| File: | out/../deps/icu-small/source/common/utext.cpp |
| Warning: | line 2547, column 13 Value stored to 'limit32' is never read |
Press '?' to see keyboard shortcuts
Keyboard shortcuts:
| 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) 2005-2016, International Business Machines |
| 7 | * Corporation and others. All Rights Reserved. |
| 8 | * |
| 9 | ******************************************************************************* |
| 10 | * file name: utext.cpp |
| 11 | * encoding: UTF-8 |
| 12 | * tab size: 8 (not used) |
| 13 | * indentation:4 |
| 14 | * |
| 15 | * created on: 2005apr12 |
| 16 | * created by: Markus W. Scherer |
| 17 | */ |
| 18 | |
| 19 | #include <cstddef> |
| 20 | |
| 21 | #include "unicode/utypes.h" |
| 22 | #include "unicode/ustring.h" |
| 23 | #include "unicode/unistr.h" |
| 24 | #include "unicode/chariter.h" |
| 25 | #include "unicode/utext.h" |
| 26 | #include "unicode/utf.h" |
| 27 | #include "unicode/utf8.h" |
| 28 | #include "unicode/utf16.h" |
| 29 | #include "ustr_imp.h" |
| 30 | #include "cmemory.h" |
| 31 | #include "cstring.h" |
| 32 | #include "uassert.h" |
| 33 | #include "putilimp.h" |
| 34 | |
| 35 | U_NAMESPACE_USEusing namespace icu_71; |
| 36 | |
| 37 | #define I32_FLAG(bitIndex)((int32_t)1<<(bitIndex)) ((int32_t)1<<(bitIndex)) |
| 38 | |
| 39 | |
| 40 | static UBool |
| 41 | utext_access(UText *ut, int64_t index, UBool forward) { |
| 42 | return ut->pFuncs->access(ut, index, forward); |
| 43 | } |
| 44 | |
| 45 | |
| 46 | |
| 47 | U_CAPIextern "C" UBool U_EXPORT2 |
| 48 | utext_moveIndex32utext_moveIndex32_71(UText *ut, int32_t delta) { |
| 49 | UChar32 c; |
| 50 | if (delta > 0) { |
| 51 | do { |
| 52 | if(ut->chunkOffset>=ut->chunkLength && !utext_access(ut, ut->chunkNativeLimit, TRUE1)) { |
| 53 | return FALSE0; |
| 54 | } |
| 55 | c = ut->chunkContents[ut->chunkOffset]; |
| 56 | if (U16_IS_SURROGATE(c)(((c)&0xfffff800)==0xd800)) { |
| 57 | c = utext_next32utext_next32_71(ut); |
| 58 | if (c == U_SENTINEL(-1)) { |
| 59 | return FALSE0; |
| 60 | } |
| 61 | } else { |
| 62 | ut->chunkOffset++; |
| 63 | } |
| 64 | } while(--delta>0); |
| 65 | |
| 66 | } else if (delta<0) { |
| 67 | do { |
| 68 | if(ut->chunkOffset<=0 && !utext_access(ut, ut->chunkNativeStart, FALSE0)) { |
| 69 | return FALSE0; |
| 70 | } |
| 71 | c = ut->chunkContents[ut->chunkOffset-1]; |
| 72 | if (U16_IS_SURROGATE(c)(((c)&0xfffff800)==0xd800)) { |
| 73 | c = utext_previous32utext_previous32_71(ut); |
| 74 | if (c == U_SENTINEL(-1)) { |
| 75 | return FALSE0; |
| 76 | } |
| 77 | } else { |
| 78 | ut->chunkOffset--; |
| 79 | } |
| 80 | } while(++delta<0); |
| 81 | } |
| 82 | |
| 83 | return TRUE1; |
| 84 | } |
| 85 | |
| 86 | |
| 87 | U_CAPIextern "C" int64_t U_EXPORT2 |
| 88 | utext_nativeLengthutext_nativeLength_71(UText *ut) { |
| 89 | return ut->pFuncs->nativeLength(ut); |
| 90 | } |
| 91 | |
| 92 | |
| 93 | U_CAPIextern "C" UBool U_EXPORT2 |
| 94 | utext_isLengthExpensiveutext_isLengthExpensive_71(const UText *ut) { |
| 95 | UBool r = (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE))) != 0; |
| 96 | return r; |
| 97 | } |
| 98 | |
| 99 | |
| 100 | U_CAPIextern "C" int64_t U_EXPORT2 |
| 101 | utext_getNativeIndexutext_getNativeIndex_71(const UText *ut) { |
| 102 | if(ut->chunkOffset <= ut->nativeIndexingLimit) { |
| 103 | return ut->chunkNativeStart+ut->chunkOffset; |
| 104 | } else { |
| 105 | return ut->pFuncs->mapOffsetToNative(ut); |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | |
| 110 | U_CAPIextern "C" void U_EXPORT2 |
| 111 | utext_setNativeIndexutext_setNativeIndex_71(UText *ut, int64_t index) { |
| 112 | if(index<ut->chunkNativeStart || index>=ut->chunkNativeLimit) { |
| 113 | // The desired position is outside of the current chunk. |
| 114 | // Access the new position. Assume a forward iteration from here, |
| 115 | // which will also be optimimum for a single random access. |
| 116 | // Reverse iterations may suffer slightly. |
| 117 | ut->pFuncs->access(ut, index, TRUE1); |
| 118 | } else if((int32_t)(index - ut->chunkNativeStart) <= ut->nativeIndexingLimit) { |
| 119 | // utf-16 indexing. |
| 120 | ut->chunkOffset=(int32_t)(index-ut->chunkNativeStart); |
| 121 | } else { |
| 122 | ut->chunkOffset=ut->pFuncs->mapNativeIndexToUTF16(ut, index); |
| 123 | } |
| 124 | // The convention is that the index must always be on a code point boundary. |
| 125 | // Adjust the index position if it is in the middle of a surrogate pair. |
| 126 | if (ut->chunkOffset<ut->chunkLength) { |
| 127 | UChar c= ut->chunkContents[ut->chunkOffset]; |
| 128 | if (U16_IS_TRAIL(c)(((c)&0xfffffc00)==0xdc00)) { |
| 129 | if (ut->chunkOffset==0) { |
| 130 | ut->pFuncs->access(ut, ut->chunkNativeStart, FALSE0); |
| 131 | } |
| 132 | if (ut->chunkOffset>0) { |
| 133 | UChar lead = ut->chunkContents[ut->chunkOffset-1]; |
| 134 | if (U16_IS_LEAD(lead)(((lead)&0xfffffc00)==0xd800)) { |
| 135 | ut->chunkOffset--; |
| 136 | } |
| 137 | } |
| 138 | } |
| 139 | } |
| 140 | } |
| 141 | |
| 142 | |
| 143 | |
| 144 | U_CAPIextern "C" int64_t U_EXPORT2 |
| 145 | utext_getPreviousNativeIndexutext_getPreviousNativeIndex_71(UText *ut) { |
| 146 | // |
| 147 | // Fast-path the common case. |
| 148 | // Common means current position is not at the beginning of a chunk |
| 149 | // and the preceding character is not supplementary. |
| 150 | // |
| 151 | int32_t i = ut->chunkOffset - 1; |
| 152 | int64_t result; |
| 153 | if (i >= 0) { |
| 154 | UChar c = ut->chunkContents[i]; |
| 155 | if (U16_IS_TRAIL(c)(((c)&0xfffffc00)==0xdc00) == FALSE0) { |
| 156 | if (i <= ut->nativeIndexingLimit) { |
| 157 | result = ut->chunkNativeStart + i; |
| 158 | } else { |
| 159 | ut->chunkOffset = i; |
| 160 | result = ut->pFuncs->mapOffsetToNative(ut); |
| 161 | ut->chunkOffset++; |
| 162 | } |
| 163 | return result; |
| 164 | } |
| 165 | } |
| 166 | |
| 167 | // If at the start of text, simply return 0. |
| 168 | if (ut->chunkOffset==0 && ut->chunkNativeStart==0) { |
| 169 | return 0; |
| 170 | } |
| 171 | |
| 172 | // Harder, less common cases. We are at a chunk boundary, or on a surrogate. |
| 173 | // Keep it simple, use other functions to handle the edges. |
| 174 | // |
| 175 | utext_previous32utext_previous32_71(ut); |
| 176 | result = UTEXT_GETNATIVEINDEX(ut)((ut)->chunkOffset <= (ut)->nativeIndexingLimit? (ut )->chunkNativeStart+(ut)->chunkOffset : (ut)->pFuncs ->mapOffsetToNative(ut)); |
| 177 | utext_next32utext_next32_71(ut); |
| 178 | return result; |
| 179 | } |
| 180 | |
| 181 | |
| 182 | // |
| 183 | // utext_current32. Get the UChar32 at the current position. |
| 184 | // UText iteration position is always on a code point boundary, |
| 185 | // never on the trail half of a surrogate pair. |
| 186 | // |
| 187 | U_CAPIextern "C" UChar32 U_EXPORT2 |
| 188 | utext_current32utext_current32_71(UText *ut) { |
| 189 | UChar32 c; |
| 190 | if (ut->chunkOffset==ut->chunkLength) { |
| 191 | // Current position is just off the end of the chunk. |
| 192 | if (ut->pFuncs->access(ut, ut->chunkNativeLimit, TRUE1) == FALSE0) { |
| 193 | // Off the end of the text. |
| 194 | return U_SENTINEL(-1); |
| 195 | } |
| 196 | } |
| 197 | |
| 198 | c = ut->chunkContents[ut->chunkOffset]; |
| 199 | if (U16_IS_LEAD(c)(((c)&0xfffffc00)==0xd800) == FALSE0) { |
| 200 | // Normal, non-supplementary case. |
| 201 | return c; |
| 202 | } |
| 203 | |
| 204 | // |
| 205 | // Possible supplementary char. |
| 206 | // |
| 207 | UChar32 trail = 0; |
| 208 | UChar32 supplementaryC = c; |
| 209 | if ((ut->chunkOffset+1) < ut->chunkLength) { |
| 210 | // The trail surrogate is in the same chunk. |
| 211 | trail = ut->chunkContents[ut->chunkOffset+1]; |
| 212 | } else { |
| 213 | // The trail surrogate is in a different chunk. |
| 214 | // Because we must maintain the iteration position, we need to switch forward |
| 215 | // into the new chunk, get the trail surrogate, then revert the chunk back to the |
| 216 | // original one. |
| 217 | // An edge case to be careful of: the entire text may end with an unpaired |
| 218 | // leading surrogate. The attempt to access the trail will fail, but |
| 219 | // the original position before the unpaired lead still needs to be restored. |
| 220 | int64_t nativePosition = ut->chunkNativeLimit; |
| 221 | int32_t originalOffset = ut->chunkOffset; |
| 222 | if (ut->pFuncs->access(ut, nativePosition, TRUE1)) { |
| 223 | trail = ut->chunkContents[ut->chunkOffset]; |
| 224 | } |
| 225 | UBool r = ut->pFuncs->access(ut, nativePosition, FALSE0); // reverse iteration flag loads preceding chunk |
| 226 | U_ASSERT(r==TRUE)(void)0; |
| 227 | ut->chunkOffset = originalOffset; |
| 228 | if(!r) { |
| 229 | return U_SENTINEL(-1); |
| 230 | } |
| 231 | } |
| 232 | |
| 233 | if (U16_IS_TRAIL(trail)(((trail)&0xfffffc00)==0xdc00)) { |
| 234 | supplementaryC = U16_GET_SUPPLEMENTARY(c, trail)(((UChar32)(c)<<10UL)+(UChar32)(trail)-((0xd800<< 10UL)+0xdc00-0x10000)); |
| 235 | } |
| 236 | return supplementaryC; |
| 237 | |
| 238 | } |
| 239 | |
| 240 | |
| 241 | U_CAPIextern "C" UChar32 U_EXPORT2 |
| 242 | utext_char32Atutext_char32At_71(UText *ut, int64_t nativeIndex) { |
| 243 | UChar32 c = U_SENTINEL(-1); |
| 244 | |
| 245 | // Fast path the common case. |
| 246 | if (nativeIndex>=ut->chunkNativeStart && nativeIndex < ut->chunkNativeStart + ut->nativeIndexingLimit) { |
| 247 | ut->chunkOffset = (int32_t)(nativeIndex - ut->chunkNativeStart); |
| 248 | c = ut->chunkContents[ut->chunkOffset]; |
| 249 | if (U16_IS_SURROGATE(c)(((c)&0xfffff800)==0xd800) == FALSE0) { |
| 250 | return c; |
| 251 | } |
| 252 | } |
| 253 | |
| 254 | |
| 255 | utext_setNativeIndexutext_setNativeIndex_71(ut, nativeIndex); |
| 256 | if (nativeIndex>=ut->chunkNativeStart && ut->chunkOffset<ut->chunkLength) { |
| 257 | c = ut->chunkContents[ut->chunkOffset]; |
| 258 | if (U16_IS_SURROGATE(c)(((c)&0xfffff800)==0xd800)) { |
| 259 | // For surrogates, let current32() deal with the complications |
| 260 | // of supplementaries that may span chunk boundaries. |
| 261 | c = utext_current32utext_current32_71(ut); |
| 262 | } |
| 263 | } |
| 264 | return c; |
| 265 | } |
| 266 | |
| 267 | |
| 268 | U_CAPIextern "C" UChar32 U_EXPORT2 |
| 269 | utext_next32utext_next32_71(UText *ut) { |
| 270 | UChar32 c; |
| 271 | |
| 272 | if (ut->chunkOffset >= ut->chunkLength) { |
| 273 | if (ut->pFuncs->access(ut, ut->chunkNativeLimit, TRUE1) == FALSE0) { |
| 274 | return U_SENTINEL(-1); |
| 275 | } |
| 276 | } |
| 277 | |
| 278 | c = ut->chunkContents[ut->chunkOffset++]; |
| 279 | if (U16_IS_LEAD(c)(((c)&0xfffffc00)==0xd800) == FALSE0) { |
| 280 | // Normal case, not supplementary. |
| 281 | // (A trail surrogate seen here is just returned as is, as a surrogate value. |
| 282 | // It cannot be part of a pair.) |
| 283 | return c; |
| 284 | } |
| 285 | |
| 286 | if (ut->chunkOffset >= ut->chunkLength) { |
| 287 | if (ut->pFuncs->access(ut, ut->chunkNativeLimit, TRUE1) == FALSE0) { |
| 288 | // c is an unpaired lead surrogate at the end of the text. |
| 289 | // return it as it is. |
| 290 | return c; |
| 291 | } |
| 292 | } |
| 293 | UChar32 trail = ut->chunkContents[ut->chunkOffset]; |
| 294 | if (U16_IS_TRAIL(trail)(((trail)&0xfffffc00)==0xdc00) == FALSE0) { |
| 295 | // c was an unpaired lead surrogate, not at the end of the text. |
| 296 | // return it as it is (unpaired). Iteration position is on the |
| 297 | // following character, possibly in the next chunk, where the |
| 298 | // trail surrogate would have been if it had existed. |
| 299 | return c; |
| 300 | } |
| 301 | |
| 302 | UChar32 supplementary = U16_GET_SUPPLEMENTARY(c, trail)(((UChar32)(c)<<10UL)+(UChar32)(trail)-((0xd800<< 10UL)+0xdc00-0x10000)); |
| 303 | ut->chunkOffset++; // move iteration position over the trail surrogate. |
| 304 | return supplementary; |
| 305 | } |
| 306 | |
| 307 | |
| 308 | U_CAPIextern "C" UChar32 U_EXPORT2 |
| 309 | utext_previous32utext_previous32_71(UText *ut) { |
| 310 | UChar32 c; |
| 311 | |
| 312 | if (ut->chunkOffset <= 0) { |
| 313 | if (ut->pFuncs->access(ut, ut->chunkNativeStart, FALSE0) == FALSE0) { |
| 314 | return U_SENTINEL(-1); |
| 315 | } |
| 316 | } |
| 317 | ut->chunkOffset--; |
| 318 | c = ut->chunkContents[ut->chunkOffset]; |
| 319 | if (U16_IS_TRAIL(c)(((c)&0xfffffc00)==0xdc00) == FALSE0) { |
| 320 | // Normal case, not supplementary. |
| 321 | // (A lead surrogate seen here is just returned as is, as a surrogate value. |
| 322 | // It cannot be part of a pair.) |
| 323 | return c; |
| 324 | } |
| 325 | |
| 326 | if (ut->chunkOffset <= 0) { |
| 327 | if (ut->pFuncs->access(ut, ut->chunkNativeStart, FALSE0) == FALSE0) { |
| 328 | // c is an unpaired trail surrogate at the start of the text. |
| 329 | // return it as it is. |
| 330 | return c; |
| 331 | } |
| 332 | } |
| 333 | |
| 334 | UChar32 lead = ut->chunkContents[ut->chunkOffset-1]; |
| 335 | if (U16_IS_LEAD(lead)(((lead)&0xfffffc00)==0xd800) == FALSE0) { |
| 336 | // c was an unpaired trail surrogate, not at the end of the text. |
| 337 | // return it as it is (unpaired). Iteration position is at c |
| 338 | return c; |
| 339 | } |
| 340 | |
| 341 | UChar32 supplementary = U16_GET_SUPPLEMENTARY(lead, c)(((UChar32)(lead)<<10UL)+(UChar32)(c)-((0xd800<<10UL )+0xdc00-0x10000)); |
| 342 | ut->chunkOffset--; // move iteration position over the lead surrogate. |
| 343 | return supplementary; |
| 344 | } |
| 345 | |
| 346 | |
| 347 | |
| 348 | U_CAPIextern "C" UChar32 U_EXPORT2 |
| 349 | utext_next32Fromutext_next32From_71(UText *ut, int64_t index) { |
| 350 | UChar32 c = U_SENTINEL(-1); |
| 351 | |
| 352 | if(index<ut->chunkNativeStart || index>=ut->chunkNativeLimit) { |
| 353 | // Desired position is outside of the current chunk. |
| 354 | if(!ut->pFuncs->access(ut, index, TRUE1)) { |
| 355 | // no chunk available here |
| 356 | return U_SENTINEL(-1); |
| 357 | } |
| 358 | } else if (index - ut->chunkNativeStart <= (int64_t)ut->nativeIndexingLimit) { |
| 359 | // Desired position is in chunk, with direct 1:1 native to UTF16 indexing |
| 360 | ut->chunkOffset = (int32_t)(index - ut->chunkNativeStart); |
| 361 | } else { |
| 362 | // Desired position is in chunk, with non-UTF16 indexing. |
| 363 | ut->chunkOffset = ut->pFuncs->mapNativeIndexToUTF16(ut, index); |
| 364 | } |
| 365 | |
| 366 | c = ut->chunkContents[ut->chunkOffset++]; |
| 367 | if (U16_IS_SURROGATE(c)(((c)&0xfffff800)==0xd800)) { |
| 368 | // Surrogates. Many edge cases. Use other functions that already |
| 369 | // deal with the problems. |
| 370 | utext_setNativeIndexutext_setNativeIndex_71(ut, index); |
| 371 | c = utext_next32utext_next32_71(ut); |
| 372 | } |
| 373 | return c; |
| 374 | } |
| 375 | |
| 376 | |
| 377 | U_CAPIextern "C" UChar32 U_EXPORT2 |
| 378 | utext_previous32Fromutext_previous32From_71(UText *ut, int64_t index) { |
| 379 | // |
| 380 | // Return the character preceding the specified index. |
| 381 | // Leave the iteration position at the start of the character that was returned. |
| 382 | // |
| 383 | UChar32 cPrev; // The character preceding cCurr, which is what we will return. |
| 384 | |
| 385 | // Address the chunk containing the position preceding the incoming index |
| 386 | // A tricky edge case: |
| 387 | // We try to test the requested native index against the chunkNativeStart to determine |
| 388 | // whether the character preceding the one at the index is in the current chunk. |
| 389 | // BUT, this test can fail with UTF-8 (or any other multibyte encoding), when the |
| 390 | // requested index is on something other than the first position of the first char. |
| 391 | // |
| 392 | if(index<=ut->chunkNativeStart || index>ut->chunkNativeLimit) { |
| 393 | // Requested native index is outside of the current chunk. |
| 394 | if(!ut->pFuncs->access(ut, index, FALSE0)) { |
| 395 | // no chunk available here |
| 396 | return U_SENTINEL(-1); |
| 397 | } |
| 398 | } else if(index - ut->chunkNativeStart <= (int64_t)ut->nativeIndexingLimit) { |
| 399 | // Direct UTF-16 indexing. |
| 400 | ut->chunkOffset = (int32_t)(index - ut->chunkNativeStart); |
| 401 | } else { |
| 402 | ut->chunkOffset=ut->pFuncs->mapNativeIndexToUTF16(ut, index); |
| 403 | if (ut->chunkOffset==0 && !ut->pFuncs->access(ut, index, FALSE0)) { |
| 404 | // no chunk available here |
| 405 | return U_SENTINEL(-1); |
| 406 | } |
| 407 | } |
| 408 | |
| 409 | // |
| 410 | // Simple case with no surrogates. |
| 411 | // |
| 412 | ut->chunkOffset--; |
| 413 | cPrev = ut->chunkContents[ut->chunkOffset]; |
| 414 | |
| 415 | if (U16_IS_SURROGATE(cPrev)(((cPrev)&0xfffff800)==0xd800)) { |
| 416 | // Possible supplementary. Many edge cases. |
| 417 | // Let other functions do the heavy lifting. |
| 418 | utext_setNativeIndexutext_setNativeIndex_71(ut, index); |
| 419 | cPrev = utext_previous32utext_previous32_71(ut); |
| 420 | } |
| 421 | return cPrev; |
| 422 | } |
| 423 | |
| 424 | |
| 425 | U_CAPIextern "C" int32_t U_EXPORT2 |
| 426 | utext_extractutext_extract_71(UText *ut, |
| 427 | int64_t start, int64_t limit, |
| 428 | UChar *dest, int32_t destCapacity, |
| 429 | UErrorCode *status) { |
| 430 | return ut->pFuncs->extract(ut, start, limit, dest, destCapacity, status); |
| 431 | } |
| 432 | |
| 433 | |
| 434 | |
| 435 | U_CAPIextern "C" UBool U_EXPORT2 |
| 436 | utext_equalsutext_equals_71(const UText *a, const UText *b) { |
| 437 | if (a==NULL__null || b==NULL__null || |
| 438 | a->magic != UTEXT_MAGIC || |
| 439 | b->magic != UTEXT_MAGIC) { |
| 440 | // Null or invalid arguments don't compare equal to anything. |
| 441 | return FALSE0; |
| 442 | } |
| 443 | |
| 444 | if (a->pFuncs != b->pFuncs) { |
| 445 | // Different types of text providers. |
| 446 | return FALSE0; |
| 447 | } |
| 448 | |
| 449 | if (a->context != b->context) { |
| 450 | // Different sources (different strings) |
| 451 | return FALSE0; |
| 452 | } |
| 453 | if (utext_getNativeIndexutext_getNativeIndex_71(a) != utext_getNativeIndexutext_getNativeIndex_71(b)) { |
| 454 | // Different current position in the string. |
| 455 | return FALSE0; |
| 456 | } |
| 457 | |
| 458 | return TRUE1; |
| 459 | } |
| 460 | |
| 461 | U_CAPIextern "C" UBool U_EXPORT2 |
| 462 | utext_isWritableutext_isWritable_71(const UText *ut) |
| 463 | { |
| 464 | UBool b = (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_WRITABLE)((int32_t)1<<(UTEXT_PROVIDER_WRITABLE))) != 0; |
| 465 | return b; |
| 466 | } |
| 467 | |
| 468 | |
| 469 | U_CAPIextern "C" void U_EXPORT2 |
| 470 | utext_freezeutext_freeze_71(UText *ut) { |
| 471 | // Zero out the WRITABLE flag. |
| 472 | ut->providerProperties &= ~(I32_FLAG(UTEXT_PROVIDER_WRITABLE)((int32_t)1<<(UTEXT_PROVIDER_WRITABLE))); |
| 473 | } |
| 474 | |
| 475 | |
| 476 | U_CAPIextern "C" UBool U_EXPORT2 |
| 477 | utext_hasMetaDatautext_hasMetaData_71(const UText *ut) |
| 478 | { |
| 479 | UBool b = (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_HAS_META_DATA)((int32_t)1<<(UTEXT_PROVIDER_HAS_META_DATA))) != 0; |
| 480 | return b; |
| 481 | } |
| 482 | |
| 483 | |
| 484 | |
| 485 | U_CAPIextern "C" int32_t U_EXPORT2 |
| 486 | utext_replaceutext_replace_71(UText *ut, |
| 487 | int64_t nativeStart, int64_t nativeLimit, |
| 488 | const UChar *replacementText, int32_t replacementLength, |
| 489 | UErrorCode *status) |
| 490 | { |
| 491 | if (U_FAILURE(*status)) { |
| 492 | return 0; |
| 493 | } |
| 494 | if ((ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_WRITABLE)((int32_t)1<<(UTEXT_PROVIDER_WRITABLE))) == 0) { |
| 495 | *status = U_NO_WRITE_PERMISSION; |
| 496 | return 0; |
| 497 | } |
| 498 | int32_t i = ut->pFuncs->replace(ut, nativeStart, nativeLimit, replacementText, replacementLength, status); |
| 499 | return i; |
| 500 | } |
| 501 | |
| 502 | U_CAPIextern "C" void U_EXPORT2 |
| 503 | utext_copyutext_copy_71(UText *ut, |
| 504 | int64_t nativeStart, int64_t nativeLimit, |
| 505 | int64_t destIndex, |
| 506 | UBool move, |
| 507 | UErrorCode *status) |
| 508 | { |
| 509 | if (U_FAILURE(*status)) { |
| 510 | return; |
| 511 | } |
| 512 | if ((ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_WRITABLE)((int32_t)1<<(UTEXT_PROVIDER_WRITABLE))) == 0) { |
| 513 | *status = U_NO_WRITE_PERMISSION; |
| 514 | return; |
| 515 | } |
| 516 | ut->pFuncs->copy(ut, nativeStart, nativeLimit, destIndex, move, status); |
| 517 | } |
| 518 | |
| 519 | |
| 520 | |
| 521 | U_CAPIextern "C" UText * U_EXPORT2 |
| 522 | utext_cloneutext_clone_71(UText *dest, const UText *src, UBool deep, UBool readOnly, UErrorCode *status) { |
| 523 | if (U_FAILURE(*status)) { |
| 524 | return dest; |
| 525 | } |
| 526 | UText *result = src->pFuncs->clone(dest, src, deep, status); |
| 527 | if (U_FAILURE(*status)) { |
| 528 | return result; |
| 529 | } |
| 530 | if (result == NULL__null) { |
| 531 | *status = U_MEMORY_ALLOCATION_ERROR; |
| 532 | return result; |
| 533 | } |
| 534 | if (readOnly) { |
| 535 | utext_freezeutext_freeze_71(result); |
| 536 | } |
| 537 | return result; |
| 538 | } |
| 539 | |
| 540 | |
| 541 | |
| 542 | //------------------------------------------------------------------------------ |
| 543 | // |
| 544 | // UText common functions implementation |
| 545 | // |
| 546 | //------------------------------------------------------------------------------ |
| 547 | |
| 548 | // |
| 549 | // UText.flags bit definitions |
| 550 | // |
| 551 | enum { |
| 552 | UTEXT_HEAP_ALLOCATED = 1, // 1 if ICU has allocated this UText struct on the heap. |
| 553 | // 0 if caller provided storage for the UText. |
| 554 | |
| 555 | UTEXT_EXTRA_HEAP_ALLOCATED = 2, // 1 if ICU has allocated extra storage as a separate |
| 556 | // heap block. |
| 557 | // 0 if there is no separate allocation. Either no extra |
| 558 | // storage was requested, or it is appended to the end |
| 559 | // of the main UText storage. |
| 560 | |
| 561 | UTEXT_OPEN = 4 // 1 if this UText is currently open |
| 562 | // 0 if this UText is not open. |
| 563 | }; |
| 564 | |
| 565 | |
| 566 | // |
| 567 | // Extended form of a UText. The purpose is to aid in computing the total size required |
| 568 | // when a provider asks for a UText to be allocated with extra storage. |
| 569 | |
| 570 | struct ExtendedUText { |
| 571 | UText ut; |
| 572 | std::max_align_t extension; |
| 573 | }; |
| 574 | |
| 575 | static const UText emptyText = UTEXT_INITIALIZER{ UTEXT_MAGIC, 0, 0, sizeof(UText), 0, 0, 0, 0, 0, 0, __null, __null, __null, __null, __null, __null, __null, __null, 0, 0 , 0, 0, 0, 0 }; |
| 576 | |
| 577 | U_CAPIextern "C" UText * U_EXPORT2 |
| 578 | utext_setuputext_setup_71(UText *ut, int32_t extraSpace, UErrorCode *status) { |
| 579 | if (U_FAILURE(*status)) { |
| 580 | return ut; |
| 581 | } |
| 582 | |
| 583 | if (ut == NULL__null) { |
| 584 | // We need to heap-allocate storage for the new UText |
| 585 | int32_t spaceRequired = sizeof(UText); |
| 586 | if (extraSpace > 0) { |
| 587 | spaceRequired = sizeof(ExtendedUText) + extraSpace - sizeof(std::max_align_t); |
| 588 | } |
| 589 | ut = (UText *)uprv_mallocuprv_malloc_71(spaceRequired); |
| 590 | if (ut == NULL__null) { |
| 591 | *status = U_MEMORY_ALLOCATION_ERROR; |
| 592 | return NULL__null; |
| 593 | } else { |
| 594 | *ut = emptyText; |
| 595 | ut->flags |= UTEXT_HEAP_ALLOCATED; |
| 596 | if (spaceRequired>0) { |
| 597 | ut->extraSize = extraSpace; |
| 598 | ut->pExtra = &((ExtendedUText *)ut)->extension; |
| 599 | } |
| 600 | } |
| 601 | } else { |
| 602 | // We have been supplied with an already existing UText. |
| 603 | // Verify that it really appears to be a UText. |
| 604 | if (ut->magic != UTEXT_MAGIC) { |
| 605 | *status = U_ILLEGAL_ARGUMENT_ERROR; |
| 606 | return ut; |
| 607 | } |
| 608 | // If the ut is already open and there's a provider supplied close |
| 609 | // function, call it. |
| 610 | if ((ut->flags & UTEXT_OPEN) && ut->pFuncs->close != NULL__null) { |
| 611 | ut->pFuncs->close(ut); |
| 612 | } |
| 613 | ut->flags &= ~UTEXT_OPEN; |
| 614 | |
| 615 | // If extra space was requested by our caller, check whether |
| 616 | // sufficient already exists, and allocate new if needed. |
| 617 | if (extraSpace > ut->extraSize) { |
| 618 | // Need more space. If there is existing separately allocated space, |
| 619 | // delete it first, then allocate new space. |
| 620 | if (ut->flags & UTEXT_EXTRA_HEAP_ALLOCATED) { |
| 621 | uprv_freeuprv_free_71(ut->pExtra); |
| 622 | ut->extraSize = 0; |
| 623 | } |
| 624 | ut->pExtra = uprv_mallocuprv_malloc_71(extraSpace); |
| 625 | if (ut->pExtra == NULL__null) { |
| 626 | *status = U_MEMORY_ALLOCATION_ERROR; |
| 627 | } else { |
| 628 | ut->extraSize = extraSpace; |
| 629 | ut->flags |= UTEXT_EXTRA_HEAP_ALLOCATED; |
| 630 | } |
| 631 | } |
| 632 | } |
| 633 | if (U_SUCCESS(*status)) { |
| 634 | ut->flags |= UTEXT_OPEN; |
| 635 | |
| 636 | // Initialize all remaining fields of the UText. |
| 637 | // |
| 638 | ut->context = NULL__null; |
| 639 | ut->chunkContents = NULL__null; |
| 640 | ut->p = NULL__null; |
| 641 | ut->q = NULL__null; |
| 642 | ut->r = NULL__null; |
| 643 | ut->a = 0; |
| 644 | ut->b = 0; |
| 645 | ut->c = 0; |
| 646 | ut->chunkOffset = 0; |
| 647 | ut->chunkLength = 0; |
| 648 | ut->chunkNativeStart = 0; |
| 649 | ut->chunkNativeLimit = 0; |
| 650 | ut->nativeIndexingLimit = 0; |
| 651 | ut->providerProperties = 0; |
| 652 | ut->privA = 0; |
| 653 | ut->privB = 0; |
| 654 | ut->privC = 0; |
| 655 | ut->privP = NULL__null; |
| 656 | if (ut->pExtra!=NULL__null && ut->extraSize>0) |
| 657 | uprv_memset(ut->pExtra, 0, ut->extraSize):: memset(ut->pExtra, 0, ut->extraSize); |
| 658 | |
| 659 | } |
| 660 | return ut; |
| 661 | } |
| 662 | |
| 663 | |
| 664 | U_CAPIextern "C" UText * U_EXPORT2 |
| 665 | utext_closeutext_close_71(UText *ut) { |
| 666 | if (ut==NULL__null || |
| 667 | ut->magic != UTEXT_MAGIC || |
| 668 | (ut->flags & UTEXT_OPEN) == 0) |
| 669 | { |
| 670 | // The supplied ut is not an open UText. |
| 671 | // Do nothing. |
| 672 | return ut; |
| 673 | } |
| 674 | |
| 675 | // If the provider gave us a close function, call it now. |
| 676 | // This will clean up anything allocated specifically by the provider. |
| 677 | if (ut->pFuncs->close != NULL__null) { |
| 678 | ut->pFuncs->close(ut); |
| 679 | } |
| 680 | ut->flags &= ~UTEXT_OPEN; |
| 681 | |
| 682 | // If we (the framework) allocated the UText or subsidiary storage, |
| 683 | // delete it. |
| 684 | if (ut->flags & UTEXT_EXTRA_HEAP_ALLOCATED) { |
| 685 | uprv_freeuprv_free_71(ut->pExtra); |
| 686 | ut->pExtra = NULL__null; |
| 687 | ut->flags &= ~UTEXT_EXTRA_HEAP_ALLOCATED; |
| 688 | ut->extraSize = 0; |
| 689 | } |
| 690 | |
| 691 | // Zero out function table of the closed UText. This is a defensive move, |
| 692 | // intended to cause applications that inadvertently use a closed |
| 693 | // utext to crash with null pointer errors. |
| 694 | ut->pFuncs = NULL__null; |
| 695 | |
| 696 | if (ut->flags & UTEXT_HEAP_ALLOCATED) { |
| 697 | // This UText was allocated by UText setup. We need to free it. |
| 698 | // Clear magic, so we can detect if the user messes up and immediately |
| 699 | // tries to reopen another UText using the deleted storage. |
| 700 | ut->magic = 0; |
| 701 | uprv_freeuprv_free_71(ut); |
| 702 | ut = NULL__null; |
| 703 | } |
| 704 | return ut; |
| 705 | } |
| 706 | |
| 707 | |
| 708 | |
| 709 | |
| 710 | // |
| 711 | // invalidateChunk Reset a chunk to have no contents, so that the next call |
| 712 | // to access will cause new data to load. |
| 713 | // This is needed when copy/move/replace operate directly on the |
| 714 | // backing text, potentially putting it out of sync with the |
| 715 | // contents in the chunk. |
| 716 | // |
| 717 | static void |
| 718 | invalidateChunk(UText *ut) { |
| 719 | ut->chunkLength = 0; |
| 720 | ut->chunkNativeLimit = 0; |
| 721 | ut->chunkNativeStart = 0; |
| 722 | ut->chunkOffset = 0; |
| 723 | ut->nativeIndexingLimit = 0; |
| 724 | } |
| 725 | |
| 726 | // |
| 727 | // pinIndex Do range pinning on a native index parameter. |
| 728 | // 64 bit pinning is done in place. |
| 729 | // 32 bit truncated result is returned as a convenience for |
| 730 | // use in providers that don't need 64 bits. |
| 731 | static int32_t |
| 732 | pinIndex(int64_t &index, int64_t limit) { |
| 733 | if (index<0) { |
| 734 | index = 0; |
| 735 | } else if (index > limit) { |
| 736 | index = limit; |
| 737 | } |
| 738 | return (int32_t)index; |
| 739 | } |
| 740 | |
| 741 | |
| 742 | U_CDECL_BEGINextern "C" { |
| 743 | |
| 744 | // |
| 745 | // Pointer relocation function, |
| 746 | // a utility used by shallow clone. |
| 747 | // Adjust a pointer that refers to something within one UText (the source) |
| 748 | // to refer to the same relative offset within a another UText (the target) |
| 749 | // |
| 750 | static void adjustPointer(UText *dest, const void **destPtr, const UText *src) { |
| 751 | // convert all pointers to (char *) so that byte address arithmetic will work. |
| 752 | char *dptr = (char *)*destPtr; |
| 753 | char *dUText = (char *)dest; |
| 754 | char *sUText = (char *)src; |
| 755 | |
| 756 | if (dptr >= (char *)src->pExtra && dptr < ((char*)src->pExtra)+src->extraSize) { |
| 757 | // target ptr was to something within the src UText's pExtra storage. |
| 758 | // relocate it into the target UText's pExtra region. |
| 759 | *destPtr = ((char *)dest->pExtra) + (dptr - (char *)src->pExtra); |
| 760 | } else if (dptr>=sUText && dptr < sUText+src->sizeOfStruct) { |
| 761 | // target ptr was pointing to somewhere within the source UText itself. |
| 762 | // Move it to the same offset within the target UText. |
| 763 | *destPtr = dUText + (dptr-sUText); |
| 764 | } |
| 765 | } |
| 766 | |
| 767 | |
| 768 | // |
| 769 | // Clone. This is a generic copy-the-utext-by-value clone function that can be |
| 770 | // used as-is with some utext types, and as a helper by other clones. |
| 771 | // |
| 772 | static UText * U_CALLCONV |
| 773 | shallowTextClone(UText * dest, const UText * src, UErrorCode * status) { |
| 774 | if (U_FAILURE(*status)) { |
| 775 | return NULL__null; |
| 776 | } |
| 777 | int32_t srcExtraSize = src->extraSize; |
| 778 | |
| 779 | // |
| 780 | // Use the generic text_setup to allocate storage if required. |
| 781 | // |
| 782 | dest = utext_setuputext_setup_71(dest, srcExtraSize, status); |
| 783 | if (U_FAILURE(*status)) { |
| 784 | return dest; |
| 785 | } |
| 786 | |
| 787 | // |
| 788 | // flags (how the UText was allocated) and the pointer to the |
| 789 | // extra storage must retain the values in the cloned utext that |
| 790 | // were set up by utext_setup. Save them separately before |
| 791 | // copying the whole struct. |
| 792 | // |
| 793 | void *destExtra = dest->pExtra; |
| 794 | int32_t flags = dest->flags; |
| 795 | |
| 796 | |
| 797 | // |
| 798 | // Copy the whole UText struct by value. |
| 799 | // Any "Extra" storage is copied also. |
| 800 | // |
| 801 | int sizeToCopy = src->sizeOfStruct; |
| 802 | if (sizeToCopy > dest->sizeOfStruct) { |
| 803 | sizeToCopy = dest->sizeOfStruct; |
| 804 | } |
| 805 | uprv_memcpy(dest, src, sizeToCopy)do { clang diagnostic push
clang diagnostic ignored "-Waddress" (void)0; (void)0; clang diagnostic pop :: memcpy(dest, src , sizeToCopy); } while (false); |
| 806 | dest->pExtra = destExtra; |
| 807 | dest->flags = flags; |
| 808 | if (srcExtraSize > 0) { |
| 809 | uprv_memcpy(dest->pExtra, src->pExtra, srcExtraSize)do { clang diagnostic push
clang diagnostic ignored "-Waddress" (void)0; (void)0; clang diagnostic pop :: memcpy(dest-> pExtra, src->pExtra, srcExtraSize); } while (false); |
| 810 | } |
| 811 | |
| 812 | // |
| 813 | // Relocate any pointers in the target that refer to the UText itself |
| 814 | // to point to the cloned copy rather than the original source. |
| 815 | // |
| 816 | adjustPointer(dest, &dest->context, src); |
| 817 | adjustPointer(dest, &dest->p, src); |
| 818 | adjustPointer(dest, &dest->q, src); |
| 819 | adjustPointer(dest, &dest->r, src); |
| 820 | adjustPointer(dest, (const void **)&dest->chunkContents, src); |
| 821 | |
| 822 | // The newly shallow-cloned UText does _not_ own the underlying storage for the text. |
| 823 | // (The source for the clone may or may not have owned the text.) |
| 824 | |
| 825 | dest->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT)); |
| 826 | |
| 827 | return dest; |
| 828 | } |
| 829 | |
| 830 | |
| 831 | U_CDECL_END} |
| 832 | |
| 833 | |
| 834 | |
| 835 | //------------------------------------------------------------------------------ |
| 836 | // |
| 837 | // UText implementation for UTF-8 char * strings (read-only) |
| 838 | // Limitation: string length must be <= 0x7fffffff in length. |
| 839 | // (length must for in an int32_t variable) |
| 840 | // |
| 841 | // Use of UText data members: |
| 842 | // context pointer to UTF-8 string |
| 843 | // utext.b is the input string length (bytes). |
| 844 | // utext.c Length scanned so far in string |
| 845 | // (for optimizing finding length of zero terminated strings.) |
| 846 | // utext.p pointer to the current buffer |
| 847 | // utext.q pointer to the other buffer. |
| 848 | // |
| 849 | //------------------------------------------------------------------------------ |
| 850 | |
| 851 | // Chunk size. |
| 852 | // Must be less than 85 (256/3), because of byte mapping from UChar indexes to native indexes. |
| 853 | // Worst case is three native bytes to one UChar. (Supplemenaries are 4 native bytes |
| 854 | // to two UChars.) |
| 855 | // The longest illegal byte sequence treated as a single error (and converted to U+FFFD) |
| 856 | // is a three-byte sequence (truncated four-byte sequence). |
| 857 | // |
| 858 | enum { UTF8_TEXT_CHUNK_SIZE=32 }; |
| 859 | |
| 860 | // |
| 861 | // UTF8Buf Two of these structs will be set up in the UText's extra allocated space. |
| 862 | // Each contains the UChar chunk buffer, the to and from native maps, and |
| 863 | // header info. |
| 864 | // |
| 865 | // because backwards iteration fills the buffers starting at the end and |
| 866 | // working towards the front, the filled part of the buffers may not begin |
| 867 | // at the start of the available storage for the buffers. |
| 868 | // |
| 869 | // Buffer size is one bigger than the specified UTF8_TEXT_CHUNK_SIZE to allow for |
| 870 | // the last character added being a supplementary, and thus requiring a surrogate |
| 871 | // pair. Doing this is simpler than checking for the edge case. |
| 872 | // |
| 873 | |
| 874 | struct UTF8Buf { |
| 875 | int32_t bufNativeStart; // Native index of first char in UChar buf |
| 876 | int32_t bufNativeLimit; // Native index following last char in buf. |
| 877 | int32_t bufStartIdx; // First filled position in buf. |
| 878 | int32_t bufLimitIdx; // Limit of filled range in buf. |
| 879 | int32_t bufNILimit; // Limit of native indexing part of buf |
| 880 | int32_t toUCharsMapStart; // Native index corresponding to |
| 881 | // mapToUChars[0]. |
| 882 | // Set to bufNativeStart when filling forwards. |
| 883 | // Set to computed value when filling backwards. |
| 884 | |
| 885 | UChar buf[UTF8_TEXT_CHUNK_SIZE+4]; // The UChar buffer. Requires one extra position beyond the |
| 886 | // the chunk size, to allow for surrogate at the end. |
| 887 | // Length must be identical to mapToNative array, below, |
| 888 | // because of the way indexing works when the array is |
| 889 | // filled backwards during a reverse iteration. Thus, |
| 890 | // the additional extra size. |
| 891 | uint8_t mapToNative[UTF8_TEXT_CHUNK_SIZE+4]; // map UChar index in buf to |
| 892 | // native offset from bufNativeStart. |
| 893 | // Requires two extra slots, |
| 894 | // one for a supplementary starting in the last normal position, |
| 895 | // and one for an entry for the buffer limit position. |
| 896 | uint8_t mapToUChars[UTF8_TEXT_CHUNK_SIZE*3+6]; // Map native offset from bufNativeStart to |
| 897 | // corresponding offset in filled part of buf. |
| 898 | int32_t align; |
| 899 | }; |
| 900 | |
| 901 | U_CDECL_BEGINextern "C" { |
| 902 | |
| 903 | // |
| 904 | // utf8TextLength |
| 905 | // |
| 906 | // Get the length of the string. If we don't already know it, |
| 907 | // we'll need to scan for the trailing nul. |
| 908 | // |
| 909 | static int64_t U_CALLCONV |
| 910 | utf8TextLength(UText *ut) { |
| 911 | if (ut->b < 0) { |
| 912 | // Zero terminated string, and we haven't scanned to the end yet. |
| 913 | // Scan it now. |
| 914 | const char *r = (const char *)ut->context + ut->c; |
| 915 | while (*r != 0) { |
| 916 | r++; |
| 917 | } |
| 918 | if ((r - (const char *)ut->context) < 0x7fffffff) { |
| 919 | ut->b = (int32_t)(r - (const char *)ut->context); |
| 920 | } else { |
| 921 | // Actual string was bigger (more than 2 gig) than we |
| 922 | // can handle. Clip it to 2 GB. |
| 923 | ut->b = 0x7fffffff; |
| 924 | } |
| 925 | ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)); |
| 926 | } |
| 927 | return ut->b; |
| 928 | } |
| 929 | |
| 930 | |
| 931 | |
| 932 | |
| 933 | |
| 934 | |
| 935 | static UBool U_CALLCONV |
| 936 | utf8TextAccess(UText *ut, int64_t index, UBool forward) { |
| 937 | // |
| 938 | // Apologies to those who are allergic to goto statements. |
| 939 | // Consider each goto to a labelled block to be the equivalent of |
| 940 | // call the named block as if it were a function(); |
| 941 | // return; |
| 942 | // |
| 943 | const uint8_t *s8=(const uint8_t *)ut->context; |
| 944 | UTF8Buf *u8b = NULL__null; |
| 945 | int32_t length = ut->b; // Length of original utf-8 |
| 946 | int32_t ix= (int32_t)index; // Requested index, trimmed to 32 bits. |
| 947 | int32_t mapIndex = 0; |
| 948 | if (index<0) { |
| 949 | ix=0; |
| 950 | } else if (index > 0x7fffffff) { |
| 951 | // Strings with 64 bit lengths not supported by this UTF-8 provider. |
| 952 | ix = 0x7fffffff; |
| 953 | } |
| 954 | |
| 955 | // Pin requested index to the string length. |
| 956 | if (ix>length) { |
| 957 | if (length>=0) { |
| 958 | ix=length; |
| 959 | } else if (ix>=ut->c) { |
| 960 | // Zero terminated string, and requested index is beyond |
| 961 | // the region that has already been scanned. |
| 962 | // Scan up to either the end of the string or to the |
| 963 | // requested position, whichever comes first. |
| 964 | while (ut->c<ix && s8[ut->c]!=0) { |
| 965 | ut->c++; |
| 966 | } |
| 967 | // TODO: support for null terminated string length > 32 bits. |
| 968 | if (s8[ut->c] == 0) { |
| 969 | // We just found the actual length of the string. |
| 970 | // Trim the requested index back to that. |
| 971 | ix = ut->c; |
| 972 | ut->b = ut->c; |
| 973 | length = ut->c; |
| 974 | ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)); |
| 975 | } |
| 976 | } |
| 977 | } |
| 978 | |
| 979 | // |
| 980 | // Dispatch to the appropriate action for a forward iteration request. |
| 981 | // |
| 982 | if (forward) { |
| 983 | if (ix==ut->chunkNativeLimit) { |
| 984 | // Check for normal sequential iteration cases first. |
| 985 | if (ix==length) { |
| 986 | // Just reached end of string |
| 987 | // Don't swap buffers, but do set the |
| 988 | // current buffer position. |
| 989 | ut->chunkOffset = ut->chunkLength; |
| 990 | return FALSE0; |
| 991 | } else { |
| 992 | // End of current buffer. |
| 993 | // check whether other buffer already has what we need. |
| 994 | UTF8Buf *altB = (UTF8Buf *)ut->q; |
| 995 | if (ix>=altB->bufNativeStart && ix<altB->bufNativeLimit) { |
| 996 | goto swapBuffers; |
| 997 | } |
| 998 | } |
| 999 | } |
| 1000 | |
| 1001 | // A random access. Desired index could be in either or niether buf. |
| 1002 | // For optimizing the order of testing, first check for the index |
| 1003 | // being in the other buffer. This will be the case for uses that |
| 1004 | // move back and forth over a fairly limited range |
| 1005 | { |
| 1006 | u8b = (UTF8Buf *)ut->q; // the alternate buffer |
| 1007 | if (ix>=u8b->bufNativeStart && ix<u8b->bufNativeLimit) { |
| 1008 | // Requested index is in the other buffer. |
| 1009 | goto swapBuffers; |
| 1010 | } |
| 1011 | if (ix == length) { |
| 1012 | // Requested index is end-of-string. |
| 1013 | // (this is the case of randomly seeking to the end. |
| 1014 | // The case of iterating off the end is handled earlier.) |
| 1015 | if (ix == ut->chunkNativeLimit) { |
| 1016 | // Current buffer extends up to the end of the string. |
| 1017 | // Leave it as the current buffer. |
| 1018 | ut->chunkOffset = ut->chunkLength; |
| 1019 | return FALSE0; |
| 1020 | } |
| 1021 | if (ix == u8b->bufNativeLimit) { |
| 1022 | // Alternate buffer extends to the end of string. |
| 1023 | // Swap it in as the current buffer. |
| 1024 | goto swapBuffersAndFail; |
| 1025 | } |
| 1026 | |
| 1027 | // Neither existing buffer extends to the end of the string. |
| 1028 | goto makeStubBuffer; |
| 1029 | } |
| 1030 | |
| 1031 | if (ix<ut->chunkNativeStart || ix>=ut->chunkNativeLimit) { |
| 1032 | // Requested index is in neither buffer. |
| 1033 | goto fillForward; |
| 1034 | } |
| 1035 | |
| 1036 | // Requested index is in this buffer. |
| 1037 | u8b = (UTF8Buf *)ut->p; // the current buffer |
| 1038 | mapIndex = ix - u8b->toUCharsMapStart; |
| 1039 | U_ASSERT(mapIndex < (int32_t)sizeof(UTF8Buf::mapToUChars))(void)0; |
| 1040 | ut->chunkOffset = u8b->mapToUChars[mapIndex] - u8b->bufStartIdx; |
| 1041 | return TRUE1; |
| 1042 | |
| 1043 | } |
| 1044 | } |
| 1045 | |
| 1046 | |
| 1047 | // |
| 1048 | // Dispatch to the appropriate action for a |
| 1049 | // Backwards Direction iteration request. |
| 1050 | // |
| 1051 | if (ix==ut->chunkNativeStart) { |
| 1052 | // Check for normal sequential iteration cases first. |
| 1053 | if (ix==0) { |
| 1054 | // Just reached the start of string |
| 1055 | // Don't swap buffers, but do set the |
| 1056 | // current buffer position. |
| 1057 | ut->chunkOffset = 0; |
| 1058 | return FALSE0; |
| 1059 | } else { |
| 1060 | // Start of current buffer. |
| 1061 | // check whether other buffer already has what we need. |
| 1062 | UTF8Buf *altB = (UTF8Buf *)ut->q; |
| 1063 | if (ix>altB->bufNativeStart && ix<=altB->bufNativeLimit) { |
| 1064 | goto swapBuffers; |
| 1065 | } |
| 1066 | } |
| 1067 | } |
| 1068 | |
| 1069 | // A random access. Desired index could be in either or niether buf. |
| 1070 | // For optimizing the order of testing, |
| 1071 | // Most likely case: in the other buffer. |
| 1072 | // Second most likely: in neither buffer. |
| 1073 | // Unlikely, but must work: in the current buffer. |
| 1074 | u8b = (UTF8Buf *)ut->q; // the alternate buffer |
| 1075 | if (ix>u8b->bufNativeStart && ix<=u8b->bufNativeLimit) { |
| 1076 | // Requested index is in the other buffer. |
| 1077 | goto swapBuffers; |
| 1078 | } |
| 1079 | // Requested index is start-of-string. |
| 1080 | // (this is the case of randomly seeking to the start. |
| 1081 | // The case of iterating off the start is handled earlier.) |
| 1082 | if (ix==0) { |
| 1083 | if (u8b->bufNativeStart==0) { |
| 1084 | // Alternate buffer contains the data for the start string. |
| 1085 | // Make it be the current buffer. |
| 1086 | goto swapBuffersAndFail; |
| 1087 | } else { |
| 1088 | // Request for data before the start of string, |
| 1089 | // neither buffer is usable. |
| 1090 | // set up a zero-length buffer. |
| 1091 | goto makeStubBuffer; |
| 1092 | } |
| 1093 | } |
| 1094 | |
| 1095 | if (ix<=ut->chunkNativeStart || ix>ut->chunkNativeLimit) { |
| 1096 | // Requested index is in neither buffer. |
| 1097 | goto fillReverse; |
| 1098 | } |
| 1099 | |
| 1100 | // Requested index is in this buffer. |
| 1101 | // Set the utf16 buffer index. |
| 1102 | u8b = (UTF8Buf *)ut->p; |
| 1103 | mapIndex = ix - u8b->toUCharsMapStart; |
| 1104 | ut->chunkOffset = u8b->mapToUChars[mapIndex] - u8b->bufStartIdx; |
| 1105 | if (ut->chunkOffset==0) { |
| 1106 | // This occurs when the first character in the text is |
| 1107 | // a multi-byte UTF-8 char, and the requested index is to |
| 1108 | // one of the trailing bytes. Because there is no preceding , |
| 1109 | // character, this access fails. We can't pick up on the |
| 1110 | // situation sooner because the requested index is not zero. |
| 1111 | return FALSE0; |
| 1112 | } else { |
| 1113 | return TRUE1; |
| 1114 | } |
| 1115 | |
| 1116 | |
| 1117 | |
| 1118 | swapBuffers: |
| 1119 | // The alternate buffer (ut->q) has the string data that was requested. |
| 1120 | // Swap the primary and alternate buffers, and set the |
| 1121 | // chunk index into the new primary buffer. |
| 1122 | { |
| 1123 | u8b = (UTF8Buf *)ut->q; |
| 1124 | ut->q = ut->p; |
| 1125 | ut->p = u8b; |
| 1126 | ut->chunkContents = &u8b->buf[u8b->bufStartIdx]; |
| 1127 | ut->chunkLength = u8b->bufLimitIdx - u8b->bufStartIdx; |
| 1128 | ut->chunkNativeStart = u8b->bufNativeStart; |
| 1129 | ut->chunkNativeLimit = u8b->bufNativeLimit; |
| 1130 | ut->nativeIndexingLimit = u8b->bufNILimit; |
| 1131 | |
| 1132 | // Index into the (now current) chunk |
| 1133 | // Use the map to set the chunk index. It's more trouble than it's worth |
| 1134 | // to check whether native indexing can be used. |
| 1135 | U_ASSERT(ix>=u8b->bufNativeStart)(void)0; |
| 1136 | U_ASSERT(ix<=u8b->bufNativeLimit)(void)0; |
| 1137 | mapIndex = ix - u8b->toUCharsMapStart; |
| 1138 | U_ASSERT(mapIndex>=0)(void)0; |
| 1139 | U_ASSERT(mapIndex<(int32_t)sizeof(u8b->mapToUChars))(void)0; |
| 1140 | ut->chunkOffset = u8b->mapToUChars[mapIndex] - u8b->bufStartIdx; |
| 1141 | |
| 1142 | return TRUE1; |
| 1143 | } |
| 1144 | |
| 1145 | |
| 1146 | swapBuffersAndFail: |
| 1147 | // We got a request for either the start or end of the string, |
| 1148 | // with iteration continuing in the out-of-bounds direction. |
| 1149 | // The alternate buffer already contains the data up to the |
| 1150 | // start/end. |
| 1151 | // Swap the buffers, then return failure, indicating that we couldn't |
| 1152 | // make things correct for continuing the iteration in the requested |
| 1153 | // direction. The position & buffer are correct should the |
| 1154 | // user decide to iterate in the opposite direction. |
| 1155 | u8b = (UTF8Buf *)ut->q; |
| 1156 | ut->q = ut->p; |
| 1157 | ut->p = u8b; |
| 1158 | ut->chunkContents = &u8b->buf[u8b->bufStartIdx]; |
| 1159 | ut->chunkLength = u8b->bufLimitIdx - u8b->bufStartIdx; |
| 1160 | ut->chunkNativeStart = u8b->bufNativeStart; |
| 1161 | ut->chunkNativeLimit = u8b->bufNativeLimit; |
| 1162 | ut->nativeIndexingLimit = u8b->bufNILimit; |
| 1163 | |
| 1164 | // Index into the (now current) chunk |
| 1165 | // For this function (swapBuffersAndFail), the requested index |
| 1166 | // will always be at either the start or end of the chunk. |
| 1167 | if (ix==u8b->bufNativeLimit) { |
| 1168 | ut->chunkOffset = ut->chunkLength; |
| 1169 | } else { |
| 1170 | ut->chunkOffset = 0; |
| 1171 | U_ASSERT(ix == u8b->bufNativeStart)(void)0; |
| 1172 | } |
| 1173 | return FALSE0; |
| 1174 | |
| 1175 | makeStubBuffer: |
| 1176 | // The user has done a seek/access past the start or end |
| 1177 | // of the string. Rather than loading data that is likely |
| 1178 | // to never be used, just set up a zero-length buffer at |
| 1179 | // the position. |
| 1180 | u8b = (UTF8Buf *)ut->q; |
| 1181 | u8b->bufNativeStart = ix; |
| 1182 | u8b->bufNativeLimit = ix; |
| 1183 | u8b->bufStartIdx = 0; |
| 1184 | u8b->bufLimitIdx = 0; |
| 1185 | u8b->bufNILimit = 0; |
| 1186 | u8b->toUCharsMapStart = ix; |
| 1187 | u8b->mapToNative[0] = 0; |
| 1188 | u8b->mapToUChars[0] = 0; |
| 1189 | goto swapBuffersAndFail; |
| 1190 | |
| 1191 | |
| 1192 | |
| 1193 | fillForward: |
| 1194 | { |
| 1195 | // Move the incoming index to a code point boundary. |
| 1196 | U8_SET_CP_START(s8, 0, ix)do { if(((int8_t)((s8)[(ix)])<-0x40)) { (ix)=utf8_back1SafeBody_71 (s8, 0, (ix)); } } while (false); |
| 1197 | |
| 1198 | // Swap the UText buffers. |
| 1199 | // We want to fill what was previously the alternate buffer, |
| 1200 | // and make what was the current buffer be the new alternate. |
| 1201 | UTF8Buf *u8b_swap = (UTF8Buf *)ut->q; |
| 1202 | ut->q = ut->p; |
| 1203 | ut->p = u8b_swap; |
| 1204 | |
| 1205 | int32_t strLen = ut->b; |
| 1206 | UBool nulTerminated = FALSE0; |
| 1207 | if (strLen < 0) { |
| 1208 | strLen = 0x7fffffff; |
| 1209 | nulTerminated = TRUE1; |
| 1210 | } |
| 1211 | |
| 1212 | UChar *buf = u8b_swap->buf; |
| 1213 | uint8_t *mapToNative = u8b_swap->mapToNative; |
| 1214 | uint8_t *mapToUChars = u8b_swap->mapToUChars; |
| 1215 | int32_t destIx = 0; |
| 1216 | int32_t srcIx = ix; |
| 1217 | UBool seenNonAscii = FALSE0; |
| 1218 | UChar32 c = 0; |
| 1219 | |
| 1220 | // Fill the chunk buffer and mapping arrays. |
| 1221 | while (destIx<UTF8_TEXT_CHUNK_SIZE) { |
| 1222 | c = s8[srcIx]; |
| 1223 | if (c>0 && c<0x80) { |
| 1224 | // Special case ASCII range for speed. |
| 1225 | // zero is excluded to simplify bounds checking. |
| 1226 | buf[destIx] = (UChar)c; |
| 1227 | mapToNative[destIx] = (uint8_t)(srcIx - ix); |
| 1228 | mapToUChars[srcIx-ix] = (uint8_t)destIx; |
| 1229 | srcIx++; |
| 1230 | destIx++; |
| 1231 | } else { |
| 1232 | // General case, handle everything. |
| 1233 | if (seenNonAscii == FALSE0) { |
| 1234 | seenNonAscii = TRUE1; |
| 1235 | u8b_swap->bufNILimit = destIx; |
| 1236 | } |
| 1237 | |
| 1238 | int32_t cIx = srcIx; |
| 1239 | int32_t dIx = destIx; |
| 1240 | int32_t dIxSaved = destIx; |
| 1241 | U8_NEXT_OR_FFFD(s8, srcIx, strLen, c)do { (c)=(uint8_t)(s8)[(srcIx)++]; if(!(((c)&0x80)==0)) { uint8_t __t = 0; if((srcIx)!=(strLen) && ((c)>=0xe0 ? ((c)<0xf0 ? "\x20\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x30\x10\x30\x30" [(c)&=0xf]&(1<<((__t=(s8)[srcIx])>>5)) && (__t&=0x3f, 1) : ((c)-=0xf0)<=4 && "\x00\x00\x00\x00\x00\x00\x00\x00\x1E\x0F\x0F\x0F\x00\x00\x00\x00" [(__t=(s8)[srcIx])>>4]&(1<<(c)) && (( c)=((c)<<6)|(__t&0x3f), ++(srcIx)!=(strLen)) && (__t=(s8)[srcIx]-0x80)<=0x3f) && ((c)=((c)<< 6)|__t, ++(srcIx)!=(strLen)) : (c)>=0xc2 && ((c)&= 0x1f, 1)) && (__t=(s8)[srcIx]-0x80)<=0x3f && ((c)=((c)<<6)|__t, ++(srcIx), 1)) { } else { (c)=(0xfffd ); } } } while (false); |
| 1242 | if (c==0 && nulTerminated) { |
| 1243 | srcIx--; |
| 1244 | break; |
| 1245 | } |
| 1246 | |
| 1247 | U16_APPEND_UNSAFE(buf, destIx, c)do { if((uint32_t)(c)<=0xffff) { (buf)[(destIx)++]=(uint16_t )(c); } else { (buf)[(destIx)++]=(uint16_t)(((c)>>10)+0xd7c0 ); (buf)[(destIx)++]=(uint16_t)(((c)&0x3ff)|0xdc00); } } while (false); |
| 1248 | do { |
| 1249 | mapToNative[dIx++] = (uint8_t)(cIx - ix); |
| 1250 | } while (dIx < destIx); |
| 1251 | |
| 1252 | do { |
| 1253 | mapToUChars[cIx++ - ix] = (uint8_t)dIxSaved; |
| 1254 | } while (cIx < srcIx); |
| 1255 | } |
| 1256 | if (srcIx>=strLen) { |
| 1257 | break; |
| 1258 | } |
| 1259 | |
| 1260 | } |
| 1261 | |
| 1262 | // store Native <--> Chunk Map entries for the end of the buffer. |
| 1263 | // There is no actual character here, but the index position is valid. |
| 1264 | mapToNative[destIx] = (uint8_t)(srcIx - ix); |
| 1265 | mapToUChars[srcIx - ix] = (uint8_t)destIx; |
| 1266 | |
| 1267 | // fill in Buffer descriptor |
| 1268 | u8b_swap->bufNativeStart = ix; |
| 1269 | u8b_swap->bufNativeLimit = srcIx; |
| 1270 | u8b_swap->bufStartIdx = 0; |
| 1271 | u8b_swap->bufLimitIdx = destIx; |
| 1272 | if (seenNonAscii == FALSE0) { |
| 1273 | u8b_swap->bufNILimit = destIx; |
| 1274 | } |
| 1275 | u8b_swap->toUCharsMapStart = u8b_swap->bufNativeStart; |
| 1276 | |
| 1277 | // Set UText chunk to refer to this buffer. |
| 1278 | ut->chunkContents = buf; |
| 1279 | ut->chunkOffset = 0; |
| 1280 | ut->chunkLength = u8b_swap->bufLimitIdx; |
| 1281 | ut->chunkNativeStart = u8b_swap->bufNativeStart; |
| 1282 | ut->chunkNativeLimit = u8b_swap->bufNativeLimit; |
| 1283 | ut->nativeIndexingLimit = u8b_swap->bufNILimit; |
| 1284 | |
| 1285 | // For zero terminated strings, keep track of the maximum point |
| 1286 | // scanned so far. |
| 1287 | if (nulTerminated && srcIx>ut->c) { |
| 1288 | ut->c = srcIx; |
| 1289 | if (c==0) { |
| 1290 | // We scanned to the end. |
| 1291 | // Remember the actual length. |
| 1292 | ut->b = srcIx; |
| 1293 | ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)); |
| 1294 | } |
| 1295 | } |
| 1296 | return TRUE1; |
| 1297 | } |
| 1298 | |
| 1299 | |
| 1300 | fillReverse: |
| 1301 | { |
| 1302 | // Move the incoming index to a code point boundary. |
| 1303 | // Can only do this if the incoming index is somewhere in the interior of the string. |
| 1304 | // If index is at the end, there is no character there to look at. |
| 1305 | if (ix != ut->b) { |
| 1306 | // Note: this function will only move the index back if it is on a trail byte |
| 1307 | // and there is a preceding lead byte and the sequence from the lead |
| 1308 | // through this trail could be part of a valid UTF-8 sequence |
| 1309 | // Otherwise the index remains unchanged. |
| 1310 | U8_SET_CP_START(s8, 0, ix)do { if(((int8_t)((s8)[(ix)])<-0x40)) { (ix)=utf8_back1SafeBody_71 (s8, 0, (ix)); } } while (false); |
| 1311 | } |
| 1312 | |
| 1313 | // Swap the UText buffers. |
| 1314 | // We want to fill what was previously the alternate buffer, |
| 1315 | // and make what was the current buffer be the new alternate. |
| 1316 | UTF8Buf *u8b_swap = (UTF8Buf *)ut->q; |
| 1317 | ut->q = ut->p; |
| 1318 | ut->p = u8b_swap; |
| 1319 | |
| 1320 | UChar *buf = u8b_swap->buf; |
| 1321 | uint8_t *mapToNative = u8b_swap->mapToNative; |
| 1322 | uint8_t *mapToUChars = u8b_swap->mapToUChars; |
| 1323 | int32_t toUCharsMapStart = ix - sizeof(UTF8Buf::mapToUChars) + 1; |
| 1324 | // Note that toUCharsMapStart can be negative. Happens when the remaining |
| 1325 | // text from current position to the beginning is less than the buffer size. |
| 1326 | // + 1 because mapToUChars must have a slot at the end for the bufNativeLimit entry. |
| 1327 | int32_t destIx = UTF8_TEXT_CHUNK_SIZE+2; // Start in the overflow region |
| 1328 | // at end of buffer to leave room |
| 1329 | // for a surrogate pair at the |
| 1330 | // buffer start. |
| 1331 | int32_t srcIx = ix; |
| 1332 | int32_t bufNILimit = destIx; |
| 1333 | UChar32 c; |
| 1334 | |
| 1335 | // Map to/from Native Indexes, fill in for the position at the end of |
| 1336 | // the buffer. |
| 1337 | // |
| 1338 | mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart); |
| 1339 | mapToUChars[srcIx - toUCharsMapStart] = (uint8_t)destIx; |
| 1340 | |
| 1341 | // Fill the chunk buffer |
| 1342 | // Work backwards, filling from the end of the buffer towards the front. |
| 1343 | // |
| 1344 | while (destIx>2 && (srcIx - toUCharsMapStart > 5) && (srcIx > 0)) { |
| 1345 | srcIx--; |
| 1346 | destIx--; |
| 1347 | |
| 1348 | // Get last byte of the UTF-8 character |
| 1349 | c = s8[srcIx]; |
| 1350 | if (c<0x80) { |
| 1351 | // Special case ASCII range for speed. |
| 1352 | buf[destIx] = (UChar)c; |
| 1353 | U_ASSERT(toUCharsMapStart <= srcIx)(void)0; |
| 1354 | mapToUChars[srcIx - toUCharsMapStart] = (uint8_t)destIx; |
| 1355 | mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart); |
| 1356 | } else { |
| 1357 | // General case, handle everything non-ASCII. |
| 1358 | |
| 1359 | int32_t sIx = srcIx; // ix of last byte of multi-byte u8 char |
| 1360 | |
| 1361 | // Get the full character from the UTF8 string. |
| 1362 | // use code derived from the macros in utf8.h |
| 1363 | // Leaves srcIx pointing at the first byte of the UTF-8 char. |
| 1364 | // |
| 1365 | c=utf8_prevCharSafeBodyutf8_prevCharSafeBody_71(s8, 0, &srcIx, c, -3); |
| 1366 | // leaves srcIx at first byte of the multi-byte char. |
| 1367 | |
| 1368 | // Store the character in UTF-16 buffer. |
| 1369 | if (c<0x10000) { |
| 1370 | buf[destIx] = (UChar)c; |
| 1371 | mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart); |
| 1372 | } else { |
| 1373 | buf[destIx] = U16_TRAIL(c)(UChar)(((c)&0x3ff)|0xdc00); |
| 1374 | mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart); |
| 1375 | buf[--destIx] = U16_LEAD(c)(UChar)(((c)>>10)+0xd7c0); |
| 1376 | mapToNative[destIx] = (uint8_t)(srcIx - toUCharsMapStart); |
| 1377 | } |
| 1378 | |
| 1379 | // Fill in the map from native indexes to UChars buf index. |
| 1380 | do { |
| 1381 | mapToUChars[sIx-- - toUCharsMapStart] = (uint8_t)destIx; |
| 1382 | } while (sIx >= srcIx); |
| 1383 | U_ASSERT(toUCharsMapStart <= (srcIx+1))(void)0; |
| 1384 | |
| 1385 | // Set native indexing limit to be the current position. |
| 1386 | // We are processing a non-ascii, non-native-indexing char now; |
| 1387 | // the limit will be here if the rest of the chars to be |
| 1388 | // added to this buffer are ascii. |
| 1389 | bufNILimit = destIx; |
| 1390 | } |
| 1391 | } |
| 1392 | u8b_swap->bufNativeStart = srcIx; |
| 1393 | u8b_swap->bufNativeLimit = ix; |
| 1394 | u8b_swap->bufStartIdx = destIx; |
| 1395 | u8b_swap->bufLimitIdx = UTF8_TEXT_CHUNK_SIZE+2; |
| 1396 | u8b_swap->bufNILimit = bufNILimit - u8b_swap->bufStartIdx; |
| 1397 | u8b_swap->toUCharsMapStart = toUCharsMapStart; |
| 1398 | |
| 1399 | ut->chunkContents = &buf[u8b_swap->bufStartIdx]; |
| 1400 | ut->chunkLength = u8b_swap->bufLimitIdx - u8b_swap->bufStartIdx; |
| 1401 | ut->chunkOffset = ut->chunkLength; |
| 1402 | ut->chunkNativeStart = u8b_swap->bufNativeStart; |
| 1403 | ut->chunkNativeLimit = u8b_swap->bufNativeLimit; |
| 1404 | ut->nativeIndexingLimit = u8b_swap->bufNILimit; |
| 1405 | return TRUE1; |
| 1406 | } |
| 1407 | |
| 1408 | } |
| 1409 | |
| 1410 | |
| 1411 | |
| 1412 | // |
| 1413 | // This is a slightly modified copy of u_strFromUTF8, |
| 1414 | // Inserts a Replacement Char rather than failing on invalid UTF-8 |
| 1415 | // Removes unnecessary features. |
| 1416 | // |
| 1417 | static UChar* |
| 1418 | utext_strFromUTF8(UChar *dest, |
| 1419 | int32_t destCapacity, |
| 1420 | int32_t *pDestLength, |
| 1421 | const char* src, |
| 1422 | int32_t srcLength, // required. NUL terminated not supported. |
| 1423 | UErrorCode *pErrorCode |
| 1424 | ) |
| 1425 | { |
| 1426 | |
| 1427 | UChar *pDest = dest; |
| 1428 | UChar *pDestLimit = (dest!=NULL__null)?(dest+destCapacity):NULL__null; |
| 1429 | UChar32 ch=0; |
| 1430 | int32_t index = 0; |
| 1431 | int32_t reqLength = 0; |
| 1432 | uint8_t* pSrc = (uint8_t*) src; |
| 1433 | |
| 1434 | |
| 1435 | while((index < srcLength)&&(pDest<pDestLimit)){ |
| 1436 | ch = pSrc[index++]; |
| 1437 | if(ch <=0x7f){ |
| 1438 | *pDest++=(UChar)ch; |
| 1439 | }else{ |
| 1440 | ch=utf8_nextCharSafeBodyutf8_nextCharSafeBody_71(pSrc, &index, srcLength, ch, -3); |
| 1441 | if(U_IS_BMP(ch)((uint32_t)(ch)<=0xffff)){ |
| 1442 | *(pDest++)=(UChar)ch; |
| 1443 | }else{ |
| 1444 | *(pDest++)=U16_LEAD(ch)(UChar)(((ch)>>10)+0xd7c0); |
| 1445 | if(pDest<pDestLimit){ |
| 1446 | *(pDest++)=U16_TRAIL(ch)(UChar)(((ch)&0x3ff)|0xdc00); |
| 1447 | }else{ |
| 1448 | reqLength++; |
| 1449 | break; |
| 1450 | } |
| 1451 | } |
| 1452 | } |
| 1453 | } |
| 1454 | /* donot fill the dest buffer just count the UChars needed */ |
| 1455 | while(index < srcLength){ |
| 1456 | ch = pSrc[index++]; |
| 1457 | if(ch <= 0x7f){ |
| 1458 | reqLength++; |
| 1459 | }else{ |
| 1460 | ch=utf8_nextCharSafeBodyutf8_nextCharSafeBody_71(pSrc, &index, srcLength, ch, -3); |
| 1461 | reqLength+=U16_LENGTH(ch)((uint32_t)(ch)<=0xffff ? 1 : 2); |
| 1462 | } |
| 1463 | } |
| 1464 | |
| 1465 | reqLength+=(int32_t)(pDest - dest); |
| 1466 | |
| 1467 | if(pDestLength){ |
| 1468 | *pDestLength = reqLength; |
| 1469 | } |
| 1470 | |
| 1471 | /* Terminate the buffer */ |
| 1472 | u_terminateUCharsu_terminateUChars_71(dest,destCapacity,reqLength,pErrorCode); |
| 1473 | |
| 1474 | return dest; |
| 1475 | } |
| 1476 | |
| 1477 | |
| 1478 | |
| 1479 | static int32_t U_CALLCONV |
| 1480 | utf8TextExtract(UText *ut, |
| 1481 | int64_t start, int64_t limit, |
| 1482 | UChar *dest, int32_t destCapacity, |
| 1483 | UErrorCode *pErrorCode) { |
| 1484 | if(U_FAILURE(*pErrorCode)) { |
| 1485 | return 0; |
| 1486 | } |
| 1487 | if(destCapacity<0 || (dest==NULL__null && destCapacity>0)) { |
| 1488 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 1489 | return 0; |
| 1490 | } |
| 1491 | int32_t length = ut->b; |
| 1492 | int32_t start32 = pinIndex(start, length); |
| 1493 | int32_t limit32 = pinIndex(limit, length); |
| 1494 | |
| 1495 | if(start32>limit32) { |
| 1496 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
| 1497 | return 0; |
| 1498 | } |
| 1499 | |
| 1500 | |
| 1501 | // adjust the incoming indexes to land on code point boundaries if needed. |
| 1502 | // adjust by no more than three, because that is the largest number of trail bytes |
| 1503 | // in a well formed UTF8 character. |
| 1504 | const uint8_t *buf = (const uint8_t *)ut->context; |
| 1505 | int i; |
| 1506 | if (start32 < ut->chunkNativeLimit) { |
| 1507 | for (i=0; i<3; i++) { |
| 1508 | if (U8_IS_SINGLE(buf[start32])(((buf[start32])&0x80)==0) || U8_IS_LEAD(buf[start32])((uint8_t)((buf[start32])-0xc2)<=0x32) || start32==0) { |
| 1509 | break; |
| 1510 | } |
| 1511 | start32--; |
| 1512 | } |
| 1513 | } |
| 1514 | |
| 1515 | if (limit32 < ut->chunkNativeLimit) { |
| 1516 | for (i=0; i<3; i++) { |
| 1517 | if (U8_IS_SINGLE(buf[limit32])(((buf[limit32])&0x80)==0) || U8_IS_LEAD(buf[limit32])((uint8_t)((buf[limit32])-0xc2)<=0x32) || limit32==0) { |
| 1518 | break; |
| 1519 | } |
| 1520 | limit32--; |
| 1521 | } |
| 1522 | } |
| 1523 | |
| 1524 | // Do the actual extract. |
| 1525 | int32_t destLength=0; |
| 1526 | utext_strFromUTF8(dest, destCapacity, &destLength, |
| 1527 | (const char *)ut->context+start32, limit32-start32, |
| 1528 | pErrorCode); |
| 1529 | utf8TextAccess(ut, limit32, TRUE1); |
| 1530 | return destLength; |
| 1531 | } |
| 1532 | |
| 1533 | // |
| 1534 | // utf8TextMapOffsetToNative |
| 1535 | // |
| 1536 | // Map a chunk (UTF-16) offset to a native index. |
| 1537 | static int64_t U_CALLCONV |
| 1538 | utf8TextMapOffsetToNative(const UText *ut) { |
| 1539 | // |
| 1540 | UTF8Buf *u8b = (UTF8Buf *)ut->p; |
| 1541 | U_ASSERT(ut->chunkOffset>ut->nativeIndexingLimit && ut->chunkOffset<=ut->chunkLength)(void)0; |
| 1542 | int32_t nativeOffset = u8b->mapToNative[ut->chunkOffset + u8b->bufStartIdx] + u8b->toUCharsMapStart; |
| 1543 | U_ASSERT(nativeOffset >= ut->chunkNativeStart && nativeOffset <= ut->chunkNativeLimit)(void)0; |
| 1544 | return nativeOffset; |
| 1545 | } |
| 1546 | |
| 1547 | // |
| 1548 | // Map a native index to the corresponding chunk offset |
| 1549 | // |
| 1550 | static int32_t U_CALLCONV |
| 1551 | utf8TextMapIndexToUTF16(const UText *ut, int64_t index64) { |
| 1552 | U_ASSERT(index64 <= 0x7fffffff)(void)0; |
| 1553 | int32_t index = (int32_t)index64; |
| 1554 | UTF8Buf *u8b = (UTF8Buf *)ut->p; |
| 1555 | U_ASSERT(index>=ut->chunkNativeStart+ut->nativeIndexingLimit)(void)0; |
| 1556 | U_ASSERT(index<=ut->chunkNativeLimit)(void)0; |
| 1557 | int32_t mapIndex = index - u8b->toUCharsMapStart; |
| 1558 | U_ASSERT(mapIndex < (int32_t)sizeof(UTF8Buf::mapToUChars))(void)0; |
| 1559 | int32_t offset = u8b->mapToUChars[mapIndex] - u8b->bufStartIdx; |
| 1560 | U_ASSERT(offset>=0 && offset<=ut->chunkLength)(void)0; |
| 1561 | return offset; |
| 1562 | } |
| 1563 | |
| 1564 | static UText * U_CALLCONV |
| 1565 | utf8TextClone(UText *dest, const UText *src, UBool deep, UErrorCode *status) |
| 1566 | { |
| 1567 | // First do a generic shallow clone. Does everything needed for the UText struct itself. |
| 1568 | dest = shallowTextClone(dest, src, status); |
| 1569 | |
| 1570 | // For deep clones, make a copy of the string. |
| 1571 | // The copied storage is owned by the newly created clone. |
| 1572 | // |
| 1573 | // TODO: There is an issue with using utext_nativeLength(). |
| 1574 | // That function is non-const in cases where the input was NUL terminated |
| 1575 | // and the length has not yet been determined. |
| 1576 | // This function (clone()) is const. |
| 1577 | // There potentially a thread safety issue lurking here. |
| 1578 | // |
| 1579 | if (deep && U_SUCCESS(*status)) { |
| 1580 | int32_t len = (int32_t)utext_nativeLengthutext_nativeLength_71((UText *)src); |
| 1581 | char *copyStr = (char *)uprv_mallocuprv_malloc_71(len+1); |
| 1582 | if (copyStr == NULL__null) { |
| 1583 | *status = U_MEMORY_ALLOCATION_ERROR; |
| 1584 | } else { |
| 1585 | uprv_memcpy(copyStr, src->context, len+1)do { clang diagnostic push
clang diagnostic ignored "-Waddress" (void)0; (void)0; clang diagnostic pop :: memcpy(copyStr, src ->context, len+1); } while (false); |
| 1586 | dest->context = copyStr; |
| 1587 | dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT)); |
| 1588 | } |
| 1589 | } |
| 1590 | return dest; |
| 1591 | } |
| 1592 | |
| 1593 | |
| 1594 | static void U_CALLCONV |
| 1595 | utf8TextClose(UText *ut) { |
| 1596 | // Most of the work of close is done by the generic UText framework close. |
| 1597 | // All that needs to be done here is to delete the UTF8 string if the UText |
| 1598 | // owns it. This occurs if the UText was created by cloning. |
| 1599 | if (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT))) { |
| 1600 | char *s = (char *)ut->context; |
| 1601 | uprv_freeuprv_free_71(s); |
| 1602 | ut->context = NULL__null; |
| 1603 | } |
| 1604 | } |
| 1605 | |
| 1606 | U_CDECL_END} |
| 1607 | |
| 1608 | |
| 1609 | static const struct UTextFuncs utf8Funcs = |
| 1610 | { |
| 1611 | sizeof(UTextFuncs), |
| 1612 | 0, 0, 0, // Reserved alignment padding |
| 1613 | utf8TextClone, |
| 1614 | utf8TextLength, |
| 1615 | utf8TextAccess, |
| 1616 | utf8TextExtract, |
| 1617 | NULL__null, /* replace*/ |
| 1618 | NULL__null, /* copy */ |
| 1619 | utf8TextMapOffsetToNative, |
| 1620 | utf8TextMapIndexToUTF16, |
| 1621 | utf8TextClose, |
| 1622 | NULL__null, // spare 1 |
| 1623 | NULL__null, // spare 2 |
| 1624 | NULL__null // spare 3 |
| 1625 | }; |
| 1626 | |
| 1627 | |
| 1628 | static const char gEmptyString[] = {0}; |
| 1629 | |
| 1630 | U_CAPIextern "C" UText * U_EXPORT2 |
| 1631 | utext_openUTF8utext_openUTF8_71(UText *ut, const char *s, int64_t length, UErrorCode *status) { |
| 1632 | if(U_FAILURE(*status)) { |
| 1633 | return NULL__null; |
| 1634 | } |
| 1635 | if(s==NULL__null && length==0) { |
| 1636 | s = gEmptyString; |
| 1637 | } |
| 1638 | |
| 1639 | if(s==NULL__null || length<-1 || length>INT32_MAX(2147483647)) { |
| 1640 | *status=U_ILLEGAL_ARGUMENT_ERROR; |
| 1641 | return NULL__null; |
| 1642 | } |
| 1643 | |
| 1644 | ut = utext_setuputext_setup_71(ut, sizeof(UTF8Buf) * 2, status); |
| 1645 | if (U_FAILURE(*status)) { |
| 1646 | return ut; |
| 1647 | } |
| 1648 | |
| 1649 | ut->pFuncs = &utf8Funcs; |
| 1650 | ut->context = s; |
| 1651 | ut->b = (int32_t)length; |
| 1652 | ut->c = (int32_t)length; |
| 1653 | if (ut->c < 0) { |
| 1654 | ut->c = 0; |
| 1655 | ut->providerProperties |= I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)); |
| 1656 | } |
| 1657 | ut->p = ut->pExtra; |
| 1658 | ut->q = (char *)ut->pExtra + sizeof(UTF8Buf); |
| 1659 | return ut; |
| 1660 | |
| 1661 | } |
| 1662 | |
| 1663 | |
| 1664 | |
| 1665 | |
| 1666 | |
| 1667 | |
| 1668 | |
| 1669 | |
| 1670 | //------------------------------------------------------------------------------ |
| 1671 | // |
| 1672 | // UText implementation wrapper for Replaceable (read/write) |
| 1673 | // |
| 1674 | // Use of UText data members: |
| 1675 | // context pointer to Replaceable. |
| 1676 | // p pointer to Replaceable if it is owned by the UText. |
| 1677 | // |
| 1678 | //------------------------------------------------------------------------------ |
| 1679 | |
| 1680 | |
| 1681 | |
| 1682 | // minimum chunk size for this implementation: 3 |
| 1683 | // to allow for possible trimming for code point boundaries |
| 1684 | enum { REP_TEXT_CHUNK_SIZE=10 }; |
| 1685 | |
| 1686 | struct ReplExtra { |
| 1687 | /* |
| 1688 | * Chunk UChars. |
| 1689 | * +1 to simplify filling with surrogate pair at the end. |
| 1690 | */ |
| 1691 | UChar s[REP_TEXT_CHUNK_SIZE+1]; |
| 1692 | }; |
| 1693 | |
| 1694 | |
| 1695 | U_CDECL_BEGINextern "C" { |
| 1696 | |
| 1697 | static UText * U_CALLCONV |
| 1698 | repTextClone(UText *dest, const UText *src, UBool deep, UErrorCode *status) { |
| 1699 | // First do a generic shallow clone. Does everything needed for the UText struct itself. |
| 1700 | dest = shallowTextClone(dest, src, status); |
| 1701 | |
| 1702 | // For deep clones, make a copy of the Replaceable. |
| 1703 | // The copied Replaceable storage is owned by the newly created UText clone. |
| 1704 | // A non-NULL pointer in UText.p is the signal to the close() function to delete |
| 1705 | // it. |
| 1706 | // |
| 1707 | if (deep && U_SUCCESS(*status)) { |
| 1708 | const Replaceable *replSrc = (const Replaceable *)src->context; |
| 1709 | dest->context = replSrc->clone(); |
| 1710 | dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT)); |
| 1711 | |
| 1712 | // with deep clone, the copy is writable, even when the source is not. |
| 1713 | dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_WRITABLE)((int32_t)1<<(UTEXT_PROVIDER_WRITABLE)); |
| 1714 | } |
| 1715 | return dest; |
| 1716 | } |
| 1717 | |
| 1718 | |
| 1719 | static void U_CALLCONV |
| 1720 | repTextClose(UText *ut) { |
| 1721 | // Most of the work of close is done by the generic UText framework close. |
| 1722 | // All that needs to be done here is delete the Replaceable if the UText |
| 1723 | // owns it. This occurs if the UText was created by cloning. |
| 1724 | if (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT))) { |
| 1725 | Replaceable *rep = (Replaceable *)ut->context; |
| 1726 | delete rep; |
| 1727 | ut->context = NULL__null; |
| 1728 | } |
| 1729 | } |
| 1730 | |
| 1731 | |
| 1732 | static int64_t U_CALLCONV |
| 1733 | repTextLength(UText *ut) { |
| 1734 | const Replaceable *replSrc = (const Replaceable *)ut->context; |
| 1735 | int32_t len = replSrc->length(); |
| 1736 | return len; |
| 1737 | } |
| 1738 | |
| 1739 | |
| 1740 | static UBool U_CALLCONV |
| 1741 | repTextAccess(UText *ut, int64_t index, UBool forward) { |
| 1742 | const Replaceable *rep=(const Replaceable *)ut->context; |
| 1743 | int32_t length=rep->length(); // Full length of the input text (bigger than a chunk) |
| 1744 | |
| 1745 | // clip the requested index to the limits of the text. |
| 1746 | int32_t index32 = pinIndex(index, length); |
| 1747 | U_ASSERT(index<=INT32_MAX)(void)0; |
| 1748 | |
| 1749 | |
| 1750 | /* |
| 1751 | * Compute start/limit boundaries around index, for a segment of text |
| 1752 | * to be extracted. |
| 1753 | * To allow for the possibility that our user gave an index to the trailing |
| 1754 | * half of a surrogate pair, we must request one extra preceding UChar when |
| 1755 | * going in the forward direction. This will ensure that the buffer has the |
| 1756 | * entire code point at the specified index. |
| 1757 | */ |
| 1758 | if(forward) { |
| 1759 | |
| 1760 | if (index32>=ut->chunkNativeStart && index32<ut->chunkNativeLimit) { |
| 1761 | // Buffer already contains the requested position. |
| 1762 | ut->chunkOffset = (int32_t)(index - ut->chunkNativeStart); |
| 1763 | return TRUE1; |
| 1764 | } |
| 1765 | if (index32>=length && ut->chunkNativeLimit==length) { |
| 1766 | // Request for end of string, and buffer already extends up to it. |
| 1767 | // Can't get the data, but don't change the buffer. |
| 1768 | ut->chunkOffset = length - (int32_t)ut->chunkNativeStart; |
| 1769 | return FALSE0; |
| 1770 | } |
| 1771 | |
| 1772 | ut->chunkNativeLimit = index + REP_TEXT_CHUNK_SIZE - 1; |
| 1773 | // Going forward, so we want to have the buffer with stuff at and beyond |
| 1774 | // the requested index. The -1 gets us one code point before the |
| 1775 | // requested index also, to handle the case of the index being on |
| 1776 | // a trail surrogate of a surrogate pair. |
| 1777 | if(ut->chunkNativeLimit > length) { |
| 1778 | ut->chunkNativeLimit = length; |
| 1779 | } |
| 1780 | // unless buffer ran off end, start is index-1. |
| 1781 | ut->chunkNativeStart = ut->chunkNativeLimit - REP_TEXT_CHUNK_SIZE; |
| 1782 | if(ut->chunkNativeStart < 0) { |
| 1783 | ut->chunkNativeStart = 0; |
| 1784 | } |
| 1785 | } else { |
| 1786 | // Reverse iteration. Fill buffer with data preceding the requested index. |
| 1787 | if (index32>ut->chunkNativeStart && index32<=ut->chunkNativeLimit) { |
| 1788 | // Requested position already in buffer. |
| 1789 | ut->chunkOffset = index32 - (int32_t)ut->chunkNativeStart; |
| 1790 | return TRUE1; |
| 1791 | } |
| 1792 | if (index32==0 && ut->chunkNativeStart==0) { |
| 1793 | // Request for start, buffer already begins at start. |
| 1794 | // No data, but keep the buffer as is. |
| 1795 | ut->chunkOffset = 0; |
| 1796 | return FALSE0; |
| 1797 | } |
| 1798 | |
| 1799 | // Figure out the bounds of the chunk to extract for reverse iteration. |
| 1800 | // Need to worry about chunk not splitting surrogate pairs, and while still |
| 1801 | // containing the data we need. |
| 1802 | // Fix by requesting a chunk that includes an extra UChar at the end. |
| 1803 | // If this turns out to be a lead surrogate, we can lop it off and still have |
| 1804 | // the data we wanted. |
| 1805 | ut->chunkNativeStart = index32 + 1 - REP_TEXT_CHUNK_SIZE; |
| 1806 | if (ut->chunkNativeStart < 0) { |
| 1807 | ut->chunkNativeStart = 0; |
| 1808 | } |
| 1809 | |
| 1810 | ut->chunkNativeLimit = index32 + 1; |
| 1811 | if (ut->chunkNativeLimit > length) { |
| 1812 | ut->chunkNativeLimit = length; |
| 1813 | } |
| 1814 | } |
| 1815 | |
| 1816 | // Extract the new chunk of text from the Replaceable source. |
| 1817 | ReplExtra *ex = (ReplExtra *)ut->pExtra; |
| 1818 | // UnicodeString with its buffer a writable alias to the chunk buffer |
| 1819 | UnicodeString buffer(ex->s, 0 /*buffer length*/, REP_TEXT_CHUNK_SIZE /*buffer capacity*/); |
| 1820 | rep->extractBetween((int32_t)ut->chunkNativeStart, (int32_t)ut->chunkNativeLimit, buffer); |
| 1821 | |
| 1822 | ut->chunkContents = ex->s; |
| 1823 | ut->chunkLength = (int32_t)(ut->chunkNativeLimit - ut->chunkNativeStart); |
| 1824 | ut->chunkOffset = (int32_t)(index32 - ut->chunkNativeStart); |
| 1825 | |
| 1826 | // Surrogate pairs from the input text must not span chunk boundaries. |
| 1827 | // If end of chunk could be the start of a surrogate, trim it off. |
| 1828 | if (ut->chunkNativeLimit < length && |
| 1829 | U16_IS_LEAD(ex->s[ut->chunkLength-1])(((ex->s[ut->chunkLength-1])&0xfffffc00)==0xd800)) { |
| 1830 | ut->chunkLength--; |
| 1831 | ut->chunkNativeLimit--; |
| 1832 | if (ut->chunkOffset > ut->chunkLength) { |
| 1833 | ut->chunkOffset = ut->chunkLength; |
| 1834 | } |
| 1835 | } |
| 1836 | |
| 1837 | // if the first UChar in the chunk could be the trailing half of a surrogate pair, |
| 1838 | // trim it off. |
| 1839 | if(ut->chunkNativeStart>0 && U16_IS_TRAIL(ex->s[0])(((ex->s[0])&0xfffffc00)==0xdc00)) { |
| 1840 | ++(ut->chunkContents); |
| 1841 | ++(ut->chunkNativeStart); |
| 1842 | --(ut->chunkLength); |
| 1843 | --(ut->chunkOffset); |
| 1844 | } |
| 1845 | |
| 1846 | // adjust the index/chunkOffset to a code point boundary |
| 1847 | U16_SET_CP_START(ut->chunkContents, 0, ut->chunkOffset)do { if(((((ut->chunkContents)[ut->chunkOffset])&0xfffffc00 )==0xdc00) && (ut->chunkOffset)>(0) && ( (((ut->chunkContents)[(ut->chunkOffset)-1])&0xfffffc00 )==0xd800)) { --(ut->chunkOffset); } } while (false); |
| 1848 | |
| 1849 | // Use fast indexing for get/setNativeIndex() |
| 1850 | ut->nativeIndexingLimit = ut->chunkLength; |
| 1851 | |
| 1852 | return TRUE1; |
| 1853 | } |
| 1854 | |
| 1855 | |
| 1856 | |
| 1857 | static int32_t U_CALLCONV |
| 1858 | repTextExtract(UText *ut, |
| 1859 | int64_t start, int64_t limit, |
| 1860 | UChar *dest, int32_t destCapacity, |
| 1861 | UErrorCode *status) { |
| 1862 | const Replaceable *rep=(const Replaceable *)ut->context; |
| 1863 | int32_t length=rep->length(); |
| 1864 | |
| 1865 | if(U_FAILURE(*status)) { |
| 1866 | return 0; |
| 1867 | } |
| 1868 | if(destCapacity<0 || (dest==NULL__null && destCapacity>0)) { |
| 1869 | *status=U_ILLEGAL_ARGUMENT_ERROR; |
| 1870 | } |
| 1871 | if(start>limit) { |
| 1872 | *status=U_INDEX_OUTOFBOUNDS_ERROR; |
| 1873 | return 0; |
| 1874 | } |
| 1875 | |
| 1876 | int32_t start32 = pinIndex(start, length); |
| 1877 | int32_t limit32 = pinIndex(limit, length); |
| 1878 | |
| 1879 | // adjust start, limit if they point to trail half of surrogates |
| 1880 | if (start32<length && U16_IS_TRAIL(rep->charAt(start32))(((rep->charAt(start32))&0xfffffc00)==0xdc00) && |
| 1881 | U_IS_SUPPLEMENTARY(rep->char32At(start32))((uint32_t)((rep->char32At(start32))-0x10000)<=0xfffff)){ |
| 1882 | start32--; |
| 1883 | } |
| 1884 | if (limit32<length && U16_IS_TRAIL(rep->charAt(limit32))(((rep->charAt(limit32))&0xfffffc00)==0xdc00) && |
| 1885 | U_IS_SUPPLEMENTARY(rep->char32At(limit32))((uint32_t)((rep->char32At(limit32))-0x10000)<=0xfffff)){ |
| 1886 | limit32--; |
| 1887 | } |
| 1888 | |
| 1889 | length=limit32-start32; |
| 1890 | if(length>destCapacity) { |
| 1891 | limit32 = start32 + destCapacity; |
| 1892 | } |
| 1893 | UnicodeString buffer(dest, 0, destCapacity); // writable alias |
| 1894 | rep->extractBetween(start32, limit32, buffer); |
| 1895 | repTextAccess(ut, limit32, TRUE1); |
| 1896 | |
| 1897 | return u_terminateUCharsu_terminateUChars_71(dest, destCapacity, length, status); |
| 1898 | } |
| 1899 | |
| 1900 | static int32_t U_CALLCONV |
| 1901 | repTextReplace(UText *ut, |
| 1902 | int64_t start, int64_t limit, |
| 1903 | const UChar *src, int32_t length, |
| 1904 | UErrorCode *status) { |
| 1905 | Replaceable *rep=(Replaceable *)ut->context; |
| 1906 | int32_t oldLength; |
| 1907 | |
| 1908 | if(U_FAILURE(*status)) { |
| 1909 | return 0; |
| 1910 | } |
| 1911 | if(src==NULL__null && length!=0) { |
| 1912 | *status=U_ILLEGAL_ARGUMENT_ERROR; |
| 1913 | return 0; |
| 1914 | } |
| 1915 | oldLength=rep->length(); // will subtract from new length |
| 1916 | if(start>limit ) { |
| 1917 | *status=U_INDEX_OUTOFBOUNDS_ERROR; |
| 1918 | return 0; |
| 1919 | } |
| 1920 | |
| 1921 | int32_t start32 = pinIndex(start, oldLength); |
| 1922 | int32_t limit32 = pinIndex(limit, oldLength); |
| 1923 | |
| 1924 | // Snap start & limit to code point boundaries. |
| 1925 | if (start32<oldLength && U16_IS_TRAIL(rep->charAt(start32))(((rep->charAt(start32))&0xfffffc00)==0xdc00) && |
| 1926 | start32>0 && U16_IS_LEAD(rep->charAt(start32-1))(((rep->charAt(start32-1))&0xfffffc00)==0xd800)) |
| 1927 | { |
| 1928 | start32--; |
| 1929 | } |
| 1930 | if (limit32<oldLength && U16_IS_LEAD(rep->charAt(limit32-1))(((rep->charAt(limit32-1))&0xfffffc00)==0xd800) && |
| 1931 | U16_IS_TRAIL(rep->charAt(limit32))(((rep->charAt(limit32))&0xfffffc00)==0xdc00)) |
| 1932 | { |
| 1933 | limit32++; |
| 1934 | } |
| 1935 | |
| 1936 | // Do the actual replace operation using methods of the Replaceable class |
| 1937 | UnicodeString replStr((UBool)(length<0), src, length); // read-only alias |
| 1938 | rep->handleReplaceBetween(start32, limit32, replStr); |
| 1939 | int32_t newLength = rep->length(); |
| 1940 | int32_t lengthDelta = newLength - oldLength; |
| 1941 | |
| 1942 | // Is the UText chunk buffer OK? |
| 1943 | if (ut->chunkNativeLimit > start32) { |
| 1944 | // this replace operation may have impacted the current chunk. |
| 1945 | // invalidate it, which will force a reload on the next access. |
| 1946 | invalidateChunk(ut); |
| 1947 | } |
| 1948 | |
| 1949 | // set the iteration position to the end of the newly inserted replacement text. |
| 1950 | int32_t newIndexPos = limit32 + lengthDelta; |
| 1951 | repTextAccess(ut, newIndexPos, TRUE1); |
| 1952 | |
| 1953 | return lengthDelta; |
| 1954 | } |
| 1955 | |
| 1956 | |
| 1957 | static void U_CALLCONV |
| 1958 | repTextCopy(UText *ut, |
| 1959 | int64_t start, int64_t limit, |
| 1960 | int64_t destIndex, |
| 1961 | UBool move, |
| 1962 | UErrorCode *status) |
| 1963 | { |
| 1964 | Replaceable *rep=(Replaceable *)ut->context; |
| 1965 | int32_t length=rep->length(); |
| 1966 | |
| 1967 | if(U_FAILURE(*status)) { |
| 1968 | return; |
| 1969 | } |
| 1970 | if (start>limit || (start<destIndex && destIndex<limit)) |
| 1971 | { |
| 1972 | *status=U_INDEX_OUTOFBOUNDS_ERROR; |
| 1973 | return; |
| 1974 | } |
| 1975 | |
| 1976 | int32_t start32 = pinIndex(start, length); |
| 1977 | int32_t limit32 = pinIndex(limit, length); |
| 1978 | int32_t destIndex32 = pinIndex(destIndex, length); |
| 1979 | |
| 1980 | // TODO: snap input parameters to code point boundaries. |
| 1981 | |
| 1982 | if(move) { |
| 1983 | // move: copy to destIndex, then replace original with nothing |
| 1984 | int32_t segLength=limit32-start32; |
| 1985 | rep->copy(start32, limit32, destIndex32); |
| 1986 | if(destIndex32<start32) { |
| 1987 | start32+=segLength; |
| 1988 | limit32+=segLength; |
| 1989 | } |
| 1990 | rep->handleReplaceBetween(start32, limit32, UnicodeString()); |
| 1991 | } else { |
| 1992 | // copy |
| 1993 | rep->copy(start32, limit32, destIndex32); |
| 1994 | } |
| 1995 | |
| 1996 | // If the change to the text touched the region in the chunk buffer, |
| 1997 | // invalidate the buffer. |
| 1998 | int32_t firstAffectedIndex = destIndex32; |
| 1999 | if (move && start32<firstAffectedIndex) { |
| 2000 | firstAffectedIndex = start32; |
| 2001 | } |
| 2002 | if (firstAffectedIndex < ut->chunkNativeLimit) { |
| 2003 | // changes may have affected range covered by the chunk |
| 2004 | invalidateChunk(ut); |
| 2005 | } |
| 2006 | |
| 2007 | // Put iteration position at the newly inserted (moved) block, |
| 2008 | int32_t nativeIterIndex = destIndex32 + limit32 - start32; |
| 2009 | if (move && destIndex32>start32) { |
| 2010 | // moved a block of text towards the end of the string. |
| 2011 | nativeIterIndex = destIndex32; |
| 2012 | } |
| 2013 | |
| 2014 | // Set position, reload chunk if needed. |
| 2015 | repTextAccess(ut, nativeIterIndex, TRUE1); |
| 2016 | } |
| 2017 | |
| 2018 | static const struct UTextFuncs repFuncs = |
| 2019 | { |
| 2020 | sizeof(UTextFuncs), |
| 2021 | 0, 0, 0, // Reserved alignment padding |
| 2022 | repTextClone, |
| 2023 | repTextLength, |
| 2024 | repTextAccess, |
| 2025 | repTextExtract, |
| 2026 | repTextReplace, |
| 2027 | repTextCopy, |
| 2028 | NULL__null, // MapOffsetToNative, |
| 2029 | NULL__null, // MapIndexToUTF16, |
| 2030 | repTextClose, |
| 2031 | NULL__null, // spare 1 |
| 2032 | NULL__null, // spare 2 |
| 2033 | NULL__null // spare 3 |
| 2034 | }; |
| 2035 | |
| 2036 | |
| 2037 | U_CAPIextern "C" UText * U_EXPORT2 |
| 2038 | utext_openReplaceableutext_openReplaceable_71(UText *ut, Replaceable *rep, UErrorCode *status) |
| 2039 | { |
| 2040 | if(U_FAILURE(*status)) { |
| 2041 | return NULL__null; |
| 2042 | } |
| 2043 | if(rep==NULL__null) { |
| 2044 | *status=U_ILLEGAL_ARGUMENT_ERROR; |
| 2045 | return NULL__null; |
| 2046 | } |
| 2047 | ut = utext_setuputext_setup_71(ut, sizeof(ReplExtra), status); |
| 2048 | if(U_FAILURE(*status)) { |
| 2049 | return ut; |
| 2050 | } |
| 2051 | |
| 2052 | ut->providerProperties = I32_FLAG(UTEXT_PROVIDER_WRITABLE)((int32_t)1<<(UTEXT_PROVIDER_WRITABLE)); |
| 2053 | if(rep->hasMetaData()) { |
| 2054 | ut->providerProperties |=I32_FLAG(UTEXT_PROVIDER_HAS_META_DATA)((int32_t)1<<(UTEXT_PROVIDER_HAS_META_DATA)); |
| 2055 | } |
| 2056 | |
| 2057 | ut->pFuncs = &repFuncs; |
| 2058 | ut->context = rep; |
| 2059 | return ut; |
| 2060 | } |
| 2061 | |
| 2062 | U_CDECL_END} |
| 2063 | |
| 2064 | |
| 2065 | |
| 2066 | |
| 2067 | |
| 2068 | |
| 2069 | |
| 2070 | |
| 2071 | //------------------------------------------------------------------------------ |
| 2072 | // |
| 2073 | // UText implementation for UnicodeString (read/write) and |
| 2074 | // for const UnicodeString (read only) |
| 2075 | // (same implementation, only the flags are different) |
| 2076 | // |
| 2077 | // Use of UText data members: |
| 2078 | // context pointer to UnicodeString |
| 2079 | // p pointer to UnicodeString IF this UText owns the string |
| 2080 | // and it must be deleted on close(). NULL otherwise. |
| 2081 | // |
| 2082 | //------------------------------------------------------------------------------ |
| 2083 | |
| 2084 | U_CDECL_BEGINextern "C" { |
| 2085 | |
| 2086 | |
| 2087 | static UText * U_CALLCONV |
| 2088 | unistrTextClone(UText *dest, const UText *src, UBool deep, UErrorCode *status) { |
| 2089 | // First do a generic shallow clone. Does everything needed for the UText struct itself. |
| 2090 | dest = shallowTextClone(dest, src, status); |
| 2091 | |
| 2092 | // For deep clones, make a copy of the UnicodeSring. |
| 2093 | // The copied UnicodeString storage is owned by the newly created UText clone. |
| 2094 | // A non-NULL pointer in UText.p is the signal to the close() function to delete |
| 2095 | // the UText. |
| 2096 | // |
| 2097 | if (deep && U_SUCCESS(*status)) { |
| 2098 | const UnicodeString *srcString = (const UnicodeString *)src->context; |
| 2099 | dest->context = new UnicodeString(*srcString); |
| 2100 | dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT)); |
| 2101 | |
| 2102 | // with deep clone, the copy is writable, even when the source is not. |
| 2103 | dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_WRITABLE)((int32_t)1<<(UTEXT_PROVIDER_WRITABLE)); |
| 2104 | } |
| 2105 | return dest; |
| 2106 | } |
| 2107 | |
| 2108 | static void U_CALLCONV |
| 2109 | unistrTextClose(UText *ut) { |
| 2110 | // Most of the work of close is done by the generic UText framework close. |
| 2111 | // All that needs to be done here is delete the UnicodeString if the UText |
| 2112 | // owns it. This occurs if the UText was created by cloning. |
| 2113 | if (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT))) { |
| 2114 | UnicodeString *str = (UnicodeString *)ut->context; |
| 2115 | delete str; |
| 2116 | ut->context = NULL__null; |
| 2117 | } |
| 2118 | } |
| 2119 | |
| 2120 | |
| 2121 | static int64_t U_CALLCONV |
| 2122 | unistrTextLength(UText *t) { |
| 2123 | return ((const UnicodeString *)t->context)->length(); |
| 2124 | } |
| 2125 | |
| 2126 | |
| 2127 | static UBool U_CALLCONV |
| 2128 | unistrTextAccess(UText *ut, int64_t index, UBool forward) { |
| 2129 | int32_t length = ut->chunkLength; |
| 2130 | ut->chunkOffset = pinIndex(index, length); |
| 2131 | |
| 2132 | // Check whether request is at the start or end |
| 2133 | UBool retVal = (forward && index<length) || (!forward && index>0); |
| 2134 | return retVal; |
| 2135 | } |
| 2136 | |
| 2137 | |
| 2138 | |
| 2139 | static int32_t U_CALLCONV |
| 2140 | unistrTextExtract(UText *t, |
| 2141 | int64_t start, int64_t limit, |
| 2142 | UChar *dest, int32_t destCapacity, |
| 2143 | UErrorCode *pErrorCode) { |
| 2144 | const UnicodeString *us=(const UnicodeString *)t->context; |
| 2145 | int32_t length=us->length(); |
| 2146 | |
| 2147 | if(U_FAILURE(*pErrorCode)) { |
| 2148 | return 0; |
| 2149 | } |
| 2150 | if(destCapacity<0 || (dest==NULL__null && destCapacity>0)) { |
| 2151 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 2152 | } |
| 2153 | if(start<0 || start>limit) { |
| 2154 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
| 2155 | return 0; |
| 2156 | } |
| 2157 | |
| 2158 | int32_t start32 = start<length ? us->getChar32Start((int32_t)start) : length; |
| 2159 | int32_t limit32 = limit<length ? us->getChar32Start((int32_t)limit) : length; |
| 2160 | |
| 2161 | length=limit32-start32; |
| 2162 | if (destCapacity>0 && dest!=NULL__null) { |
| 2163 | int32_t trimmedLength = length; |
| 2164 | if(trimmedLength>destCapacity) { |
| 2165 | trimmedLength=destCapacity; |
| 2166 | } |
| 2167 | us->extract(start32, trimmedLength, dest); |
| 2168 | t->chunkOffset = start32+trimmedLength; |
| 2169 | } else { |
| 2170 | t->chunkOffset = start32; |
| 2171 | } |
| 2172 | u_terminateUCharsu_terminateUChars_71(dest, destCapacity, length, pErrorCode); |
| 2173 | return length; |
| 2174 | } |
| 2175 | |
| 2176 | static int32_t U_CALLCONV |
| 2177 | unistrTextReplace(UText *ut, |
| 2178 | int64_t start, int64_t limit, |
| 2179 | const UChar *src, int32_t length, |
| 2180 | UErrorCode *pErrorCode) { |
| 2181 | UnicodeString *us=(UnicodeString *)ut->context; |
| 2182 | int32_t oldLength; |
| 2183 | |
| 2184 | if(U_FAILURE(*pErrorCode)) { |
| 2185 | return 0; |
| 2186 | } |
| 2187 | if(src==NULL__null && length!=0) { |
| 2188 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 2189 | } |
| 2190 | if(start>limit) { |
| 2191 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
| 2192 | return 0; |
| 2193 | } |
| 2194 | oldLength=us->length(); |
| 2195 | int32_t start32 = pinIndex(start, oldLength); |
| 2196 | int32_t limit32 = pinIndex(limit, oldLength); |
| 2197 | if (start32 < oldLength) { |
| 2198 | start32 = us->getChar32Start(start32); |
| 2199 | } |
| 2200 | if (limit32 < oldLength) { |
| 2201 | limit32 = us->getChar32Start(limit32); |
| 2202 | } |
| 2203 | |
| 2204 | // replace |
| 2205 | us->replace(start32, limit32-start32, src, length); |
| 2206 | int32_t newLength = us->length(); |
| 2207 | |
| 2208 | // Update the chunk description. |
| 2209 | ut->chunkContents = us->getBuffer(); |
| 2210 | ut->chunkLength = newLength; |
| 2211 | ut->chunkNativeLimit = newLength; |
| 2212 | ut->nativeIndexingLimit = newLength; |
| 2213 | |
| 2214 | // Set iteration position to the point just following the newly inserted text. |
| 2215 | int32_t lengthDelta = newLength - oldLength; |
| 2216 | ut->chunkOffset = limit32 + lengthDelta; |
| 2217 | |
| 2218 | return lengthDelta; |
| 2219 | } |
| 2220 | |
| 2221 | static void U_CALLCONV |
| 2222 | unistrTextCopy(UText *ut, |
| 2223 | int64_t start, int64_t limit, |
| 2224 | int64_t destIndex, |
| 2225 | UBool move, |
| 2226 | UErrorCode *pErrorCode) { |
| 2227 | UnicodeString *us=(UnicodeString *)ut->context; |
| 2228 | int32_t length=us->length(); |
| 2229 | |
| 2230 | if(U_FAILURE(*pErrorCode)) { |
| 2231 | return; |
| 2232 | } |
| 2233 | int32_t start32 = pinIndex(start, length); |
| 2234 | int32_t limit32 = pinIndex(limit, length); |
| 2235 | int32_t destIndex32 = pinIndex(destIndex, length); |
| 2236 | |
| 2237 | if( start32>limit32 || (start32<destIndex32 && destIndex32<limit32)) { |
| 2238 | *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; |
| 2239 | return; |
| 2240 | } |
| 2241 | |
| 2242 | if(move) { |
| 2243 | // move: copy to destIndex, then remove original |
| 2244 | int32_t segLength=limit32-start32; |
| 2245 | us->copy(start32, limit32, destIndex32); |
| 2246 | if(destIndex32<start32) { |
| 2247 | start32+=segLength; |
| 2248 | } |
| 2249 | us->remove(start32, segLength); |
| 2250 | } else { |
| 2251 | // copy |
| 2252 | us->copy(start32, limit32, destIndex32); |
| 2253 | } |
| 2254 | |
| 2255 | // update chunk description, set iteration position. |
| 2256 | ut->chunkContents = us->getBuffer(); |
| 2257 | if (move==FALSE0) { |
| 2258 | // copy operation, string length grows |
| 2259 | ut->chunkLength += limit32-start32; |
| 2260 | ut->chunkNativeLimit = ut->chunkLength; |
| 2261 | ut->nativeIndexingLimit = ut->chunkLength; |
| 2262 | } |
| 2263 | |
| 2264 | // Iteration position to end of the newly inserted text. |
| 2265 | ut->chunkOffset = destIndex32+limit32-start32; |
| 2266 | if (move && destIndex32>start32) { |
| 2267 | ut->chunkOffset = destIndex32; |
| 2268 | } |
| 2269 | |
| 2270 | } |
| 2271 | |
| 2272 | static const struct UTextFuncs unistrFuncs = |
| 2273 | { |
| 2274 | sizeof(UTextFuncs), |
| 2275 | 0, 0, 0, // Reserved alignment padding |
| 2276 | unistrTextClone, |
| 2277 | unistrTextLength, |
| 2278 | unistrTextAccess, |
| 2279 | unistrTextExtract, |
| 2280 | unistrTextReplace, |
| 2281 | unistrTextCopy, |
| 2282 | NULL__null, // MapOffsetToNative, |
| 2283 | NULL__null, // MapIndexToUTF16, |
| 2284 | unistrTextClose, |
| 2285 | NULL__null, // spare 1 |
| 2286 | NULL__null, // spare 2 |
| 2287 | NULL__null // spare 3 |
| 2288 | }; |
| 2289 | |
| 2290 | |
| 2291 | |
| 2292 | U_CDECL_END} |
| 2293 | |
| 2294 | |
| 2295 | U_CAPIextern "C" UText * U_EXPORT2 |
| 2296 | utext_openUnicodeStringutext_openUnicodeString_71(UText *ut, UnicodeString *s, UErrorCode *status) { |
| 2297 | ut = utext_openConstUnicodeStringutext_openConstUnicodeString_71(ut, s, status); |
| 2298 | if (U_SUCCESS(*status)) { |
| 2299 | ut->providerProperties |= I32_FLAG(UTEXT_PROVIDER_WRITABLE)((int32_t)1<<(UTEXT_PROVIDER_WRITABLE)); |
| 2300 | } |
| 2301 | return ut; |
| 2302 | } |
| 2303 | |
| 2304 | |
| 2305 | |
| 2306 | U_CAPIextern "C" UText * U_EXPORT2 |
| 2307 | utext_openConstUnicodeStringutext_openConstUnicodeString_71(UText *ut, const UnicodeString *s, UErrorCode *status) { |
| 2308 | if (U_SUCCESS(*status) && s->isBogus()) { |
| 2309 | // The UnicodeString is bogus, but we still need to detach the UText |
| 2310 | // from whatever it was hooked to before, if anything. |
| 2311 | utext_openUCharsutext_openUChars_71(ut, NULL__null, 0, status); |
| 2312 | *status = U_ILLEGAL_ARGUMENT_ERROR; |
| 2313 | return ut; |
| 2314 | } |
| 2315 | ut = utext_setuputext_setup_71(ut, 0, status); |
| 2316 | // note: use the standard (writable) function table for UnicodeString. |
| 2317 | // The flag settings disable writing, so having the functions in |
| 2318 | // the table is harmless. |
| 2319 | if (U_SUCCESS(*status)) { |
| 2320 | ut->pFuncs = &unistrFuncs; |
| 2321 | ut->context = s; |
| 2322 | ut->providerProperties = I32_FLAG(UTEXT_PROVIDER_STABLE_CHUNKS)((int32_t)1<<(UTEXT_PROVIDER_STABLE_CHUNKS)); |
| 2323 | ut->chunkContents = s->getBuffer(); |
| 2324 | ut->chunkLength = s->length(); |
| 2325 | ut->chunkNativeStart = 0; |
| 2326 | ut->chunkNativeLimit = ut->chunkLength; |
| 2327 | ut->nativeIndexingLimit = ut->chunkLength; |
| 2328 | } |
| 2329 | return ut; |
| 2330 | } |
| 2331 | |
| 2332 | //------------------------------------------------------------------------------ |
| 2333 | // |
| 2334 | // UText implementation for const UChar * strings |
| 2335 | // |
| 2336 | // Use of UText data members: |
| 2337 | // context pointer to UnicodeString |
| 2338 | // a length. -1 if not yet known. |
| 2339 | // |
| 2340 | // TODO: support 64 bit lengths. |
| 2341 | // |
| 2342 | //------------------------------------------------------------------------------ |
| 2343 | |
| 2344 | U_CDECL_BEGINextern "C" { |
| 2345 | |
| 2346 | |
| 2347 | static UText * U_CALLCONV |
| 2348 | ucstrTextClone(UText *dest, const UText * src, UBool deep, UErrorCode * status) { |
| 2349 | // First do a generic shallow clone. |
| 2350 | dest = shallowTextClone(dest, src, status); |
| 2351 | |
| 2352 | // For deep clones, make a copy of the string. |
| 2353 | // The copied storage is owned by the newly created clone. |
| 2354 | // A non-NULL pointer in UText.p is the signal to the close() function to delete |
| 2355 | // it. |
| 2356 | // |
| 2357 | if (deep && U_SUCCESS(*status)) { |
| 2358 | U_ASSERT(utext_nativeLength(dest) < INT32_MAX)(void)0; |
| 2359 | int32_t len = (int32_t)utext_nativeLengthutext_nativeLength_71(dest); |
| 2360 | |
| 2361 | // The cloned string IS going to be NUL terminated, whether or not the original was. |
| 2362 | const UChar *srcStr = (const UChar *)src->context; |
| 2363 | UChar *copyStr = (UChar *)uprv_mallocuprv_malloc_71((len+1) * sizeof(UChar)); |
| 2364 | if (copyStr == NULL__null) { |
| 2365 | *status = U_MEMORY_ALLOCATION_ERROR; |
| 2366 | } else { |
| 2367 | int64_t i; |
| 2368 | for (i=0; i<len; i++) { |
| 2369 | copyStr[i] = srcStr[i]; |
| 2370 | } |
| 2371 | copyStr[len] = 0; |
| 2372 | dest->context = copyStr; |
| 2373 | dest->providerProperties |= I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT)); |
| 2374 | } |
| 2375 | } |
| 2376 | return dest; |
| 2377 | } |
| 2378 | |
| 2379 | |
| 2380 | static void U_CALLCONV |
| 2381 | ucstrTextClose(UText *ut) { |
| 2382 | // Most of the work of close is done by the generic UText framework close. |
| 2383 | // All that needs to be done here is delete the string if the UText |
| 2384 | // owns it. This occurs if the UText was created by cloning. |
| 2385 | if (ut->providerProperties & I32_FLAG(UTEXT_PROVIDER_OWNS_TEXT)((int32_t)1<<(UTEXT_PROVIDER_OWNS_TEXT))) { |
| 2386 | UChar *s = (UChar *)ut->context; |
| 2387 | uprv_freeuprv_free_71(s); |
| 2388 | ut->context = NULL__null; |
| 2389 | } |
| 2390 | } |
| 2391 | |
| 2392 | |
| 2393 | |
| 2394 | static int64_t U_CALLCONV |
| 2395 | ucstrTextLength(UText *ut) { |
| 2396 | if (ut->a < 0) { |
| 2397 | // null terminated, we don't yet know the length. Scan for it. |
| 2398 | // Access is not convenient for doing this |
| 2399 | // because the current iteration position can't be changed. |
| 2400 | const UChar *str = (const UChar *)ut->context; |
| 2401 | for (;;) { |
| 2402 | if (str[ut->chunkNativeLimit] == 0) { |
| 2403 | break; |
| 2404 | } |
| 2405 | ut->chunkNativeLimit++; |
| 2406 | } |
| 2407 | ut->a = ut->chunkNativeLimit; |
| 2408 | ut->chunkLength = (int32_t)ut->chunkNativeLimit; |
| 2409 | ut->nativeIndexingLimit = ut->chunkLength; |
| 2410 | ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)); |
| 2411 | } |
| 2412 | return ut->a; |
| 2413 | } |
| 2414 | |
| 2415 | |
| 2416 | static UBool U_CALLCONV |
| 2417 | ucstrTextAccess(UText *ut, int64_t index, UBool forward) { |
| 2418 | const UChar *str = (const UChar *)ut->context; |
| 2419 | |
| 2420 | // pin the requested index to the bounds of the string, |
| 2421 | // and set current iteration position. |
| 2422 | if (index<0) { |
| 2423 | index = 0; |
| 2424 | } else if (index < ut->chunkNativeLimit) { |
| 2425 | // The request data is within the chunk as it is known so far. |
| 2426 | // Put index on a code point boundary. |
| 2427 | U16_SET_CP_START(str, 0, index)do { if(((((str)[index])&0xfffffc00)==0xdc00) && ( index)>(0) && ((((str)[(index)-1])&0xfffffc00) ==0xd800)) { --(index); } } while (false); |
| 2428 | } else if (ut->a >= 0) { |
| 2429 | // We know the length of this string, and the user is requesting something |
| 2430 | // at or beyond the length. Pin the requested index to the length. |
| 2431 | index = ut->a; |
| 2432 | } else { |
| 2433 | // Null terminated string, length not yet known, and the requested index |
| 2434 | // is beyond where we have scanned so far. |
| 2435 | // Scan to 32 UChars beyond the requested index. The strategy here is |
| 2436 | // to avoid fully scanning a long string when the caller only wants to |
| 2437 | // see a few characters at its beginning. |
| 2438 | int32_t scanLimit = (int32_t)index + 32; |
| 2439 | if ((index + 32)>INT32_MAX(2147483647) || (index + 32)<0 ) { // note: int64 expression |
| 2440 | scanLimit = INT32_MAX(2147483647); |
| 2441 | } |
| 2442 | |
| 2443 | int32_t chunkLimit = (int32_t)ut->chunkNativeLimit; |
| 2444 | for (; chunkLimit<scanLimit; chunkLimit++) { |
| 2445 | if (str[chunkLimit] == 0) { |
| 2446 | // We found the end of the string. Remember it, pin the requested index to it, |
| 2447 | // and bail out of here. |
| 2448 | ut->a = chunkLimit; |
| 2449 | ut->chunkLength = chunkLimit; |
| 2450 | ut->nativeIndexingLimit = chunkLimit; |
| 2451 | if (index >= chunkLimit) { |
| 2452 | index = chunkLimit; |
| 2453 | } else { |
| 2454 | U16_SET_CP_START(str, 0, index)do { if(((((str)[index])&0xfffffc00)==0xdc00) && ( index)>(0) && ((((str)[(index)-1])&0xfffffc00) ==0xd800)) { --(index); } } while (false); |
| 2455 | } |
| 2456 | |
| 2457 | ut->chunkNativeLimit = chunkLimit; |
| 2458 | ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)); |
| 2459 | goto breakout; |
| 2460 | } |
| 2461 | } |
| 2462 | // We scanned through the next batch of UChars without finding the end. |
| 2463 | U16_SET_CP_START(str, 0, index)do { if(((((str)[index])&0xfffffc00)==0xdc00) && ( index)>(0) && ((((str)[(index)-1])&0xfffffc00) ==0xd800)) { --(index); } } while (false); |
| 2464 | if (chunkLimit == INT32_MAX(2147483647)) { |
| 2465 | // Scanned to the limit of a 32 bit length. |
| 2466 | // Forceably trim the overlength string back so length fits in int32 |
| 2467 | // TODO: add support for 64 bit strings. |
| 2468 | ut->a = chunkLimit; |
| 2469 | ut->chunkLength = chunkLimit; |
| 2470 | ut->nativeIndexingLimit = chunkLimit; |
| 2471 | if (index > chunkLimit) { |
| 2472 | index = chunkLimit; |
| 2473 | } |
| 2474 | ut->chunkNativeLimit = chunkLimit; |
| 2475 | ut->providerProperties &= ~I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)); |
| 2476 | } else { |
| 2477 | // The endpoint of a chunk must not be left in the middle of a surrogate pair. |
| 2478 | // If the current end is on a lead surrogate, back the end up by one. |
| 2479 | // It doesn't matter if the end char happens to be an unpaired surrogate, |
| 2480 | // and it's simpler not to worry about it. |
| 2481 | if (U16_IS_LEAD(str[chunkLimit-1])(((str[chunkLimit-1])&0xfffffc00)==0xd800)) { |
| 2482 | --chunkLimit; |
| 2483 | } |
| 2484 | // Null-terminated chunk with end still unknown. |
| 2485 | // Update the chunk length to reflect what has been scanned thus far. |
| 2486 | // That the full length is still unknown is (still) flagged by |
| 2487 | // ut->a being < 0. |
| 2488 | ut->chunkNativeLimit = chunkLimit; |
| 2489 | ut->nativeIndexingLimit = chunkLimit; |
| 2490 | ut->chunkLength = chunkLimit; |
| 2491 | } |
| 2492 | |
| 2493 | } |
| 2494 | breakout: |
| 2495 | U_ASSERT(index<=INT32_MAX)(void)0; |
| 2496 | ut->chunkOffset = (int32_t)index; |
| 2497 | |
| 2498 | // Check whether request is at the start or end |
| 2499 | UBool retVal = (forward && index<ut->chunkNativeLimit) || (!forward && index>0); |
| 2500 | return retVal; |
| 2501 | } |
| 2502 | |
| 2503 | |
| 2504 | |
| 2505 | static int32_t U_CALLCONV |
| 2506 | ucstrTextExtract(UText *ut, |
| 2507 | int64_t start, int64_t limit, |
| 2508 | UChar *dest, int32_t destCapacity, |
| 2509 | UErrorCode *pErrorCode) |
| 2510 | { |
| 2511 | if(U_FAILURE(*pErrorCode)) { |
| 2512 | return 0; |
| 2513 | } |
| 2514 | if(destCapacity<0 || (dest==NULL__null && destCapacity>0) || start>limit) { |
| 2515 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
| 2516 | return 0; |
| 2517 | } |
| 2518 | |
| 2519 | //const UChar *s=(const UChar *)ut->context; |
| 2520 | int32_t si, di; |
| 2521 | |
| 2522 | int32_t start32; |
| 2523 | int32_t limit32; |
| 2524 | |
| 2525 | // Access the start. Does two things we need: |
| 2526 | // Pins 'start' to the length of the string, if it came in out-of-bounds. |
| 2527 | // Snaps 'start' to the beginning of a code point. |
| 2528 | ucstrTextAccess(ut, start, TRUE1); |
| 2529 | const UChar *s=ut->chunkContents; |
| 2530 | start32 = ut->chunkOffset; |
| 2531 | |
| 2532 | int32_t strLength=(int32_t)ut->a; |
| 2533 | if (strLength >= 0) { |
| 2534 | limit32 = pinIndex(limit, strLength); |
| 2535 | } else { |
| 2536 | limit32 = pinIndex(limit, INT32_MAX(2147483647)); |
| 2537 | } |
| 2538 | di = 0; |
| 2539 | for (si=start32; si<limit32; si++) { |
| 2540 | if (strLength<0 && s[si]==0) { |
| 2541 | // Just hit the end of a null-terminated string. |
| 2542 | ut->a = si; // set string length for this UText |
| 2543 | ut->chunkNativeLimit = si; |
| 2544 | ut->chunkLength = si; |
| 2545 | ut->nativeIndexingLimit = si; |
| 2546 | strLength = si; |
| 2547 | limit32 = si; |
Value stored to 'limit32' is never read | |
| 2548 | break; |
| 2549 | } |
| 2550 | U_ASSERT(di>=0)(void)0; /* to ensure di never exceeds INT32_MAX, which must not happen logically */ |
| 2551 | if (di<destCapacity) { |
| 2552 | // only store if there is space. |
| 2553 | dest[di] = s[si]; |
| 2554 | } else { |
| 2555 | if (strLength>=0) { |
| 2556 | // We have filled the destination buffer, and the string length is known. |
| 2557 | // Cut the loop short. There is no need to scan string termination. |
| 2558 | di = limit32 - start32; |
| 2559 | si = limit32; |
| 2560 | break; |
| 2561 | } |
| 2562 | } |
| 2563 | di++; |
| 2564 | } |
| 2565 | |
| 2566 | // If the limit index points to a lead surrogate of a pair, |
| 2567 | // add the corresponding trail surrogate to the destination. |
| 2568 | if (si>0 && U16_IS_LEAD(s[si-1])(((s[si-1])&0xfffffc00)==0xd800) && |
| 2569 | ((si<strLength || strLength<0) && U16_IS_TRAIL(s[si])(((s[si])&0xfffffc00)==0xdc00))) |
| 2570 | { |
| 2571 | if (di<destCapacity) { |
| 2572 | // store only if there is space in the output buffer. |
| 2573 | dest[di++] = s[si]; |
| 2574 | } |
| 2575 | si++; |
| 2576 | } |
| 2577 | |
| 2578 | // Put iteration position at the point just following the extracted text |
| 2579 | if (si <= ut->chunkNativeLimit) { |
| 2580 | ut->chunkOffset = si; |
| 2581 | } else { |
| 2582 | ucstrTextAccess(ut, si, TRUE1); |
| 2583 | } |
| 2584 | |
| 2585 | // Add a terminating NUL if space in the buffer permits, |
| 2586 | // and set the error status as required. |
| 2587 | u_terminateUCharsu_terminateUChars_71(dest, destCapacity, di, pErrorCode); |
| 2588 | return di; |
| 2589 | } |
| 2590 | |
| 2591 | static const struct UTextFuncs ucstrFuncs = |
| 2592 | { |
| 2593 | sizeof(UTextFuncs), |
| 2594 | 0, 0, 0, // Reserved alignment padding |
| 2595 | ucstrTextClone, |
| 2596 | ucstrTextLength, |
| 2597 | ucstrTextAccess, |
| 2598 | ucstrTextExtract, |
| 2599 | NULL__null, // Replace |
| 2600 | NULL__null, // Copy |
| 2601 | NULL__null, // MapOffsetToNative, |
| 2602 | NULL__null, // MapIndexToUTF16, |
| 2603 | ucstrTextClose, |
| 2604 | NULL__null, // spare 1 |
| 2605 | NULL__null, // spare 2 |
| 2606 | NULL__null, // spare 3 |
| 2607 | }; |
| 2608 | |
| 2609 | U_CDECL_END} |
| 2610 | |
| 2611 | static const UChar gEmptyUString[] = {0}; |
| 2612 | |
| 2613 | U_CAPIextern "C" UText * U_EXPORT2 |
| 2614 | utext_openUCharsutext_openUChars_71(UText *ut, const UChar *s, int64_t length, UErrorCode *status) { |
| 2615 | if (U_FAILURE(*status)) { |
| 2616 | return NULL__null; |
| 2617 | } |
| 2618 | if(s==NULL__null && length==0) { |
| 2619 | s = gEmptyUString; |
| 2620 | } |
| 2621 | if (s==NULL__null || length < -1 || length>INT32_MAX(2147483647)) { |
| 2622 | *status = U_ILLEGAL_ARGUMENT_ERROR; |
| 2623 | return NULL__null; |
| 2624 | } |
| 2625 | ut = utext_setuputext_setup_71(ut, 0, status); |
| 2626 | if (U_SUCCESS(*status)) { |
| 2627 | ut->pFuncs = &ucstrFuncs; |
| 2628 | ut->context = s; |
| 2629 | ut->providerProperties = I32_FLAG(UTEXT_PROVIDER_STABLE_CHUNKS)((int32_t)1<<(UTEXT_PROVIDER_STABLE_CHUNKS)); |
| 2630 | if (length==-1) { |
| 2631 | ut->providerProperties |= I32_FLAG(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)((int32_t)1<<(UTEXT_PROVIDER_LENGTH_IS_EXPENSIVE)); |
| 2632 | } |
| 2633 | ut->a = length; |
| 2634 | ut->chunkContents = s; |
| 2635 | ut->chunkNativeStart = 0; |
| 2636 | ut->chunkNativeLimit = length>=0? length : 0; |
| 2637 | ut->chunkLength = (int32_t)ut->chunkNativeLimit; |
| 2638 | ut->chunkOffset = 0; |
| 2639 | ut->nativeIndexingLimit = ut->chunkLength; |
| 2640 | } |
| 2641 | return ut; |
| 2642 | } |
| 2643 | |
| 2644 | |
| 2645 | //------------------------------------------------------------------------------ |
| 2646 | // |
| 2647 | // UText implementation for text from ICU CharacterIterators |
| 2648 | // |
| 2649 | // Use of UText data members: |
| 2650 | // context pointer to the CharacterIterator |
| 2651 | // a length of the full text. |
| 2652 | // p pointer to buffer 1 |
| 2653 | // b start index of local buffer 1 contents |
| 2654 | // q pointer to buffer 2 |
| 2655 | // c start index of local buffer 2 contents |
| 2656 | // r pointer to the character iterator if the UText owns it. |
| 2657 | // Null otherwise. |
| 2658 | // |
| 2659 | //------------------------------------------------------------------------------ |
| 2660 | #define CIBufSize16 16 |
| 2661 | |
| 2662 | U_CDECL_BEGINextern "C" { |
| 2663 | static void U_CALLCONV |
| 2664 | charIterTextClose(UText *ut) { |
| 2665 | // Most of the work of close is done by the generic UText framework close. |
| 2666 | // All that needs to be done here is delete the CharacterIterator if the UText |
| 2667 | // owns it. This occurs if the UText was created by cloning. |
| 2668 | CharacterIterator *ci = (CharacterIterator *)ut->r; |
| 2669 | delete ci; |
| 2670 | ut->r = NULL__null; |
| 2671 | } |
| 2672 | |
| 2673 | static int64_t U_CALLCONV |
| 2674 | charIterTextLength(UText *ut) { |
| 2675 | return (int32_t)ut->a; |
| 2676 | } |
| 2677 | |
| 2678 | static UBool U_CALLCONV |
| 2679 | charIterTextAccess(UText *ut, int64_t index, UBool forward) { |
| 2680 | CharacterIterator *ci = (CharacterIterator *)ut->context; |
| 2681 | |
| 2682 | int32_t clippedIndex = (int32_t)index; |
| 2683 | if (clippedIndex<0) { |
| 2684 | clippedIndex=0; |
| 2685 | } else if (clippedIndex>=ut->a) { |
| 2686 | clippedIndex=(int32_t)ut->a; |
| 2687 | } |
| 2688 | int32_t neededIndex = clippedIndex; |
| 2689 | if (!forward && neededIndex>0) { |
| 2690 | // reverse iteration, want the position just before what was asked for. |
| 2691 | neededIndex--; |
| 2692 | } else if (forward && neededIndex==ut->a && neededIndex>0) { |
| 2693 | // Forward iteration, don't ask for something past the end of the text. |
| 2694 | neededIndex--; |
| 2695 | } |
| 2696 | |
| 2697 | // Find the native index of the start of the buffer containing what we want. |
| 2698 | neededIndex -= neededIndex % CIBufSize16; |
| 2699 | |
| 2700 | UChar *buf = NULL__null; |
| 2701 | UBool needChunkSetup = TRUE1; |
| 2702 | int i; |
| 2703 | if (ut->chunkNativeStart == neededIndex) { |
| 2704 | // The buffer we want is already the current chunk. |
| 2705 | needChunkSetup = FALSE0; |
| 2706 | } else if (ut->b == neededIndex) { |
| 2707 | // The first buffer (buffer p) has what we need. |
| 2708 | buf = (UChar *)ut->p; |
| 2709 | } else if (ut->c == neededIndex) { |
| 2710 | // The second buffer (buffer q) has what we need. |
| 2711 | buf = (UChar *)ut->q; |
| 2712 | } else { |
| 2713 | // Neither buffer already has what we need. |
| 2714 | // Load new data from the character iterator. |
| 2715 | // Use the buf that is not the current buffer. |
| 2716 | buf = (UChar *)ut->p; |
| 2717 | if (ut->p == ut->chunkContents) { |
| 2718 | buf = (UChar *)ut->q; |
| 2719 | } |
| 2720 | ci->setIndex(neededIndex); |
| 2721 | for (i=0; i<CIBufSize16; i++) { |
| 2722 | buf[i] = ci->nextPostInc(); |
| 2723 | if (i+neededIndex > ut->a) { |
| 2724 | break; |
| 2725 | } |
| 2726 | } |
| 2727 | } |
| 2728 | |
| 2729 | // We have a buffer with the data we need. |
| 2730 | // Set it up as the current chunk, if it wasn't already. |
| 2731 | if (needChunkSetup) { |
| 2732 | ut->chunkContents = buf; |
| 2733 | ut->chunkLength = CIBufSize16; |
| 2734 | ut->chunkNativeStart = neededIndex; |
| 2735 | ut->chunkNativeLimit = neededIndex + CIBufSize16; |
| 2736 | if (ut->chunkNativeLimit > ut->a) { |
| 2737 | ut->chunkNativeLimit = ut->a; |
| 2738 | ut->chunkLength = (int32_t)(ut->chunkNativeLimit)-(int32_t)(ut->chunkNativeStart); |
| 2739 | } |
| 2740 | ut->nativeIndexingLimit = ut->chunkLength; |
| 2741 | U_ASSERT(ut->chunkOffset>=0 && ut->chunkOffset<=CIBufSize)(void)0; |
| 2742 | } |
| 2743 | ut->chunkOffset = clippedIndex - (int32_t)ut->chunkNativeStart; |
| 2744 | UBool success = (forward? ut->chunkOffset<ut->chunkLength : ut->chunkOffset>0); |
| 2745 | return success; |
| 2746 | } |
| 2747 | |
| 2748 | static UText * U_CALLCONV |
| 2749 | charIterTextClone(UText *dest, const UText *src, UBool deep, UErrorCode * status) { |
| 2750 | if (U_FAILURE(*status)) { |
| 2751 | return NULL__null; |
| 2752 | } |
| 2753 | |
| 2754 | if (deep) { |
| 2755 | // There is no CharacterIterator API for cloning the underlying text storage. |
| 2756 | *status = U_UNSUPPORTED_ERROR; |
| 2757 | return NULL__null; |
| 2758 | } else { |
| 2759 | CharacterIterator *srcCI =(CharacterIterator *)src->context; |
| 2760 | srcCI = srcCI->clone(); |
| 2761 | dest = utext_openCharacterIteratorutext_openCharacterIterator_71(dest, srcCI, status); |
| 2762 | if (U_FAILURE(*status)) { |
| 2763 | return dest; |
| 2764 | } |
| 2765 | // cast off const on getNativeIndex. |
| 2766 | // For CharacterIterator based UTexts, this is safe, the operation is const. |
| 2767 | int64_t ix = utext_getNativeIndexutext_getNativeIndex_71((UText *)src); |
| 2768 | utext_setNativeIndexutext_setNativeIndex_71(dest, ix); |
| 2769 | dest->r = srcCI; // flags that this UText owns the CharacterIterator |
| 2770 | } |
| 2771 | return dest; |
| 2772 | } |
| 2773 | |
| 2774 | static int32_t U_CALLCONV |
| 2775 | charIterTextExtract(UText *ut, |
| 2776 | int64_t start, int64_t limit, |
| 2777 | UChar *dest, int32_t destCapacity, |
| 2778 | UErrorCode *status) |
| 2779 | { |
| 2780 | if(U_FAILURE(*status)) { |
| 2781 | return 0; |
| 2782 | } |
| 2783 | if(destCapacity<0 || (dest==NULL__null && destCapacity>0) || start>limit) { |
| 2784 | *status=U_ILLEGAL_ARGUMENT_ERROR; |
| 2785 | return 0; |
| 2786 | } |
| 2787 | int32_t length = (int32_t)ut->a; |
| 2788 | int32_t start32 = pinIndex(start, length); |
| 2789 | int32_t limit32 = pinIndex(limit, length); |
| 2790 | int32_t desti = 0; |
| 2791 | int32_t srci; |
| 2792 | int32_t copyLimit; |
| 2793 | |
| 2794 | CharacterIterator *ci = (CharacterIterator *)ut->context; |
| 2795 | ci->setIndex32(start32); // Moves ix to lead of surrogate pair, if needed. |
| 2796 | srci = ci->getIndex(); |
| 2797 | copyLimit = srci; |
| 2798 | while (srci<limit32) { |
| 2799 | UChar32 c = ci->next32PostInc(); |
| 2800 | int32_t len = U16_LENGTH(c)((uint32_t)(c)<=0xffff ? 1 : 2); |
| 2801 | U_ASSERT(desti+len>0)(void)0; /* to ensure desti+len never exceeds MAX_INT32, which must not happen logically */ |
| 2802 | if (desti+len <= destCapacity) { |
| 2803 | U16_APPEND_UNSAFE(dest, desti, c)do { if((uint32_t)(c)<=0xffff) { (dest)[(desti)++]=(uint16_t )(c); } else { (dest)[(desti)++]=(uint16_t)(((c)>>10)+0xd7c0 ); (dest)[(desti)++]=(uint16_t)(((c)&0x3ff)|0xdc00); } } while (false); |
| 2804 | copyLimit = srci+len; |
| 2805 | } else { |
| 2806 | desti += len; |
| 2807 | *status = U_BUFFER_OVERFLOW_ERROR; |
| 2808 | } |
| 2809 | srci += len; |
| 2810 | } |
| 2811 | |
| 2812 | charIterTextAccess(ut, copyLimit, TRUE1); |
| 2813 | |
| 2814 | u_terminateUCharsu_terminateUChars_71(dest, destCapacity, desti, status); |
| 2815 | return desti; |
| 2816 | } |
| 2817 | |
| 2818 | static const struct UTextFuncs charIterFuncs = |
| 2819 | { |
| 2820 | sizeof(UTextFuncs), |
| 2821 | 0, 0, 0, // Reserved alignment padding |
| 2822 | charIterTextClone, |
| 2823 | charIterTextLength, |
| 2824 | charIterTextAccess, |
| 2825 | charIterTextExtract, |
| 2826 | NULL__null, // Replace |
| 2827 | NULL__null, // Copy |
| 2828 | NULL__null, // MapOffsetToNative, |
| 2829 | NULL__null, // MapIndexToUTF16, |
| 2830 | charIterTextClose, |
| 2831 | NULL__null, // spare 1 |
| 2832 | NULL__null, // spare 2 |
| 2833 | NULL__null // spare 3 |
| 2834 | }; |
| 2835 | U_CDECL_END} |
| 2836 | |
| 2837 | |
| 2838 | U_CAPIextern "C" UText * U_EXPORT2 |
| 2839 | utext_openCharacterIteratorutext_openCharacterIterator_71(UText *ut, CharacterIterator *ci, UErrorCode *status) { |
| 2840 | if (U_FAILURE(*status)) { |
| 2841 | return NULL__null; |
| 2842 | } |
| 2843 | |
| 2844 | if (ci->startIndex() > 0) { |
| 2845 | // No support for CharacterIterators that do not start indexing from zero. |
| 2846 | *status = U_UNSUPPORTED_ERROR; |
| 2847 | return NULL__null; |
| 2848 | } |
| 2849 | |
| 2850 | // Extra space in UText for 2 buffers of CIBufSize UChars each. |
| 2851 | int32_t extraSpace = 2 * CIBufSize16 * sizeof(UChar); |
| 2852 | ut = utext_setuputext_setup_71(ut, extraSpace, status); |
| 2853 | if (U_SUCCESS(*status)) { |
| 2854 | ut->pFuncs = &charIterFuncs; |
| 2855 | ut->context = ci; |
| 2856 | ut->providerProperties = 0; |
| 2857 | ut->a = ci->endIndex(); // Length of text |
| 2858 | ut->p = ut->pExtra; // First buffer |
| 2859 | ut->b = -1; // Native index of first buffer contents |
| 2860 | ut->q = (UChar*)ut->pExtra+CIBufSize16; // Second buffer |
| 2861 | ut->c = -1; // Native index of second buffer contents |
| 2862 | |
| 2863 | // Initialize current chunk contents to be empty. |
| 2864 | // First access will fault something in. |
| 2865 | // Note: The initial nativeStart and chunkOffset must sum to zero |
| 2866 | // so that getNativeIndex() will correctly compute to zero |
| 2867 | // if no call to Access() has ever been made. They can't be both |
| 2868 | // zero without Access() thinking that the chunk is valid. |
| 2869 | ut->chunkContents = (UChar *)ut->p; |
| 2870 | ut->chunkNativeStart = -1; |
| 2871 | ut->chunkOffset = 1; |
| 2872 | ut->chunkNativeLimit = 0; |
| 2873 | ut->chunkLength = 0; |
| 2874 | ut->nativeIndexingLimit = ut->chunkOffset; // enables native indexing |
| 2875 | } |
| 2876 | return ut; |
| 2877 | } |