCount = ds->readUInt16(*((uint16_t*)(inBytes+header.leadByteToScript))); // each entry = uint16 |
260 | 0 | int dataCount = ds->readUInt16(*((uint16_t*)(inBytes+header.leadByteToScript + 2))); // each entry = uint16 |
261 | 0 | ds->swapArray16(ds, inBytes+header.leadByteToScript, |
262 | 0 | 4 + (2 * indexCount) + (2 * dataCount), |
263 | 0 | outBytes+header.leadByteToScript, pErrorCode); |
264 | 0 | } |
265 | 0 | } |
266 | |
|
267 | 0 | return header.size; |
268 | 0 | } |
269 | | |
270 | | // swap formatVersion 4 or 5 ----------------------------------------------- *** |
271 | | |
272 | | // The following are copied from CollationDataReader, trading an awkward copy of constants |
273 | | // for an awkward relocation of the i18n collationdatareader.h file into the common library. |
274 | | // Keep them in sync! |
275 | | |
276 | | enum { |
277 | | IX_INDEXES_LENGTH, // 0 |
278 | | IX_OPTIONS, |
279 | | IX_RESERVED2, |
280 | | IX_RESERVED3, |
281 | | |
282 | | IX_JAMO_CE32S_START, // 4 |
283 | | IX_REORDER_CODES_OFFSET, |
284 | | IX_REORDER_TABLE_OFFSET, |
285 | | IX_TRIE_OFFSET, |
286 | | |
287 | | IX_RESERVED8_OFFSET, // 8 |
288 | | IX_CES_OFFSET, |
289 | | IX_RESERVED10_OFFSET, |
290 | | IX_CE32S_OFFSET, |
291 | | |
292 | | IX_ROOT_ELEMENTS_OFFSET, // 12 |
293 | | IX_CONTEXTS_OFFSET, |
294 | | IX_UNSAFE_BWD_OFFSET, |
295 | | IX_FAST_LATIN_TABLE_OFFSET, |
296 | | |
297 | | IX_SCRIPTS_OFFSET, // 16 |
298 | | IX_COMPRESSIBLE_BYTES_OFFSET, |
299 | | IX_RESERVED18_OFFSET, |
300 | | IX_TOTAL_SIZE |
301 | | }; |
302 | | |
303 | | int32_t |
304 | | swapFormatVersion4(const UDataSwapper *ds, |
305 | | const void *inData, int32_t length, void *outData, |
306 | 0 | UErrorCode &errorCode) { |
307 | 0 | if(U_FAILURE(errorCode)) { return 0; } |
308 | | |
309 | 0 | const uint8_t *inBytes=(const uint8_t *)inData; |
310 | 0 | uint8_t *outBytes=(uint8_t *)outData; |
311 | |
|
312 | 0 | const int32_t *inIndexes=(const int32_t *)inBytes; |
313 | 0 | int32_t indexes[IX_TOTAL_SIZE+1]; |
314 | | |
315 | | // Need at least IX_INDEXES_LENGTH and IX_OPTIONS. |
316 | 0 | if(0<=length && length<8) { |
317 | 0 | udata_printError(ds, "ucol_swap(formatVersion=4): too few bytes " |
318 | 0 | "(%d after header) for collation data\n", |
319 | 0 | length); |
320 | 0 | errorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
321 | 0 | return 0; |
322 | 0 | } |
323 | | |
324 | 0 | int32_t indexesLength=indexes[0]=udata_readInt32(ds, inIndexes[0]); |
325 | 0 | if(0<=length && length<(indexesLength*4)) { |
326 | 0 | udata_printError(ds, "ucol_swap(formatVersion=4): too few bytes " |
327 | 0 | "(%d after header) for collation data\n", |
328 | 0 | length); |
329 | 0 | errorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
330 | 0 | return 0; |
331 | 0 | } |
332 | | |
333 | 0 | for(int32_t i=1; i<=IX_TOTAL_SIZE && i<indexesLength; ++i) { |
334 | 0 | indexes[i]=udata_readInt32(ds, inIndexes[i]); |
335 | 0 | } |
336 | 0 | for(int32_t i=indexesLength; i<=IX_TOTAL_SIZE; ++i) { |
337 | 0 | indexes[i]=-1; |
338 | 0 | } |
339 | 0 | inIndexes=NULL; // Make sure we do not accidentally use these instead of indexes[]. |
340 | | |
341 | | // Get the total length of the data. |
342 | 0 | int32_t size; |
343 | 0 | if(indexesLength>IX_TOTAL_SIZE) { |
344 | 0 | size=indexes[IX_TOTAL_SIZE]; |
345 | 0 | } else if(indexesLength>IX_REORDER_CODES_OFFSET) { |
346 | 0 | size=indexes[indexesLength-1]; |
347 | 0 | } else { |
348 | 0 | size=indexesLength*4; |
349 | 0 | } |
350 | 0 | if(length<0) { return size; } |
351 | | |
352 | 0 | if(length<size) { |
353 | 0 | udata_printError(ds, "ucol_swap(formatVersion=4): too few bytes " |
354 | 0 | "(%d after header) for collation data\n", |
355 | 0 | length); |
356 | 0 | errorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
357 | 0 | return 0; |
358 | 0 | } |
359 | | |
360 | | // Copy the data for inaccessible bytes and arrays of bytes. |
361 | 0 | if(inBytes!=outBytes) { |
362 | 0 | uprv_memcpy(outBytes, inBytes, size); |
363 | 0 | } |
364 | | |
365 | | // Swap the int32_t indexes[]. |
366 | 0 | ds->swapArray32(ds, inBytes, indexesLength * 4, outBytes, &errorCode); |
367 | | |
368 | | // The following is a modified version of CollationDataReader::read(). |
369 | | // Here we use indexes[] not inIndexes[] because |
370 | | // the inIndexes[] may not be in this machine's endianness. |
371 | 0 | int32_t index; // one of the indexes[] slots |
372 | 0 | int32_t offset; // byte offset for the index part |
373 | | // int32_t length; // number of bytes in the index part |
374 | |
|
375 | 0 | index = IX_REORDER_CODES_OFFSET; |
376 | 0 | offset = indexes[index]; |
377 | 0 | length = indexes[index + 1] - offset; |
378 | 0 | if(length > 0) { |
379 | 0 | ds->swapArray32(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
380 | 0 | } |
381 | | |
382 | | // Skip the IX_REORDER_TABLE_OFFSET byte array. |
383 | |
|
384 | 0 | index = IX_TRIE_OFFSET; |
385 | 0 | offset = indexes[index]; |
386 | 0 | length = indexes[index + 1] - offset; |
387 | 0 | if(length > 0) { |
388 | 0 | utrie2_swap(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
389 | 0 | } |
390 | |
|
391 | 0 | index = IX_RESERVED8_OFFSET; |
392 | 0 | offset = indexes[index]; |
393 | 0 | length = indexes[index + 1] - offset; |
394 | 0 | if(length > 0) { |
395 | 0 | udata_printError(ds, "ucol_swap(formatVersion=4): unknown data at IX_RESERVED8_OFFSET\n", length); |
396 | 0 | errorCode = U_UNSUPPORTED_ERROR; |
397 | 0 | return 0; |
398 | 0 | } |
399 | | |
400 | 0 | index = IX_CES_OFFSET; |
401 | 0 | offset = indexes[index]; |
402 | 0 | length = indexes[index + 1] - offset; |
403 | 0 | if(length > 0) { |
404 | 0 | ds->swapArray64(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
405 | 0 | } |
406 | |
|
407 | 0 | index = IX_RESERVED10_OFFSET; |
408 | 0 | offset = indexes[index]; |
409 | 0 | length = indexes[index + 1] - offset; |
410 | 0 | if(length > 0) { |
411 | 0 | udata_printError(ds, "ucol_swap(formatVersion=4): unknown data at IX_RESERVED10_OFFSET\n", length); |
412 | 0 | errorCode = U_UNSUPPORTED_ERROR; |
413 | 0 | return 0; |
414 | 0 | } |
415 | | |
416 | 0 | index = IX_CE32S_OFFSET; |
417 | 0 | offset = indexes[index]; |
418 | 0 | length = indexes[index + 1] - offset; |
419 | 0 | if(length > 0) { |
420 | 0 | ds->swapArray32(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
421 | 0 | } |
422 | |
|
423 | 0 | index = IX_ROOT_ELEMENTS_OFFSET; |
424 | 0 | offset = indexes[index]; |
425 | 0 | length = indexes[index + 1] - offset; |
426 | 0 | if(length > 0) { |
427 | 0 | ds->swapArray32(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
428 | 0 | } |
429 | |
|
430 | 0 | index = IX_CONTEXTS_OFFSET; |
431 | 0 | offset = indexes[index]; |
432 | 0 | length = indexes[index + 1] - offset; |
433 | 0 | if(length > 0) { |
434 | 0 | ds->swapArray16(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
435 | 0 | } |
436 | |
|
437 | 0 | index = IX_UNSAFE_BWD_OFFSET; |
438 | 0 | offset = indexes[index]; |
439 | 0 | length = indexes[index + 1] - offset; |
440 | 0 | if(length > 0) { |
441 | 0 | ds->swapArray16(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
442 | 0 | } |
443 | |
|
444 | 0 | index = IX_FAST_LATIN_TABLE_OFFSET; |
445 | 0 | offset = indexes[index]; |
446 | 0 | length = indexes[index + 1] - offset; |
447 | 0 | if(length > 0) { |
448 | 0 | ds->swapArray16(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
449 | 0 | } |
450 | |
|
451 | 0 | index = IX_SCRIPTS_OFFSET; |
452 | 0 | offset = indexes[index]; |
453 | 0 | length = indexes[index + 1] - offset; |
454 | 0 | if(length > 0) { |
455 | 0 | ds->swapArray16(ds, inBytes + offset, length, outBytes + offset, &errorCode); |
456 | 0 | } |
457 | | |
458 | | // Skip the IX_COMPRESSIBLE_BYTES_OFFSET byte array. |
459 | |
|
460 | 0 | index = IX_RESERVED18_OFFSET; |
461 | 0 | offset = indexes[index]; |
462 | 0 | length = indexes[index + 1] - offset; |
463 | 0 | if(length > 0) { |
464 | 0 | udata_printError(ds, "ucol_swap(formatVersion=4): unknown data at IX_RESERVED18_OFFSET\n", length); |
465 | 0 | errorCode = U_UNSUPPORTED_ERROR; |
466 | 0 | return 0; |
467 | 0 | } |
468 | | |
469 | 0 | return size; |
470 | 0 | } |
471 | | |
472 | | } // namespace |
473 | | |
474 | | /* swap ICU collation data like ucadata.icu */ |
475 | | U_CAPI int32_t U_EXPORT2 |
476 | | ucol_swap(const UDataSwapper *ds, |
477 | | const void *inData, int32_t length, void *outData, |
478 | 0 | UErrorCode *pErrorCode) { |
479 | 0 | if(U_FAILURE(*pErrorCode)) { return 0; } |
480 | | |
481 | | /* udata_swapDataHeader checks the arguments */ |
482 | 0 | int32_t headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); |
483 | 0 | if(U_FAILURE(*pErrorCode)) { |
484 | | // Try to swap the old format version which did not have a standard data header. |
485 | 0 | *pErrorCode=U_ZERO_ERROR; |
486 | 0 | return swapFormatVersion3(ds, inData, length, outData, pErrorCode); |
487 | 0 | } |
488 | | |
489 | | /* check data format and format version */ |
490 | 0 | const UDataInfo &info=*(const UDataInfo *)((const char *)inData+4); |
491 | 0 | if(!( |
492 | 0 | info.dataFormat[0]==0x55 && // dataFormat="UCol" |
493 | 0 | info.dataFormat[1]==0x43 && |
494 | 0 | info.dataFormat[2]==0x6f && |
495 | 0 | info.dataFormat[3]==0x6c && |
496 | 0 | (3<=info.formatVersion[0] && info.formatVersion[0]<=5) |
497 | 0 | )) { |
498 | 0 | udata_printError(ds, "ucol_swap(): data format %02x.%02x.%02x.%02x " |
499 | 0 | "(format version %02x.%02x) is not recognized as collation data\n", |
500 | 0 | info.dataFormat[0], info.dataFormat[1], |
501 | 0 | info.dataFormat[2], info.dataFormat[3], |
502 | 0 | info.formatVersion[0], info.formatVersion[1]); |
503 | 0 | *pErrorCode=U_UNSUPPORTED_ERROR; |
504 | 0 | return 0; |
505 | 0 | } |
506 | | |
507 | 0 | inData=(const char *)inData+headerSize; |
508 | 0 | if(length>=0) { length-=headerSize; } |
509 | 0 | outData=(char *)outData+headerSize; |
510 | 0 | int32_t collationSize; |
511 | 0 | if(info.formatVersion[0]>=4) { |
512 | 0 | collationSize=swapFormatVersion4(ds, inData, length, outData, *pErrorCode); |
513 | 0 | } else { |
514 | 0 | collationSize=swapFormatVersion3(ds, inData, length, outData, pErrorCode); |
515 | 0 | } |
516 | 0 | if(U_SUCCESS(*pErrorCode)) { |
517 | 0 | return headerSize+collationSize; |
518 | 0 | } else { |
519 | 0 | return 0; |
520 | 0 | } |
521 | 0 | } |
522 | | |
523 | | /* swap inverse UCA collation data (invuca.icu) */ |
524 | | U_CAPI int32_t U_EXPORT2 |
525 | | ucol_swapInverseUCA(const UDataSwapper *ds, |
526 | | const void *inData, int32_t length, void *outData, |
527 | 0 | UErrorCode *pErrorCode) { |
528 | 0 | const UDataInfo *pInfo; |
529 | 0 | int32_t headerSize; |
530 | |
|
531 | 0 | const uint8_t *inBytes; |
532 | 0 | uint8_t *outBytes; |
533 | |
|
534 | 0 | const InverseUCATableHeader *inHeader; |
535 | 0 | InverseUCATableHeader *outHeader; |
536 | 0 | InverseUCATableHeader header={ 0,0,0,0,0,{0,0,0,0},{0,0,0,0,0,0,0,0} }; |
537 | | |
538 | | /* udata_swapDataHeader checks the arguments */ |
539 | 0 | headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); |
540 | 0 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
541 | 0 | return 0; |
542 | 0 | } |
543 | | |
544 | | /* check data format and format version */ |
545 | 0 | pInfo=(const UDataInfo *)((const char *)inData+4); |
546 | 0 | if(!( |
547 | 0 | pInfo->dataFormat[0]==0x49 && /* dataFormat="InvC" */ |
548 | 0 | pInfo->dataFormat[1]==0x6e && |
549 | 0 | pInfo->dataFormat[2]==0x76 && |
550 | 0 | pInfo->dataFormat[3]==0x43 && |
551 | 0 | pInfo->formatVersion[0]==2 && |
552 | 0 | pInfo->formatVersion[1]>=1 |
553 | 0 | )) { |
554 | 0 | udata_printError(ds, "ucol_swapInverseUCA(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not an inverse UCA collation file\n", |
555 | 0 | pInfo->dataFormat[0], pInfo->dataFormat[1], |
556 | 0 | pInfo->dataFormat[2], pInfo->dataFormat[3], |
557 | 0 | pInfo->formatVersion[0], pInfo->formatVersion[1]); |
558 | 0 | *pErrorCode=U_UNSUPPORTED_ERROR; |
559 | 0 | return 0; |
560 | 0 | } |
561 | | |
562 | 0 | inBytes=(const uint8_t *)inData+headerSize; |
563 | 0 | outBytes=(uint8_t *)outData+headerSize; |
564 | |
|
565 | 0 | inHeader=(const InverseUCATableHeader *)inBytes; |
566 | 0 | outHeader=(InverseUCATableHeader *)outBytes; |
567 | | |
568 | | /* |
569 | | * The inverse UCA collation binary must contain at least the InverseUCATableHeader, |
570 | | * starting with its size field. |
571 | | * sizeof(UCATableHeader)==8*4 in ICU 2.8 |
572 | | * check the length against the header size before reading the size field |
573 | | */ |
574 | 0 | if(length<0) { |
575 | 0 | header.byteSize=udata_readInt32(ds, inHeader->byteSize); |
576 | 0 | } else if( |
577 | 0 | ((length-headerSize)<(8*4) || |
578 | 0 | (uint32_t)(length-headerSize)<(header.byteSize=udata_readInt32(ds, inHeader->byteSize))) |
579 | 0 | ) { |
580 | 0 | udata_printError(ds, "ucol_swapInverseUCA(): too few bytes (%d after header) for inverse UCA collation data\n", |
581 | 0 | length); |
582 | 0 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
583 | 0 | return 0; |
584 | 0 | } |
585 | | |
586 | 0 | if(length>=0) { |
587 | | /* copy everything, takes care of data that needs no swapping */ |
588 | 0 | if(inBytes!=outBytes) { |
589 | 0 | uprv_memcpy(outBytes, inBytes, header.byteSize); |
590 | 0 | } |
591 | | |
592 | | /* swap the necessary pieces in the order of their occurrence in the data */ |
593 | | |
594 | | /* read more of the InverseUCATableHeader (the byteSize field was read above) */ |
595 | 0 | header.tableSize= ds->readUInt32(inHeader->tableSize); |
596 | 0 | header.contsSize= ds->readUInt32(inHeader->contsSize); |
597 | 0 | header.table= ds->readUInt32(inHeader->table); |
598 | 0 | header.conts= ds->readUInt32(inHeader->conts); |
599 | | |
600 | | /* swap the 32-bit integers in the header */ |
601 | 0 | ds->swapArray32(ds, inHeader, 5*4, outHeader, pErrorCode); |
602 | | |
603 | | /* swap the inverse table; tableSize counts uint32_t[3] rows */ |
604 | 0 | ds->swapArray32(ds, inBytes+header.table, header.tableSize*3*4, |
605 | 0 | outBytes+header.table, pErrorCode); |
606 | | |
607 | | /* swap the continuation table; contsSize counts UChars */ |
608 | 0 | ds->swapArray16(ds, inBytes+header.conts, header.contsSize*U_SIZEOF_UCHAR, |
609 | 0 | outBytes+header.conts, pErrorCode); |
610 | 0 | } |
611 | |
|
612 | 0 | return headerSize+header.byteSize; |
613 | 0 | } |
614 | | |
615 | | #endif /* #if !UCONFIG_NO_COLLATION */ |