Jungshik Shin | 87232d8 | 2017-05-13 21:10:13 -0700 | [diff] [blame] | 1 | // © 2016 and later: Unicode, Inc. and others. |
Jungshik Shin | 5feb9ad | 2016-10-21 12:52:48 -0700 | [diff] [blame] | 2 | // License & terms of use: http://www.unicode.org/copyright.html |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 3 | /* |
| 4 | ****************************************************************************** |
| 5 | * |
| 6 | * Copyright (C) 2001-2012, International Business Machines |
| 7 | * Corporation and others. All Rights Reserved. |
| 8 | * |
| 9 | ****************************************************************************** |
| 10 | * file name: utrie.cpp |
Jungshik Shin | 87232d8 | 2017-05-13 21:10:13 -0700 | [diff] [blame] | 11 | * encoding: UTF-8 |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 12 | * tab size: 8 (not used) |
| 13 | * indentation:4 |
| 14 | * |
| 15 | * created on: 2001oct20 |
| 16 | * created by: Markus W. Scherer |
| 17 | * |
| 18 | * This is a common implementation of a "folded" trie. |
| 19 | * It is a kind of compressed, serializable table of 16- or 32-bit values associated with |
| 20 | * Unicode code points (0..0x10ffff). |
| 21 | */ |
| 22 | |
| 23 | #ifdef UTRIE_DEBUG |
| 24 | # include <stdio.h> |
| 25 | #endif |
| 26 | |
| 27 | #include "unicode/utypes.h" |
| 28 | #include "cmemory.h" |
| 29 | #include "utrie.h" |
| 30 | |
| 31 | /* miscellaneous ------------------------------------------------------------ */ |
| 32 | |
| 33 | #undef ABS |
| 34 | #define ABS(x) ((x)>=0 ? (x) : -(x)) |
| 35 | |
| 36 | static inline UBool |
| 37 | equal_uint32(const uint32_t *s, const uint32_t *t, int32_t length) { |
| 38 | while(length>0 && *s==*t) { |
| 39 | ++s; |
| 40 | ++t; |
| 41 | --length; |
| 42 | } |
| 43 | return (UBool)(length==0); |
| 44 | } |
| 45 | |
| 46 | /* Building a trie ----------------------------------------------------------*/ |
| 47 | |
| 48 | U_CAPI UNewTrie * U_EXPORT2 |
| 49 | utrie_open(UNewTrie *fillIn, |
| 50 | uint32_t *aliasData, int32_t maxDataLength, |
| 51 | uint32_t initialValue, uint32_t leadUnitValue, |
| 52 | UBool latin1Linear) { |
| 53 | UNewTrie *trie; |
| 54 | int32_t i, j; |
| 55 | |
| 56 | if( maxDataLength<UTRIE_DATA_BLOCK_LENGTH || |
| 57 | (latin1Linear && maxDataLength<1024) |
| 58 | ) { |
| 59 | return NULL; |
| 60 | } |
| 61 | |
| 62 | if(fillIn!=NULL) { |
| 63 | trie=fillIn; |
| 64 | } else { |
| 65 | trie=(UNewTrie *)uprv_malloc(sizeof(UNewTrie)); |
| 66 | if(trie==NULL) { |
| 67 | return NULL; |
| 68 | } |
| 69 | } |
| 70 | uprv_memset(trie, 0, sizeof(UNewTrie)); |
| 71 | trie->isAllocated= (UBool)(fillIn==NULL); |
| 72 | |
| 73 | if(aliasData!=NULL) { |
| 74 | trie->data=aliasData; |
| 75 | trie->isDataAllocated=FALSE; |
| 76 | } else { |
| 77 | trie->data=(uint32_t *)uprv_malloc(maxDataLength*4); |
| 78 | if(trie->data==NULL) { |
| 79 | uprv_free(trie); |
| 80 | return NULL; |
| 81 | } |
| 82 | trie->isDataAllocated=TRUE; |
| 83 | } |
| 84 | |
| 85 | /* preallocate and reset the first data block (block index 0) */ |
| 86 | j=UTRIE_DATA_BLOCK_LENGTH; |
| 87 | |
| 88 | if(latin1Linear) { |
| 89 | /* preallocate and reset the first block (number 0) and Latin-1 (U+0000..U+00ff) after that */ |
| 90 | /* made sure above that maxDataLength>=1024 */ |
| 91 | |
| 92 | /* set indexes to point to consecutive data blocks */ |
| 93 | i=0; |
| 94 | do { |
| 95 | /* do this at least for trie->index[0] even if that block is only partly used for Latin-1 */ |
| 96 | trie->index[i++]=j; |
| 97 | j+=UTRIE_DATA_BLOCK_LENGTH; |
| 98 | } while(i<(256>>UTRIE_SHIFT)); |
| 99 | } |
| 100 | |
| 101 | /* reset the initially allocated blocks to the initial value */ |
| 102 | trie->dataLength=j; |
| 103 | while(j>0) { |
| 104 | trie->data[--j]=initialValue; |
| 105 | } |
| 106 | |
| 107 | trie->leadUnitValue=leadUnitValue; |
| 108 | trie->indexLength=UTRIE_MAX_INDEX_LENGTH; |
| 109 | trie->dataCapacity=maxDataLength; |
| 110 | trie->isLatin1Linear=latin1Linear; |
| 111 | trie->isCompacted=FALSE; |
| 112 | return trie; |
| 113 | } |
| 114 | |
| 115 | U_CAPI UNewTrie * U_EXPORT2 |
| 116 | utrie_clone(UNewTrie *fillIn, const UNewTrie *other, uint32_t *aliasData, int32_t aliasDataCapacity) { |
| 117 | UNewTrie *trie; |
| 118 | UBool isDataAllocated; |
| 119 | |
| 120 | /* do not clone if other is not valid or already compacted */ |
| 121 | if(other==NULL || other->data==NULL || other->isCompacted) { |
| 122 | return NULL; |
| 123 | } |
| 124 | |
| 125 | /* clone data */ |
| 126 | if(aliasData!=NULL && aliasDataCapacity>=other->dataCapacity) { |
| 127 | isDataAllocated=FALSE; |
| 128 | } else { |
| 129 | aliasDataCapacity=other->dataCapacity; |
| 130 | aliasData=(uint32_t *)uprv_malloc(other->dataCapacity*4); |
| 131 | if(aliasData==NULL) { |
| 132 | return NULL; |
| 133 | } |
| 134 | isDataAllocated=TRUE; |
| 135 | } |
| 136 | |
| 137 | trie=utrie_open(fillIn, aliasData, aliasDataCapacity, |
| 138 | other->data[0], other->leadUnitValue, |
| 139 | other->isLatin1Linear); |
| 140 | if(trie==NULL) { |
| 141 | uprv_free(aliasData); |
| 142 | } else { |
| 143 | uprv_memcpy(trie->index, other->index, sizeof(trie->index)); |
Jungshik Shin | 5feb9ad | 2016-10-21 12:52:48 -0700 | [diff] [blame] | 144 | uprv_memcpy(trie->data, other->data, (size_t)other->dataLength*4); |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 145 | trie->dataLength=other->dataLength; |
| 146 | trie->isDataAllocated=isDataAllocated; |
| 147 | } |
| 148 | |
| 149 | return trie; |
| 150 | } |
| 151 | |
| 152 | U_CAPI void U_EXPORT2 |
| 153 | utrie_close(UNewTrie *trie) { |
| 154 | if(trie!=NULL) { |
| 155 | if(trie->isDataAllocated) { |
| 156 | uprv_free(trie->data); |
| 157 | trie->data=NULL; |
| 158 | } |
| 159 | if(trie->isAllocated) { |
| 160 | uprv_free(trie); |
| 161 | } |
| 162 | } |
| 163 | } |
| 164 | |
| 165 | U_CAPI uint32_t * U_EXPORT2 |
| 166 | utrie_getData(UNewTrie *trie, int32_t *pLength) { |
| 167 | if(trie==NULL || pLength==NULL) { |
| 168 | return NULL; |
| 169 | } |
| 170 | |
| 171 | *pLength=trie->dataLength; |
| 172 | return trie->data; |
| 173 | } |
| 174 | |
| 175 | static int32_t |
| 176 | utrie_allocDataBlock(UNewTrie *trie) { |
| 177 | int32_t newBlock, newTop; |
| 178 | |
| 179 | newBlock=trie->dataLength; |
| 180 | newTop=newBlock+UTRIE_DATA_BLOCK_LENGTH; |
| 181 | if(newTop>trie->dataCapacity) { |
| 182 | /* out of memory in the data array */ |
| 183 | return -1; |
| 184 | } |
| 185 | trie->dataLength=newTop; |
| 186 | return newBlock; |
| 187 | } |
| 188 | |
| 189 | /** |
| 190 | * No error checking for illegal arguments. |
| 191 | * |
| 192 | * @return -1 if no new data block available (out of memory in data array) |
| 193 | * @internal |
| 194 | */ |
| 195 | static int32_t |
| 196 | utrie_getDataBlock(UNewTrie *trie, UChar32 c) { |
| 197 | int32_t indexValue, newBlock; |
| 198 | |
| 199 | c>>=UTRIE_SHIFT; |
| 200 | indexValue=trie->index[c]; |
| 201 | if(indexValue>0) { |
| 202 | return indexValue; |
| 203 | } |
| 204 | |
| 205 | /* allocate a new data block */ |
| 206 | newBlock=utrie_allocDataBlock(trie); |
| 207 | if(newBlock<0) { |
| 208 | /* out of memory in the data array */ |
| 209 | return -1; |
| 210 | } |
| 211 | trie->index[c]=newBlock; |
| 212 | |
| 213 | /* copy-on-write for a block from a setRange() */ |
| 214 | uprv_memcpy(trie->data+newBlock, trie->data-indexValue, 4*UTRIE_DATA_BLOCK_LENGTH); |
| 215 | return newBlock; |
| 216 | } |
| 217 | |
| 218 | /** |
| 219 | * @return TRUE if the value was successfully set |
| 220 | */ |
| 221 | U_CAPI UBool U_EXPORT2 |
| 222 | utrie_set32(UNewTrie *trie, UChar32 c, uint32_t value) { |
| 223 | int32_t block; |
| 224 | |
| 225 | /* valid, uncompacted trie and valid c? */ |
| 226 | if(trie==NULL || trie->isCompacted || (uint32_t)c>0x10ffff) { |
| 227 | return FALSE; |
| 228 | } |
| 229 | |
| 230 | block=utrie_getDataBlock(trie, c); |
| 231 | if(block<0) { |
| 232 | return FALSE; |
| 233 | } |
| 234 | |
| 235 | trie->data[block+(c&UTRIE_MASK)]=value; |
| 236 | return TRUE; |
| 237 | } |
| 238 | |
| 239 | U_CAPI uint32_t U_EXPORT2 |
| 240 | utrie_get32(UNewTrie *trie, UChar32 c, UBool *pInBlockZero) { |
| 241 | int32_t block; |
| 242 | |
| 243 | /* valid, uncompacted trie and valid c? */ |
| 244 | if(trie==NULL || trie->isCompacted || (uint32_t)c>0x10ffff) { |
| 245 | if(pInBlockZero!=NULL) { |
| 246 | *pInBlockZero=TRUE; |
| 247 | } |
| 248 | return 0; |
| 249 | } |
| 250 | |
| 251 | block=trie->index[c>>UTRIE_SHIFT]; |
| 252 | if(pInBlockZero!=NULL) { |
| 253 | *pInBlockZero= (UBool)(block==0); |
| 254 | } |
| 255 | |
| 256 | return trie->data[ABS(block)+(c&UTRIE_MASK)]; |
| 257 | } |
| 258 | |
| 259 | /** |
| 260 | * @internal |
| 261 | */ |
| 262 | static void |
| 263 | utrie_fillBlock(uint32_t *block, UChar32 start, UChar32 limit, |
| 264 | uint32_t value, uint32_t initialValue, UBool overwrite) { |
| 265 | uint32_t *pLimit; |
| 266 | |
| 267 | pLimit=block+limit; |
| 268 | block+=start; |
| 269 | if(overwrite) { |
| 270 | while(block<pLimit) { |
| 271 | *block++=value; |
| 272 | } |
| 273 | } else { |
| 274 | while(block<pLimit) { |
| 275 | if(*block==initialValue) { |
| 276 | *block=value; |
| 277 | } |
| 278 | ++block; |
| 279 | } |
| 280 | } |
| 281 | } |
| 282 | |
| 283 | U_CAPI UBool U_EXPORT2 |
| 284 | utrie_setRange32(UNewTrie *trie, UChar32 start, UChar32 limit, uint32_t value, UBool overwrite) { |
| 285 | /* |
| 286 | * repeat value in [start..limit[ |
| 287 | * mark index values for repeat-data blocks by setting bit 31 of the index values |
| 288 | * fill around existing values if any, if(overwrite) |
| 289 | */ |
| 290 | uint32_t initialValue; |
| 291 | int32_t block, rest, repeatBlock; |
| 292 | |
| 293 | /* valid, uncompacted trie and valid indexes? */ |
| 294 | if( trie==NULL || trie->isCompacted || |
| 295 | (uint32_t)start>0x10ffff || (uint32_t)limit>0x110000 || start>limit |
| 296 | ) { |
| 297 | return FALSE; |
| 298 | } |
| 299 | if(start==limit) { |
| 300 | return TRUE; /* nothing to do */ |
| 301 | } |
| 302 | |
| 303 | initialValue=trie->data[0]; |
| 304 | if(start&UTRIE_MASK) { |
| 305 | UChar32 nextStart; |
| 306 | |
| 307 | /* set partial block at [start..following block boundary[ */ |
| 308 | block=utrie_getDataBlock(trie, start); |
| 309 | if(block<0) { |
| 310 | return FALSE; |
| 311 | } |
| 312 | |
| 313 | nextStart=(start+UTRIE_DATA_BLOCK_LENGTH)&~UTRIE_MASK; |
| 314 | if(nextStart<=limit) { |
| 315 | utrie_fillBlock(trie->data+block, start&UTRIE_MASK, UTRIE_DATA_BLOCK_LENGTH, |
| 316 | value, initialValue, overwrite); |
| 317 | start=nextStart; |
| 318 | } else { |
| 319 | utrie_fillBlock(trie->data+block, start&UTRIE_MASK, limit&UTRIE_MASK, |
| 320 | value, initialValue, overwrite); |
| 321 | return TRUE; |
| 322 | } |
| 323 | } |
| 324 | |
| 325 | /* number of positions in the last, partial block */ |
| 326 | rest=limit&UTRIE_MASK; |
| 327 | |
| 328 | /* round down limit to a block boundary */ |
| 329 | limit&=~UTRIE_MASK; |
| 330 | |
| 331 | /* iterate over all-value blocks */ |
| 332 | if(value==initialValue) { |
| 333 | repeatBlock=0; |
| 334 | } else { |
| 335 | repeatBlock=-1; |
| 336 | } |
| 337 | while(start<limit) { |
| 338 | /* get index value */ |
| 339 | block=trie->index[start>>UTRIE_SHIFT]; |
| 340 | if(block>0) { |
| 341 | /* already allocated, fill in value */ |
| 342 | utrie_fillBlock(trie->data+block, 0, UTRIE_DATA_BLOCK_LENGTH, value, initialValue, overwrite); |
| 343 | } else if(trie->data[-block]!=value && (block==0 || overwrite)) { |
| 344 | /* set the repeatBlock instead of the current block 0 or range block */ |
| 345 | if(repeatBlock>=0) { |
| 346 | trie->index[start>>UTRIE_SHIFT]=-repeatBlock; |
| 347 | } else { |
| 348 | /* create and set and fill the repeatBlock */ |
| 349 | repeatBlock=utrie_getDataBlock(trie, start); |
| 350 | if(repeatBlock<0) { |
| 351 | return FALSE; |
| 352 | } |
| 353 | |
| 354 | /* set the negative block number to indicate that it is a repeat block */ |
| 355 | trie->index[start>>UTRIE_SHIFT]=-repeatBlock; |
| 356 | utrie_fillBlock(trie->data+repeatBlock, 0, UTRIE_DATA_BLOCK_LENGTH, value, initialValue, TRUE); |
| 357 | } |
| 358 | } |
| 359 | |
| 360 | start+=UTRIE_DATA_BLOCK_LENGTH; |
| 361 | } |
| 362 | |
| 363 | if(rest>0) { |
| 364 | /* set partial block at [last block boundary..limit[ */ |
| 365 | block=utrie_getDataBlock(trie, start); |
| 366 | if(block<0) { |
| 367 | return FALSE; |
| 368 | } |
| 369 | |
| 370 | utrie_fillBlock(trie->data+block, 0, rest, value, initialValue, overwrite); |
| 371 | } |
| 372 | |
| 373 | return TRUE; |
| 374 | } |
| 375 | |
| 376 | static int32_t |
| 377 | _findSameIndexBlock(const int32_t *idx, int32_t indexLength, |
| 378 | int32_t otherBlock) { |
| 379 | int32_t block, i; |
| 380 | |
| 381 | for(block=UTRIE_BMP_INDEX_LENGTH; block<indexLength; block+=UTRIE_SURROGATE_BLOCK_COUNT) { |
| 382 | for(i=0; i<UTRIE_SURROGATE_BLOCK_COUNT; ++i) { |
| 383 | if(idx[block+i]!=idx[otherBlock+i]) { |
| 384 | break; |
| 385 | } |
| 386 | } |
| 387 | if(i==UTRIE_SURROGATE_BLOCK_COUNT) { |
| 388 | return block; |
| 389 | } |
| 390 | } |
| 391 | return indexLength; |
| 392 | } |
| 393 | |
| 394 | /* |
| 395 | * Fold the normalization data for supplementary code points into |
| 396 | * a compact area on top of the BMP-part of the trie index, |
| 397 | * with the lead surrogates indexing this compact area. |
| 398 | * |
| 399 | * Duplicate the index values for lead surrogates: |
| 400 | * From inside the BMP area, where some may be overridden with folded values, |
| 401 | * to just after the BMP area, where they can be retrieved for |
| 402 | * code point lookups. |
| 403 | */ |
| 404 | static void |
| 405 | utrie_fold(UNewTrie *trie, UNewTrieGetFoldedValue *getFoldedValue, UErrorCode *pErrorCode) { |
| 406 | int32_t leadIndexes[UTRIE_SURROGATE_BLOCK_COUNT]; |
| 407 | int32_t *idx; |
| 408 | uint32_t value; |
| 409 | UChar32 c; |
| 410 | int32_t indexLength, block; |
| 411 | #ifdef UTRIE_DEBUG |
| 412 | int countLeadCUWithData=0; |
| 413 | #endif |
| 414 | |
| 415 | idx=trie->index; |
| 416 | |
| 417 | /* copy the lead surrogate indexes into a temporary array */ |
| 418 | uprv_memcpy(leadIndexes, idx+(0xd800>>UTRIE_SHIFT), 4*UTRIE_SURROGATE_BLOCK_COUNT); |
| 419 | |
| 420 | /* |
| 421 | * set all values for lead surrogate code *units* to leadUnitValue |
| 422 | * so that, by default, runtime lookups will find no data for associated |
| 423 | * supplementary code points, unless there is data for such code points |
| 424 | * which will result in a non-zero folding value below that is set for |
| 425 | * the respective lead units |
| 426 | * |
| 427 | * the above saved the indexes for surrogate code *points* |
| 428 | * fill the indexes with simplified code from utrie_setRange32() |
| 429 | */ |
| 430 | if(trie->leadUnitValue==trie->data[0]) { |
| 431 | block=0; /* leadUnitValue==initialValue, use all-initial-value block */ |
| 432 | } else { |
| 433 | /* create and fill the repeatBlock */ |
| 434 | block=utrie_allocDataBlock(trie); |
| 435 | if(block<0) { |
| 436 | /* data table overflow */ |
| 437 | *pErrorCode=U_MEMORY_ALLOCATION_ERROR; |
| 438 | return; |
| 439 | } |
| 440 | utrie_fillBlock(trie->data+block, 0, UTRIE_DATA_BLOCK_LENGTH, trie->leadUnitValue, trie->data[0], TRUE); |
| 441 | block=-block; /* negative block number to indicate that it is a repeat block */ |
| 442 | } |
| 443 | for(c=(0xd800>>UTRIE_SHIFT); c<(0xdc00>>UTRIE_SHIFT); ++c) { |
| 444 | trie->index[c]=block; |
| 445 | } |
| 446 | |
| 447 | /* |
| 448 | * Fold significant index values into the area just after the BMP indexes. |
| 449 | * In case the first lead surrogate has significant data, |
| 450 | * its index block must be used first (in which case the folding is a no-op). |
| 451 | * Later all folded index blocks are moved up one to insert the copied |
| 452 | * lead surrogate indexes. |
| 453 | */ |
| 454 | indexLength=UTRIE_BMP_INDEX_LENGTH; |
| 455 | |
| 456 | /* search for any index (stage 1) entries for supplementary code points */ |
| 457 | for(c=0x10000; c<0x110000;) { |
| 458 | if(idx[c>>UTRIE_SHIFT]!=0) { |
| 459 | /* there is data, treat the full block for a lead surrogate */ |
| 460 | c&=~0x3ff; |
| 461 | |
| 462 | #ifdef UTRIE_DEBUG |
| 463 | ++countLeadCUWithData; |
| 464 | /* printf("supplementary data for lead surrogate U+%04lx\n", (long)(0xd7c0+(c>>10))); */ |
| 465 | #endif |
| 466 | |
| 467 | /* is there an identical index block? */ |
| 468 | block=_findSameIndexBlock(idx, indexLength, c>>UTRIE_SHIFT); |
| 469 | |
| 470 | /* |
| 471 | * get a folded value for [c..c+0x400[ and, |
| 472 | * if different from the value for the lead surrogate code point, |
| 473 | * set it for the lead surrogate code unit |
| 474 | */ |
| 475 | value=getFoldedValue(trie, c, block+UTRIE_SURROGATE_BLOCK_COUNT); |
| 476 | if(value!=utrie_get32(trie, U16_LEAD(c), NULL)) { |
| 477 | if(!utrie_set32(trie, U16_LEAD(c), value)) { |
| 478 | /* data table overflow */ |
| 479 | *pErrorCode=U_MEMORY_ALLOCATION_ERROR; |
| 480 | return; |
| 481 | } |
| 482 | |
| 483 | /* if we did not find an identical index block... */ |
| 484 | if(block==indexLength) { |
| 485 | /* move the actual index (stage 1) entries from the supplementary position to the new one */ |
| 486 | uprv_memmove(idx+indexLength, |
| 487 | idx+(c>>UTRIE_SHIFT), |
| 488 | 4*UTRIE_SURROGATE_BLOCK_COUNT); |
| 489 | indexLength+=UTRIE_SURROGATE_BLOCK_COUNT; |
| 490 | } |
| 491 | } |
| 492 | c+=0x400; |
| 493 | } else { |
| 494 | c+=UTRIE_DATA_BLOCK_LENGTH; |
| 495 | } |
| 496 | } |
| 497 | #ifdef UTRIE_DEBUG |
| 498 | if(countLeadCUWithData>0) { |
| 499 | printf("supplementary data for %d lead surrogates\n", countLeadCUWithData); |
| 500 | } |
| 501 | #endif |
| 502 | |
| 503 | /* |
| 504 | * index array overflow? |
| 505 | * This is to guarantee that a folding offset is of the form |
| 506 | * UTRIE_BMP_INDEX_LENGTH+n*UTRIE_SURROGATE_BLOCK_COUNT with n=0..1023. |
| 507 | * If the index is too large, then n>=1024 and more than 10 bits are necessary. |
| 508 | * |
| 509 | * In fact, it can only ever become n==1024 with completely unfoldable data and |
| 510 | * the additional block of duplicated values for lead surrogates. |
| 511 | */ |
| 512 | if(indexLength>=UTRIE_MAX_INDEX_LENGTH) { |
| 513 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
| 514 | return; |
| 515 | } |
| 516 | |
| 517 | /* |
| 518 | * make space for the lead surrogate index block and |
| 519 | * insert it between the BMP indexes and the folded ones |
| 520 | */ |
| 521 | uprv_memmove(idx+UTRIE_BMP_INDEX_LENGTH+UTRIE_SURROGATE_BLOCK_COUNT, |
| 522 | idx+UTRIE_BMP_INDEX_LENGTH, |
| 523 | 4*(indexLength-UTRIE_BMP_INDEX_LENGTH)); |
| 524 | uprv_memcpy(idx+UTRIE_BMP_INDEX_LENGTH, |
| 525 | leadIndexes, |
| 526 | 4*UTRIE_SURROGATE_BLOCK_COUNT); |
| 527 | indexLength+=UTRIE_SURROGATE_BLOCK_COUNT; |
| 528 | |
| 529 | #ifdef UTRIE_DEBUG |
| 530 | printf("trie index count: BMP %ld all Unicode %ld folded %ld\n", |
| 531 | UTRIE_BMP_INDEX_LENGTH, (long)UTRIE_MAX_INDEX_LENGTH, indexLength); |
| 532 | #endif |
| 533 | |
| 534 | trie->indexLength=indexLength; |
| 535 | } |
| 536 | |
| 537 | /* |
| 538 | * Set a value in the trie index map to indicate which data block |
| 539 | * is referenced and which one is not. |
| 540 | * utrie_compact() will remove data blocks that are not used at all. |
| 541 | * Set |
| 542 | * - 0 if it is used |
| 543 | * - -1 if it is not used |
| 544 | */ |
| 545 | static void |
| 546 | _findUnusedBlocks(UNewTrie *trie) { |
| 547 | int32_t i; |
| 548 | |
| 549 | /* fill the entire map with "not used" */ |
| 550 | uprv_memset(trie->map, 0xff, (UTRIE_MAX_BUILD_TIME_DATA_LENGTH>>UTRIE_SHIFT)*4); |
| 551 | |
| 552 | /* mark each block that _is_ used with 0 */ |
| 553 | for(i=0; i<trie->indexLength; ++i) { |
| 554 | trie->map[ABS(trie->index[i])>>UTRIE_SHIFT]=0; |
| 555 | } |
| 556 | |
| 557 | /* never move the all-initial-value block 0 */ |
| 558 | trie->map[0]=0; |
| 559 | } |
| 560 | |
| 561 | static int32_t |
| 562 | _findSameDataBlock(const uint32_t *data, int32_t dataLength, |
| 563 | int32_t otherBlock, int32_t step) { |
| 564 | int32_t block; |
| 565 | |
| 566 | /* ensure that we do not even partially get past dataLength */ |
| 567 | dataLength-=UTRIE_DATA_BLOCK_LENGTH; |
| 568 | |
| 569 | for(block=0; block<=dataLength; block+=step) { |
| 570 | if(equal_uint32(data+block, data+otherBlock, UTRIE_DATA_BLOCK_LENGTH)) { |
| 571 | return block; |
| 572 | } |
| 573 | } |
| 574 | return -1; |
| 575 | } |
| 576 | |
| 577 | /* |
| 578 | * Compact a folded build-time trie. |
| 579 | * |
| 580 | * The compaction |
| 581 | * - removes blocks that are identical with earlier ones |
| 582 | * - overlaps adjacent blocks as much as possible (if overlap==TRUE) |
| 583 | * - moves blocks in steps of the data granularity |
| 584 | * - moves and overlaps blocks that overlap with multiple values in the overlap region |
| 585 | * |
| 586 | * It does not |
| 587 | * - try to move and overlap blocks that are not already adjacent |
| 588 | */ |
| 589 | static void |
| 590 | utrie_compact(UNewTrie *trie, UBool overlap, UErrorCode *pErrorCode) { |
| 591 | int32_t i, start, newStart, overlapStart; |
| 592 | |
| 593 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 594 | return; |
| 595 | } |
| 596 | |
| 597 | /* valid, uncompacted trie? */ |
| 598 | if(trie==NULL) { |
| 599 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 600 | return; |
| 601 | } |
| 602 | if(trie->isCompacted) { |
| 603 | return; /* nothing left to do */ |
| 604 | } |
| 605 | |
| 606 | /* compaction */ |
| 607 | |
| 608 | /* initialize the index map with "block is used/unused" flags */ |
| 609 | _findUnusedBlocks(trie); |
| 610 | |
| 611 | /* if Latin-1 is preallocated and linear, then do not compact Latin-1 data */ |
| 612 | if(trie->isLatin1Linear && UTRIE_SHIFT<=8) { |
| 613 | overlapStart=UTRIE_DATA_BLOCK_LENGTH+256; |
| 614 | } else { |
| 615 | overlapStart=UTRIE_DATA_BLOCK_LENGTH; |
| 616 | } |
| 617 | |
| 618 | newStart=UTRIE_DATA_BLOCK_LENGTH; |
| 619 | for(start=newStart; start<trie->dataLength;) { |
| 620 | /* |
| 621 | * start: index of first entry of current block |
| 622 | * newStart: index where the current block is to be moved |
| 623 | * (right after current end of already-compacted data) |
| 624 | */ |
| 625 | |
| 626 | /* skip blocks that are not used */ |
| 627 | if(trie->map[start>>UTRIE_SHIFT]<0) { |
| 628 | /* advance start to the next block */ |
| 629 | start+=UTRIE_DATA_BLOCK_LENGTH; |
| 630 | |
| 631 | /* leave newStart with the previous block! */ |
| 632 | continue; |
| 633 | } |
| 634 | |
| 635 | /* search for an identical block */ |
| 636 | if( start>=overlapStart && |
| 637 | (i=_findSameDataBlock(trie->data, newStart, start, |
| 638 | overlap ? UTRIE_DATA_GRANULARITY : UTRIE_DATA_BLOCK_LENGTH)) |
| 639 | >=0 |
| 640 | ) { |
| 641 | /* found an identical block, set the other block's index value for the current block */ |
| 642 | trie->map[start>>UTRIE_SHIFT]=i; |
| 643 | |
| 644 | /* advance start to the next block */ |
| 645 | start+=UTRIE_DATA_BLOCK_LENGTH; |
| 646 | |
| 647 | /* leave newStart with the previous block! */ |
| 648 | continue; |
| 649 | } |
| 650 | |
| 651 | /* see if the beginning of this block can be overlapped with the end of the previous block */ |
| 652 | if(overlap && start>=overlapStart) { |
| 653 | /* look for maximum overlap (modulo granularity) with the previous, adjacent block */ |
| 654 | for(i=UTRIE_DATA_BLOCK_LENGTH-UTRIE_DATA_GRANULARITY; |
| 655 | i>0 && !equal_uint32(trie->data+(newStart-i), trie->data+start, i); |
| 656 | i-=UTRIE_DATA_GRANULARITY) {} |
| 657 | } else { |
| 658 | i=0; |
| 659 | } |
| 660 | |
| 661 | if(i>0) { |
| 662 | /* some overlap */ |
| 663 | trie->map[start>>UTRIE_SHIFT]=newStart-i; |
| 664 | |
| 665 | /* move the non-overlapping indexes to their new positions */ |
| 666 | start+=i; |
| 667 | for(i=UTRIE_DATA_BLOCK_LENGTH-i; i>0; --i) { |
| 668 | trie->data[newStart++]=trie->data[start++]; |
| 669 | } |
| 670 | } else if(newStart<start) { |
| 671 | /* no overlap, just move the indexes to their new positions */ |
| 672 | trie->map[start>>UTRIE_SHIFT]=newStart; |
| 673 | for(i=UTRIE_DATA_BLOCK_LENGTH; i>0; --i) { |
| 674 | trie->data[newStart++]=trie->data[start++]; |
| 675 | } |
| 676 | } else /* no overlap && newStart==start */ { |
| 677 | trie->map[start>>UTRIE_SHIFT]=start; |
| 678 | newStart+=UTRIE_DATA_BLOCK_LENGTH; |
| 679 | start=newStart; |
| 680 | } |
| 681 | } |
| 682 | |
| 683 | /* now adjust the index (stage 1) table */ |
| 684 | for(i=0; i<trie->indexLength; ++i) { |
| 685 | trie->index[i]=trie->map[ABS(trie->index[i])>>UTRIE_SHIFT]; |
| 686 | } |
| 687 | |
| 688 | #ifdef UTRIE_DEBUG |
| 689 | /* we saved some space */ |
| 690 | printf("compacting trie: count of 32-bit words %lu->%lu\n", |
| 691 | (long)trie->dataLength, (long)newStart); |
| 692 | #endif |
| 693 | |
| 694 | trie->dataLength=newStart; |
| 695 | } |
| 696 | |
| 697 | /* serialization ------------------------------------------------------------ */ |
| 698 | |
| 699 | /* |
| 700 | * Default function for the folding value: |
| 701 | * Just store the offset (16 bits) if there is any non-initial-value entry. |
| 702 | * |
| 703 | * The offset parameter is never 0. |
| 704 | * Returning the offset itself is safe for UTRIE_SHIFT>=5 because |
| 705 | * for UTRIE_SHIFT==5 the maximum index length is UTRIE_MAX_INDEX_LENGTH==0x8800 |
| 706 | * which fits into 16-bit trie values; |
| 707 | * for higher UTRIE_SHIFT, UTRIE_MAX_INDEX_LENGTH decreases. |
| 708 | * |
| 709 | * Theoretically, it would be safer for all possible UTRIE_SHIFT including |
| 710 | * those of 4 and lower to return offset>>UTRIE_SURROGATE_BLOCK_BITS |
| 711 | * which would always result in a value of 0x40..0x43f |
| 712 | * (start/end 1k blocks of supplementary Unicode code points). |
| 713 | * However, this would be uglier, and would not work for some existing |
| 714 | * binary data file formats. |
| 715 | * |
| 716 | * Also, we do not plan to change UTRIE_SHIFT because it would change binary |
| 717 | * data file formats, and we would probably not make it smaller because of |
| 718 | * the then even larger BMP index length even for empty tries. |
| 719 | */ |
| 720 | static uint32_t U_CALLCONV |
| 721 | defaultGetFoldedValue(UNewTrie *trie, UChar32 start, int32_t offset) { |
| 722 | uint32_t value, initialValue; |
| 723 | UChar32 limit; |
| 724 | UBool inBlockZero; |
| 725 | |
| 726 | initialValue=trie->data[0]; |
| 727 | limit=start+0x400; |
| 728 | while(start<limit) { |
| 729 | value=utrie_get32(trie, start, &inBlockZero); |
| 730 | if(inBlockZero) { |
| 731 | start+=UTRIE_DATA_BLOCK_LENGTH; |
| 732 | } else if(value!=initialValue) { |
| 733 | return (uint32_t)offset; |
| 734 | } else { |
| 735 | ++start; |
| 736 | } |
| 737 | } |
| 738 | return 0; |
| 739 | } |
| 740 | |
| 741 | U_CAPI int32_t U_EXPORT2 |
| 742 | utrie_serialize(UNewTrie *trie, void *dt, int32_t capacity, |
| 743 | UNewTrieGetFoldedValue *getFoldedValue, |
| 744 | UBool reduceTo16Bits, |
| 745 | UErrorCode *pErrorCode) { |
| 746 | UTrieHeader *header; |
| 747 | uint32_t *p; |
| 748 | uint16_t *dest16; |
| 749 | int32_t i, length; |
| 750 | uint8_t* data = NULL; |
| 751 | |
| 752 | /* argument check */ |
| 753 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 754 | return 0; |
| 755 | } |
| 756 | |
| 757 | if(trie==NULL || capacity<0 || (capacity>0 && dt==NULL)) { |
| 758 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 759 | return 0; |
| 760 | } |
| 761 | if(getFoldedValue==NULL) { |
| 762 | getFoldedValue=defaultGetFoldedValue; |
| 763 | } |
| 764 | |
| 765 | data = (uint8_t*)dt; |
| 766 | /* fold and compact if necessary, also checks that indexLength is within limits */ |
| 767 | if(!trie->isCompacted) { |
| 768 | /* compact once without overlap to improve folding */ |
| 769 | utrie_compact(trie, FALSE, pErrorCode); |
| 770 | |
| 771 | /* fold the supplementary part of the index array */ |
| 772 | utrie_fold(trie, getFoldedValue, pErrorCode); |
| 773 | |
| 774 | /* compact again with overlap for minimum data array length */ |
| 775 | utrie_compact(trie, TRUE, pErrorCode); |
| 776 | |
| 777 | trie->isCompacted=TRUE; |
| 778 | if(U_FAILURE(*pErrorCode)) { |
| 779 | return 0; |
| 780 | } |
| 781 | } |
| 782 | |
| 783 | /* is dataLength within limits? */ |
| 784 | if( (reduceTo16Bits ? (trie->dataLength+trie->indexLength) : trie->dataLength) >= UTRIE_MAX_DATA_LENGTH) { |
| 785 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
| 786 | } |
| 787 | |
| 788 | length=sizeof(UTrieHeader)+2*trie->indexLength; |
| 789 | if(reduceTo16Bits) { |
| 790 | length+=2*trie->dataLength; |
| 791 | } else { |
| 792 | length+=4*trie->dataLength; |
| 793 | } |
| 794 | |
| 795 | if(length>capacity) { |
| 796 | return length; /* preflighting */ |
| 797 | } |
| 798 | |
| 799 | #ifdef UTRIE_DEBUG |
| 800 | printf("**UTrieLengths(serialize)** index:%6ld data:%6ld serialized:%6ld\n", |
| 801 | (long)trie->indexLength, (long)trie->dataLength, (long)length); |
| 802 | #endif |
| 803 | |
| 804 | /* set the header fields */ |
| 805 | header=(UTrieHeader *)data; |
| 806 | data+=sizeof(UTrieHeader); |
| 807 | |
| 808 | header->signature=0x54726965; /* "Trie" */ |
| 809 | header->options=UTRIE_SHIFT | (UTRIE_INDEX_SHIFT<<UTRIE_OPTIONS_INDEX_SHIFT); |
| 810 | |
| 811 | if(!reduceTo16Bits) { |
| 812 | header->options|=UTRIE_OPTIONS_DATA_IS_32_BIT; |
| 813 | } |
| 814 | if(trie->isLatin1Linear) { |
| 815 | header->options|=UTRIE_OPTIONS_LATIN1_IS_LINEAR; |
| 816 | } |
| 817 | |
| 818 | header->indexLength=trie->indexLength; |
| 819 | header->dataLength=trie->dataLength; |
| 820 | |
| 821 | /* write the index (stage 1) array and the 16/32-bit data (stage 2) array */ |
| 822 | if(reduceTo16Bits) { |
| 823 | /* write 16-bit index values shifted right by UTRIE_INDEX_SHIFT, after adding indexLength */ |
| 824 | p=(uint32_t *)trie->index; |
| 825 | dest16=(uint16_t *)data; |
| 826 | for(i=trie->indexLength; i>0; --i) { |
| 827 | *dest16++=(uint16_t)((*p++ + trie->indexLength)>>UTRIE_INDEX_SHIFT); |
| 828 | } |
| 829 | |
| 830 | /* write 16-bit data values */ |
| 831 | p=trie->data; |
| 832 | for(i=trie->dataLength; i>0; --i) { |
| 833 | *dest16++=(uint16_t)*p++; |
| 834 | } |
| 835 | } else { |
| 836 | /* write 16-bit index values shifted right by UTRIE_INDEX_SHIFT */ |
| 837 | p=(uint32_t *)trie->index; |
| 838 | dest16=(uint16_t *)data; |
| 839 | for(i=trie->indexLength; i>0; --i) { |
| 840 | *dest16++=(uint16_t)(*p++ >> UTRIE_INDEX_SHIFT); |
| 841 | } |
| 842 | |
| 843 | /* write 32-bit data values */ |
Jungshik Shin | 5feb9ad | 2016-10-21 12:52:48 -0700 | [diff] [blame] | 844 | uprv_memcpy(dest16, trie->data, 4*(size_t)trie->dataLength); |
jshin@chromium.org | 6f31ac3 | 2014-03-26 22:15:14 +0000 | [diff] [blame] | 845 | } |
| 846 | |
| 847 | return length; |
| 848 | } |
| 849 | |
| 850 | /* inverse to defaultGetFoldedValue() */ |
| 851 | U_CAPI int32_t U_EXPORT2 |
| 852 | utrie_defaultGetFoldingOffset(uint32_t data) { |
| 853 | return (int32_t)data; |
| 854 | } |
| 855 | |
| 856 | U_CAPI int32_t U_EXPORT2 |
| 857 | utrie_unserialize(UTrie *trie, const void *data, int32_t length, UErrorCode *pErrorCode) { |
| 858 | const UTrieHeader *header; |
| 859 | const uint16_t *p16; |
| 860 | uint32_t options; |
| 861 | |
| 862 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 863 | return -1; |
| 864 | } |
| 865 | |
| 866 | /* enough data for a trie header? */ |
| 867 | if(length<(int32_t)sizeof(UTrieHeader)) { |
| 868 | *pErrorCode=U_INVALID_FORMAT_ERROR; |
| 869 | return -1; |
| 870 | } |
| 871 | |
| 872 | /* check the signature */ |
| 873 | header=(const UTrieHeader *)data; |
| 874 | if(header->signature!=0x54726965) { |
| 875 | *pErrorCode=U_INVALID_FORMAT_ERROR; |
| 876 | return -1; |
| 877 | } |
| 878 | |
| 879 | /* get the options and check the shift values */ |
| 880 | options=header->options; |
| 881 | if( (options&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_SHIFT || |
| 882 | ((options>>UTRIE_OPTIONS_INDEX_SHIFT)&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_INDEX_SHIFT |
| 883 | ) { |
| 884 | *pErrorCode=U_INVALID_FORMAT_ERROR; |
| 885 | return -1; |
| 886 | } |
| 887 | trie->isLatin1Linear= (UBool)((options&UTRIE_OPTIONS_LATIN1_IS_LINEAR)!=0); |
| 888 | |
| 889 | /* get the length values */ |
| 890 | trie->indexLength=header->indexLength; |
| 891 | trie->dataLength=header->dataLength; |
| 892 | |
| 893 | length-=(int32_t)sizeof(UTrieHeader); |
| 894 | |
| 895 | /* enough data for the index? */ |
| 896 | if(length<2*trie->indexLength) { |
| 897 | *pErrorCode=U_INVALID_FORMAT_ERROR; |
| 898 | return -1; |
| 899 | } |
| 900 | p16=(const uint16_t *)(header+1); |
| 901 | trie->index=p16; |
| 902 | p16+=trie->indexLength; |
| 903 | length-=2*trie->indexLength; |
| 904 | |
| 905 | /* get the data */ |
| 906 | if(options&UTRIE_OPTIONS_DATA_IS_32_BIT) { |
| 907 | if(length<4*trie->dataLength) { |
| 908 | *pErrorCode=U_INVALID_FORMAT_ERROR; |
| 909 | return -1; |
| 910 | } |
| 911 | trie->data32=(const uint32_t *)p16; |
| 912 | trie->initialValue=trie->data32[0]; |
| 913 | length=(int32_t)sizeof(UTrieHeader)+2*trie->indexLength+4*trie->dataLength; |
| 914 | } else { |
| 915 | if(length<2*trie->dataLength) { |
| 916 | *pErrorCode=U_INVALID_FORMAT_ERROR; |
| 917 | return -1; |
| 918 | } |
| 919 | |
| 920 | /* the "data16" data is used via the index pointer */ |
| 921 | trie->data32=NULL; |
| 922 | trie->initialValue=trie->index[trie->indexLength]; |
| 923 | length=(int32_t)sizeof(UTrieHeader)+2*trie->indexLength+2*trie->dataLength; |
| 924 | } |
| 925 | |
| 926 | trie->getFoldingOffset=utrie_defaultGetFoldingOffset; |
| 927 | |
| 928 | return length; |
| 929 | } |
| 930 | |
| 931 | U_CAPI int32_t U_EXPORT2 |
| 932 | utrie_unserializeDummy(UTrie *trie, |
| 933 | void *data, int32_t length, |
| 934 | uint32_t initialValue, uint32_t leadUnitValue, |
| 935 | UBool make16BitTrie, |
| 936 | UErrorCode *pErrorCode) { |
| 937 | uint16_t *p16; |
| 938 | int32_t actualLength, latin1Length, i, limit; |
| 939 | uint16_t block; |
| 940 | |
| 941 | if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { |
| 942 | return -1; |
| 943 | } |
| 944 | |
| 945 | /* calculate the actual size of the dummy trie data */ |
| 946 | |
| 947 | /* max(Latin-1, block 0) */ |
| 948 | latin1Length= 256; /*UTRIE_SHIFT<=8 ? 256 : UTRIE_DATA_BLOCK_LENGTH;*/ |
| 949 | |
| 950 | trie->indexLength=UTRIE_BMP_INDEX_LENGTH+UTRIE_SURROGATE_BLOCK_COUNT; |
| 951 | trie->dataLength=latin1Length; |
| 952 | if(leadUnitValue!=initialValue) { |
| 953 | trie->dataLength+=UTRIE_DATA_BLOCK_LENGTH; |
| 954 | } |
| 955 | |
| 956 | actualLength=trie->indexLength*2; |
| 957 | if(make16BitTrie) { |
| 958 | actualLength+=trie->dataLength*2; |
| 959 | } else { |
| 960 | actualLength+=trie->dataLength*4; |
| 961 | } |
| 962 | |
| 963 | /* enough space for the dummy trie? */ |
| 964 | if(length<actualLength) { |
| 965 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
| 966 | return actualLength; |
| 967 | } |
| 968 | |
| 969 | trie->isLatin1Linear=TRUE; |
| 970 | trie->initialValue=initialValue; |
| 971 | |
| 972 | /* fill the index and data arrays */ |
| 973 | p16=(uint16_t *)data; |
| 974 | trie->index=p16; |
| 975 | |
| 976 | if(make16BitTrie) { |
| 977 | /* indexes to block 0 */ |
| 978 | block=(uint16_t)(trie->indexLength>>UTRIE_INDEX_SHIFT); |
| 979 | limit=trie->indexLength; |
| 980 | for(i=0; i<limit; ++i) { |
| 981 | p16[i]=block; |
| 982 | } |
| 983 | |
| 984 | if(leadUnitValue!=initialValue) { |
| 985 | /* indexes for lead surrogate code units to the block after Latin-1 */ |
| 986 | block+=(uint16_t)(latin1Length>>UTRIE_INDEX_SHIFT); |
| 987 | i=0xd800>>UTRIE_SHIFT; |
| 988 | limit=0xdc00>>UTRIE_SHIFT; |
| 989 | for(; i<limit; ++i) { |
| 990 | p16[i]=block; |
| 991 | } |
| 992 | } |
| 993 | |
| 994 | trie->data32=NULL; |
| 995 | |
| 996 | /* Latin-1 data */ |
| 997 | p16+=trie->indexLength; |
| 998 | for(i=0; i<latin1Length; ++i) { |
| 999 | p16[i]=(uint16_t)initialValue; |
| 1000 | } |
| 1001 | |
| 1002 | /* data for lead surrogate code units */ |
| 1003 | if(leadUnitValue!=initialValue) { |
| 1004 | limit=latin1Length+UTRIE_DATA_BLOCK_LENGTH; |
| 1005 | for(/* i=latin1Length */; i<limit; ++i) { |
| 1006 | p16[i]=(uint16_t)leadUnitValue; |
| 1007 | } |
| 1008 | } |
| 1009 | } else { |
| 1010 | uint32_t *p32; |
| 1011 | |
| 1012 | /* indexes to block 0 */ |
| 1013 | uprv_memset(p16, 0, trie->indexLength*2); |
| 1014 | |
| 1015 | if(leadUnitValue!=initialValue) { |
| 1016 | /* indexes for lead surrogate code units to the block after Latin-1 */ |
| 1017 | block=(uint16_t)(latin1Length>>UTRIE_INDEX_SHIFT); |
| 1018 | i=0xd800>>UTRIE_SHIFT; |
| 1019 | limit=0xdc00>>UTRIE_SHIFT; |
| 1020 | for(; i<limit; ++i) { |
| 1021 | p16[i]=block; |
| 1022 | } |
| 1023 | } |
| 1024 | |
| 1025 | trie->data32=p32=(uint32_t *)(p16+trie->indexLength); |
| 1026 | |
| 1027 | /* Latin-1 data */ |
| 1028 | for(i=0; i<latin1Length; ++i) { |
| 1029 | p32[i]=initialValue; |
| 1030 | } |
| 1031 | |
| 1032 | /* data for lead surrogate code units */ |
| 1033 | if(leadUnitValue!=initialValue) { |
| 1034 | limit=latin1Length+UTRIE_DATA_BLOCK_LENGTH; |
| 1035 | for(/* i=latin1Length */; i<limit; ++i) { |
| 1036 | p32[i]=leadUnitValue; |
| 1037 | } |
| 1038 | } |
| 1039 | } |
| 1040 | |
| 1041 | trie->getFoldingOffset=utrie_defaultGetFoldingOffset; |
| 1042 | |
| 1043 | return actualLength; |
| 1044 | } |
| 1045 | |
| 1046 | /* enumeration -------------------------------------------------------------- */ |
| 1047 | |
| 1048 | /* default UTrieEnumValue() returns the input value itself */ |
| 1049 | static uint32_t U_CALLCONV |
| 1050 | enumSameValue(const void * /*context*/, uint32_t value) { |
| 1051 | return value; |
| 1052 | } |
| 1053 | |
| 1054 | /** |
| 1055 | * Enumerate all ranges of code points with the same relevant values. |
| 1056 | * The values are transformed from the raw trie entries by the enumValue function. |
| 1057 | */ |
| 1058 | U_CAPI void U_EXPORT2 |
| 1059 | utrie_enum(const UTrie *trie, |
| 1060 | UTrieEnumValue *enumValue, UTrieEnumRange *enumRange, const void *context) { |
| 1061 | const uint32_t *data32; |
| 1062 | const uint16_t *idx; |
| 1063 | |
| 1064 | uint32_t value, prevValue, initialValue; |
| 1065 | UChar32 c, prev; |
| 1066 | int32_t l, i, j, block, prevBlock, nullBlock, offset; |
| 1067 | |
| 1068 | /* check arguments */ |
| 1069 | if(trie==NULL || trie->index==NULL || enumRange==NULL) { |
| 1070 | return; |
| 1071 | } |
| 1072 | if(enumValue==NULL) { |
| 1073 | enumValue=enumSameValue; |
| 1074 | } |
| 1075 | |
| 1076 | idx=trie->index; |
| 1077 | data32=trie->data32; |
| 1078 | |
| 1079 | /* get the enumeration value that corresponds to an initial-value trie data entry */ |
| 1080 | initialValue=enumValue(context, trie->initialValue); |
| 1081 | |
| 1082 | if(data32==NULL) { |
| 1083 | nullBlock=trie->indexLength; |
| 1084 | } else { |
| 1085 | nullBlock=0; |
| 1086 | } |
| 1087 | |
| 1088 | /* set variables for previous range */ |
| 1089 | prevBlock=nullBlock; |
| 1090 | prev=0; |
| 1091 | prevValue=initialValue; |
| 1092 | |
| 1093 | /* enumerate BMP - the main loop enumerates data blocks */ |
| 1094 | for(i=0, c=0; c<=0xffff; ++i) { |
| 1095 | if(c==0xd800) { |
| 1096 | /* skip lead surrogate code _units_, go to lead surr. code _points_ */ |
| 1097 | i=UTRIE_BMP_INDEX_LENGTH; |
| 1098 | } else if(c==0xdc00) { |
| 1099 | /* go back to regular BMP code points */ |
| 1100 | i=c>>UTRIE_SHIFT; |
| 1101 | } |
| 1102 | |
| 1103 | block=idx[i]<<UTRIE_INDEX_SHIFT; |
| 1104 | if(block==prevBlock) { |
| 1105 | /* the block is the same as the previous one, and filled with value */ |
| 1106 | c+=UTRIE_DATA_BLOCK_LENGTH; |
| 1107 | } else if(block==nullBlock) { |
| 1108 | /* this is the all-initial-value block */ |
| 1109 | if(prevValue!=initialValue) { |
| 1110 | if(prev<c) { |
| 1111 | if(!enumRange(context, prev, c, prevValue)) { |
| 1112 | return; |
| 1113 | } |
| 1114 | } |
| 1115 | prevBlock=nullBlock; |
| 1116 | prev=c; |
| 1117 | prevValue=initialValue; |
| 1118 | } |
| 1119 | c+=UTRIE_DATA_BLOCK_LENGTH; |
| 1120 | } else { |
| 1121 | prevBlock=block; |
| 1122 | for(j=0; j<UTRIE_DATA_BLOCK_LENGTH; ++j) { |
| 1123 | value=enumValue(context, data32!=NULL ? data32[block+j] : idx[block+j]); |
| 1124 | if(value!=prevValue) { |
| 1125 | if(prev<c) { |
| 1126 | if(!enumRange(context, prev, c, prevValue)) { |
| 1127 | return; |
| 1128 | } |
| 1129 | } |
| 1130 | if(j>0) { |
| 1131 | /* the block is not filled with all the same value */ |
| 1132 | prevBlock=-1; |
| 1133 | } |
| 1134 | prev=c; |
| 1135 | prevValue=value; |
| 1136 | } |
| 1137 | ++c; |
| 1138 | } |
| 1139 | } |
| 1140 | } |
| 1141 | |
| 1142 | /* enumerate supplementary code points */ |
| 1143 | for(l=0xd800; l<0xdc00;) { |
| 1144 | /* lead surrogate access */ |
| 1145 | offset=idx[l>>UTRIE_SHIFT]<<UTRIE_INDEX_SHIFT; |
| 1146 | if(offset==nullBlock) { |
| 1147 | /* no entries for a whole block of lead surrogates */ |
| 1148 | if(prevValue!=initialValue) { |
| 1149 | if(prev<c) { |
| 1150 | if(!enumRange(context, prev, c, prevValue)) { |
| 1151 | return; |
| 1152 | } |
| 1153 | } |
| 1154 | prevBlock=nullBlock; |
| 1155 | prev=c; |
| 1156 | prevValue=initialValue; |
| 1157 | } |
| 1158 | |
| 1159 | l+=UTRIE_DATA_BLOCK_LENGTH; |
| 1160 | c+=UTRIE_DATA_BLOCK_LENGTH<<10; |
| 1161 | continue; |
| 1162 | } |
| 1163 | |
| 1164 | value= data32!=NULL ? data32[offset+(l&UTRIE_MASK)] : idx[offset+(l&UTRIE_MASK)]; |
| 1165 | |
| 1166 | /* enumerate trail surrogates for this lead surrogate */ |
| 1167 | offset=trie->getFoldingOffset(value); |
| 1168 | if(offset<=0) { |
| 1169 | /* no data for this lead surrogate */ |
| 1170 | if(prevValue!=initialValue) { |
| 1171 | if(prev<c) { |
| 1172 | if(!enumRange(context, prev, c, prevValue)) { |
| 1173 | return; |
| 1174 | } |
| 1175 | } |
| 1176 | prevBlock=nullBlock; |
| 1177 | prev=c; |
| 1178 | prevValue=initialValue; |
| 1179 | } |
| 1180 | |
| 1181 | /* nothing else to do for the supplementary code points for this lead surrogate */ |
| 1182 | c+=0x400; |
| 1183 | } else { |
| 1184 | /* enumerate code points for this lead surrogate */ |
| 1185 | i=offset; |
| 1186 | offset+=UTRIE_SURROGATE_BLOCK_COUNT; |
| 1187 | do { |
| 1188 | /* copy of most of the body of the BMP loop */ |
| 1189 | block=idx[i]<<UTRIE_INDEX_SHIFT; |
| 1190 | if(block==prevBlock) { |
| 1191 | /* the block is the same as the previous one, and filled with value */ |
| 1192 | c+=UTRIE_DATA_BLOCK_LENGTH; |
| 1193 | } else if(block==nullBlock) { |
| 1194 | /* this is the all-initial-value block */ |
| 1195 | if(prevValue!=initialValue) { |
| 1196 | if(prev<c) { |
| 1197 | if(!enumRange(context, prev, c, prevValue)) { |
| 1198 | return; |
| 1199 | } |
| 1200 | } |
| 1201 | prevBlock=nullBlock; |
| 1202 | prev=c; |
| 1203 | prevValue=initialValue; |
| 1204 | } |
| 1205 | c+=UTRIE_DATA_BLOCK_LENGTH; |
| 1206 | } else { |
| 1207 | prevBlock=block; |
| 1208 | for(j=0; j<UTRIE_DATA_BLOCK_LENGTH; ++j) { |
| 1209 | value=enumValue(context, data32!=NULL ? data32[block+j] : idx[block+j]); |
| 1210 | if(value!=prevValue) { |
| 1211 | if(prev<c) { |
| 1212 | if(!enumRange(context, prev, c, prevValue)) { |
| 1213 | return; |
| 1214 | } |
| 1215 | } |
| 1216 | if(j>0) { |
| 1217 | /* the block is not filled with all the same value */ |
| 1218 | prevBlock=-1; |
| 1219 | } |
| 1220 | prev=c; |
| 1221 | prevValue=value; |
| 1222 | } |
| 1223 | ++c; |
| 1224 | } |
| 1225 | } |
| 1226 | } while(++i<offset); |
| 1227 | } |
| 1228 | |
| 1229 | ++l; |
| 1230 | } |
| 1231 | |
| 1232 | /* deliver last range */ |
| 1233 | enumRange(context, prev, c, prevValue); |
| 1234 | } |