| 1 | /* |
| 2 | * Copyright (C) 2010-2019 Apple Inc. All rights reserved. |
| 3 | * Copyright (C) 2012 Google Inc. All rights reserved. |
| 4 | * |
| 5 | * Redistribution and use in source and binary forms, with or without |
| 6 | * modification, are permitted provided that the following conditions |
| 7 | * are met: |
| 8 | * 1. Redistributions of source code must retain the above copyright |
| 9 | * notice, this list of conditions and the following disclaimer. |
| 10 | * 2. Redistributions in binary form must reproduce the above copyright |
| 11 | * notice, this list of conditions and the following disclaimer in the |
| 12 | * documentation and/or other materials provided with the distribution. |
| 13 | * |
| 14 | * THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY |
| 15 | * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 16 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
| 17 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL APPLE INC. OR |
| 18 | * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, |
| 19 | * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
| 20 | * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR |
| 21 | * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY |
| 22 | * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 23 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| 24 | * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 25 | */ |
| 26 | |
| 27 | #include "config.h" |
| 28 | #include <wtf/text/StringBuilder.h> |
| 29 | |
| 30 | #include <wtf/dtoa.h> |
| 31 | #include <wtf/MathExtras.h> |
| 32 | |
| 33 | namespace WTF { |
| 34 | |
| 35 | static constexpr unsigned maxCapacity = String::MaxLength; |
| 36 | |
| 37 | static unsigned expandedCapacity(unsigned capacity, unsigned requiredLength) |
| 38 | { |
| 39 | static constexpr unsigned minimumCapacity = 16; |
| 40 | return std::max(requiredLength, std::max(minimumCapacity, std::min(capacity * 2, maxCapacity))); |
| 41 | } |
| 42 | |
| 43 | void StringBuilder::reifyString() const |
| 44 | { |
| 45 | ASSERT(!hasOverflowed()); |
| 46 | |
| 47 | // Check if the string already exists. |
| 48 | if (!m_string.isNull()) { |
| 49 | ASSERT(m_string.length() == m_length.unsafeGet<unsigned>()); |
| 50 | return; |
| 51 | } |
| 52 | |
| 53 | #if !ASSERT_DISABLED |
| 54 | m_isReified = true; |
| 55 | #endif |
| 56 | |
| 57 | // Check for empty. |
| 58 | if (!m_length) { |
| 59 | m_string = StringImpl::empty(); |
| 60 | return; |
| 61 | } |
| 62 | |
| 63 | // Must be valid in the buffer, take a substring (unless string fills the buffer). |
| 64 | ASSERT(m_buffer && m_length.unsafeGet<unsigned>() <= m_buffer->length()); |
| 65 | if (m_length.unsafeGet<unsigned>() == m_buffer->length()) |
| 66 | m_string = m_buffer.get(); |
| 67 | else |
| 68 | m_string = StringImpl::createSubstringSharingImpl(*m_buffer, 0, m_length.unsafeGet()); |
| 69 | } |
| 70 | |
| 71 | void StringBuilder::resize(unsigned newSize) |
| 72 | { |
| 73 | if (hasOverflowed()) |
| 74 | return; |
| 75 | |
| 76 | // Check newSize < m_length, hence m_length > 0. |
| 77 | unsigned oldLength = m_length.unsafeGet(); |
| 78 | ASSERT(newSize <= oldLength); |
| 79 | if (newSize == oldLength) |
| 80 | return; |
| 81 | ASSERT(oldLength); |
| 82 | |
| 83 | m_length = newSize; |
| 84 | ASSERT(!hasOverflowed()); |
| 85 | |
| 86 | // If there is a buffer, we only need to duplicate it if it has more than one ref. |
| 87 | if (m_buffer) { |
| 88 | m_string = String(); // Clear the string to remove the reference to m_buffer if any before checking the reference count of m_buffer. |
| 89 | if (!m_buffer->hasOneRef()) { |
| 90 | if (m_buffer->is8Bit()) |
| 91 | allocateBuffer(m_buffer->characters8(), m_buffer->length()); |
| 92 | else |
| 93 | allocateBuffer(m_buffer->characters16(), m_buffer->length()); |
| 94 | } |
| 95 | ASSERT(hasOverflowed() || m_buffer->length() >= m_length.unsafeGet<unsigned>()); |
| 96 | return; |
| 97 | } |
| 98 | |
| 99 | // Since m_length && !m_buffer, the string must be valid in m_string, and m_string.length() > 0. |
| 100 | ASSERT(!m_string.isEmpty()); |
| 101 | ASSERT(oldLength == m_string.length()); |
| 102 | ASSERT(newSize < m_string.length()); |
| 103 | m_string = StringImpl::createSubstringSharingImpl(*m_string.impl(), 0, newSize); |
| 104 | } |
| 105 | |
| 106 | // Allocate a new 8 bit buffer, copying in currentCharacters (these may come from either m_string |
| 107 | // or m_buffer, neither will be reassigned until the copy has completed). |
| 108 | void StringBuilder::allocateBuffer(const LChar* currentCharacters, unsigned requiredLength) |
| 109 | { |
| 110 | ASSERT(!hasOverflowed()); |
| 111 | ASSERT(m_is8Bit); |
| 112 | // Copy the existing data into a new buffer, set result to point to the end of the existing data. |
| 113 | auto buffer = StringImpl::tryCreateUninitialized(requiredLength, m_bufferCharacters8); |
| 114 | if (UNLIKELY(!buffer)) |
| 115 | return didOverflow(); |
| 116 | std::memcpy(m_bufferCharacters8, currentCharacters, m_length.unsafeGet()); |
| 117 | |
| 118 | // Update the builder state. |
| 119 | m_buffer = WTFMove(buffer); |
| 120 | m_string = String(); |
| 121 | ASSERT(m_buffer->length() == requiredLength); |
| 122 | } |
| 123 | |
| 124 | // Allocate a new 16 bit buffer, copying in currentCharacters (these may come from either m_string |
| 125 | // or m_buffer, neither will be reassigned until the copy has completed). |
| 126 | void StringBuilder::allocateBuffer(const UChar* currentCharacters, unsigned requiredLength) |
| 127 | { |
| 128 | ASSERT(!hasOverflowed()); |
| 129 | ASSERT(!m_is8Bit); |
| 130 | // Copy the existing data into a new buffer, set result to point to the end of the existing data. |
| 131 | auto buffer = StringImpl::tryCreateUninitialized(requiredLength, m_bufferCharacters16); |
| 132 | if (UNLIKELY(!buffer)) |
| 133 | return didOverflow(); |
| 134 | std::memcpy(m_bufferCharacters16, currentCharacters, static_cast<size_t>(m_length.unsafeGet()) * sizeof(UChar)); // This can't overflow. |
| 135 | |
| 136 | // Update the builder state. |
| 137 | m_buffer = WTFMove(buffer); |
| 138 | m_string = String(); |
| 139 | ASSERT(m_buffer->length() == requiredLength); |
| 140 | } |
| 141 | |
| 142 | // Allocate a new 16 bit buffer, copying in currentCharacters (which is 8 bit and may come |
| 143 | // from either m_string or m_buffer, neither will be reassigned until the copy has completed). |
| 144 | void StringBuilder::allocateBufferUpConvert(const LChar* currentCharacters, unsigned requiredLength) |
| 145 | { |
| 146 | ASSERT(!hasOverflowed()); |
| 147 | ASSERT(m_is8Bit); |
| 148 | unsigned length = m_length.unsafeGet(); |
| 149 | ASSERT(requiredLength <= maxCapacity && requiredLength >= length); |
| 150 | // Copy the existing data into a new buffer, set result to point to the end of the existing data. |
| 151 | auto buffer = StringImpl::tryCreateUninitialized(requiredLength, m_bufferCharacters16); |
| 152 | if (UNLIKELY(!buffer)) |
| 153 | return didOverflow(); // Treat a failure to allcoate as an overflow. |
| 154 | for (unsigned i = 0; i < length; ++i) |
| 155 | m_bufferCharacters16[i] = currentCharacters[i]; |
| 156 | |
| 157 | m_is8Bit = false; |
| 158 | |
| 159 | // Update the builder state. |
| 160 | m_buffer = WTFMove(buffer); |
| 161 | m_string = String(); |
| 162 | ASSERT(m_buffer->length() == requiredLength); |
| 163 | } |
| 164 | |
| 165 | template<> |
| 166 | void StringBuilder::reallocateBuffer<LChar>(unsigned requiredLength) |
| 167 | { |
| 168 | // If the buffer has only one ref (by this StringBuilder), reallocate it, |
| 169 | // otherwise fall back to "allocate and copy" method. |
| 170 | m_string = String(); |
| 171 | |
| 172 | ASSERT(m_is8Bit); |
| 173 | ASSERT(m_buffer->is8Bit()); |
| 174 | |
| 175 | if (m_buffer->hasOneRef()) { |
| 176 | auto expectedStringImpl = StringImpl::tryReallocate(m_buffer.releaseNonNull(), requiredLength, m_bufferCharacters8); |
| 177 | if (UNLIKELY(!expectedStringImpl)) |
| 178 | return didOverflow(); |
| 179 | m_buffer = WTFMove(expectedStringImpl.value()); |
| 180 | } else |
| 181 | allocateBuffer(m_buffer->characters8(), requiredLength); |
| 182 | ASSERT(hasOverflowed() || m_buffer->length() == requiredLength); |
| 183 | } |
| 184 | |
| 185 | template<> |
| 186 | void StringBuilder::reallocateBuffer<UChar>(unsigned requiredLength) |
| 187 | { |
| 188 | // If the buffer has only one ref (by this StringBuilder), reallocate it, |
| 189 | // otherwise fall back to "allocate and copy" method. |
| 190 | m_string = String(); |
| 191 | |
| 192 | if (m_buffer->is8Bit()) |
| 193 | allocateBufferUpConvert(m_buffer->characters8(), requiredLength); |
| 194 | else if (m_buffer->hasOneRef()) { |
| 195 | auto expectedStringImpl = StringImpl::tryReallocate(m_buffer.releaseNonNull(), requiredLength, m_bufferCharacters16); |
| 196 | if (UNLIKELY(!expectedStringImpl)) |
| 197 | return didOverflow(); |
| 198 | m_buffer = WTFMove(expectedStringImpl.value()); |
| 199 | } else |
| 200 | allocateBuffer(m_buffer->characters16(), requiredLength); |
| 201 | ASSERT(hasOverflowed() || m_buffer->length() == requiredLength); |
| 202 | } |
| 203 | |
| 204 | void StringBuilder::reserveCapacity(unsigned newCapacity) |
| 205 | { |
| 206 | if (hasOverflowed()) |
| 207 | return; |
| 208 | ASSERT(newCapacity <= String::MaxLength); |
| 209 | if (m_buffer) { |
| 210 | // If there is already a buffer, then grow if necessary. |
| 211 | if (newCapacity > m_buffer->length()) { |
| 212 | if (m_buffer->is8Bit()) |
| 213 | reallocateBuffer<LChar>(newCapacity); |
| 214 | else |
| 215 | reallocateBuffer<UChar>(newCapacity); |
| 216 | } |
| 217 | } else { |
| 218 | // Grow the string, if necessary. |
| 219 | unsigned length = m_length.unsafeGet(); |
| 220 | if (newCapacity > length) { |
| 221 | if (!length) { |
| 222 | LChar* nullPlaceholder = nullptr; |
| 223 | allocateBuffer(nullPlaceholder, newCapacity); |
| 224 | } else if (m_string.is8Bit()) |
| 225 | allocateBuffer(m_string.characters8(), newCapacity); |
| 226 | else |
| 227 | allocateBuffer(m_string.characters16(), newCapacity); |
| 228 | } |
| 229 | } |
| 230 | ASSERT(hasOverflowed() || !newCapacity || m_buffer->length() >= newCapacity); |
| 231 | } |
| 232 | |
| 233 | // Make 'additionalLength' additional capacity be available in m_buffer, update m_string & m_length, |
| 234 | // return a pointer to the newly allocated storage. |
| 235 | // Returns nullptr if the size of the new builder would have overflowed |
| 236 | template<typename CharacterType> ALWAYS_INLINE CharacterType* StringBuilder::extendBufferForAppending(unsigned additionalLength) |
| 237 | { |
| 238 | ASSERT(additionalLength); |
| 239 | |
| 240 | // Calculate the new size of the builder after appending. |
| 241 | CheckedInt32 requiredLength = m_length + additionalLength; |
| 242 | if (requiredLength.hasOverflowed()) { |
| 243 | didOverflow(); |
| 244 | return nullptr; |
| 245 | } |
| 246 | |
| 247 | return extendBufferForAppendingWithoutOverflowCheck<CharacterType>(requiredLength); |
| 248 | } |
| 249 | |
| 250 | template<typename CharacterType> ALWAYS_INLINE CharacterType* StringBuilder::extendBufferForAppendingWithoutOverflowCheck(CheckedInt32 requiredLength) |
| 251 | { |
| 252 | ASSERT(!requiredLength.hasOverflowed()); |
| 253 | |
| 254 | if (m_buffer && (requiredLength.unsafeGet<unsigned>() <= m_buffer->length())) { |
| 255 | // If the buffer is valid it must be at least as long as the current builder contents! |
| 256 | ASSERT(m_buffer->length() >= m_length.unsafeGet<unsigned>()); |
| 257 | unsigned currentLength = m_length.unsafeGet(); |
| 258 | m_string = String(); |
| 259 | m_length = requiredLength; |
| 260 | return getBufferCharacters<CharacterType>() + currentLength; |
| 261 | } |
| 262 | |
| 263 | return extendBufferForAppendingSlowCase<CharacterType>(requiredLength.unsafeGet()); |
| 264 | } |
| 265 | |
| 266 | LChar* StringBuilder::extendBufferForAppending8(CheckedInt32 requiredLength) |
| 267 | { |
| 268 | if (UNLIKELY(requiredLength.hasOverflowed())) { |
| 269 | didOverflow(); |
| 270 | return nullptr; |
| 271 | } |
| 272 | return extendBufferForAppendingWithoutOverflowCheck<LChar>(requiredLength); |
| 273 | } |
| 274 | |
| 275 | UChar* StringBuilder::extendBufferForAppending16(CheckedInt32 requiredLength) |
| 276 | { |
| 277 | if (UNLIKELY(requiredLength.hasOverflowed())) { |
| 278 | didOverflow(); |
| 279 | return nullptr; |
| 280 | } |
| 281 | if (m_is8Bit) { |
| 282 | const LChar* characters; |
| 283 | if (m_buffer) { |
| 284 | ASSERT(m_buffer->length() >= m_length.unsafeGet<unsigned>()); |
| 285 | characters = m_buffer->characters8(); |
| 286 | } else { |
| 287 | ASSERT(m_string.length() == m_length.unsafeGet<unsigned>()); |
| 288 | characters = m_string.isNull() ? nullptr : m_string.characters8(); |
| 289 | } |
| 290 | allocateBufferUpConvert(characters, expandedCapacity(capacity(), requiredLength.unsafeGet())); |
| 291 | if (UNLIKELY(hasOverflowed())) |
| 292 | return nullptr; |
| 293 | unsigned oldLength = m_length.unsafeGet(); |
| 294 | m_length = requiredLength.unsafeGet(); |
| 295 | return m_bufferCharacters16 + oldLength; |
| 296 | } |
| 297 | return extendBufferForAppendingWithoutOverflowCheck<UChar>(requiredLength); |
| 298 | } |
| 299 | |
| 300 | // Make 'requiredLength' capacity be available in m_buffer, update m_string & m_length, |
| 301 | // return a pointer to the newly allocated storage. |
| 302 | template<typename CharacterType> CharacterType* StringBuilder::extendBufferForAppendingSlowCase(unsigned requiredLength) |
| 303 | { |
| 304 | ASSERT(!hasOverflowed()); |
| 305 | ASSERT(requiredLength); |
| 306 | |
| 307 | if (m_buffer) { |
| 308 | // If the buffer is valid it must be at least as long as the current builder contents! |
| 309 | ASSERT(m_buffer->length() >= m_length.unsafeGet<unsigned>()); |
| 310 | |
| 311 | reallocateBuffer<CharacterType>(expandedCapacity(capacity(), requiredLength)); |
| 312 | } else { |
| 313 | ASSERT(m_string.length() == m_length.unsafeGet<unsigned>()); |
| 314 | allocateBuffer(m_length ? m_string.characters<CharacterType>() : nullptr, expandedCapacity(capacity(), requiredLength)); |
| 315 | } |
| 316 | if (UNLIKELY(hasOverflowed())) |
| 317 | return nullptr; |
| 318 | |
| 319 | CharacterType* result = getBufferCharacters<CharacterType>() + m_length.unsafeGet(); |
| 320 | m_length = requiredLength; |
| 321 | ASSERT(!hasOverflowed()); |
| 322 | ASSERT(m_buffer->length() >= m_length.unsafeGet<unsigned>()); |
| 323 | return result; |
| 324 | } |
| 325 | |
| 326 | void StringBuilder::appendCharacters(const UChar* characters, unsigned length) |
| 327 | { |
| 328 | if (!length || hasOverflowed()) |
| 329 | return; |
| 330 | |
| 331 | ASSERT(characters); |
| 332 | |
| 333 | if (m_is8Bit && length == 1 && isLatin1(characters[0])) { |
| 334 | append(static_cast<LChar>(characters[0])); |
| 335 | return; |
| 336 | } |
| 337 | |
| 338 | // FIXME: Should we optimize memory by keeping the string 8-bit when all the characters are Latin-1? |
| 339 | |
| 340 | UChar* destination = extendBufferForAppending16(m_length + length); |
| 341 | if (UNLIKELY(!destination)) |
| 342 | return; |
| 343 | std::memcpy(destination, characters, static_cast<size_t>(length) * sizeof(UChar)); |
| 344 | ASSERT(!hasOverflowed()); |
| 345 | ASSERT(m_buffer->length() >= m_length.unsafeGet<unsigned>()); |
| 346 | } |
| 347 | |
| 348 | void StringBuilder::appendCharacters(const LChar* characters, unsigned length) |
| 349 | { |
| 350 | if (!length || hasOverflowed()) |
| 351 | return; |
| 352 | |
| 353 | ASSERT(characters); |
| 354 | |
| 355 | if (m_is8Bit) { |
| 356 | LChar* destination = extendBufferForAppending<LChar>(length); |
| 357 | if (!destination) { |
| 358 | ASSERT(hasOverflowed()); |
| 359 | return; |
| 360 | } |
| 361 | if (length > 8) |
| 362 | std::memcpy(destination, characters, length); |
| 363 | else { |
| 364 | // FIXME: How strong is our evidence that this is faster than memcpy? What platforms is this true for? |
| 365 | const LChar* end = characters + length; |
| 366 | while (characters < end) |
| 367 | *destination++ = *characters++; |
| 368 | } |
| 369 | } else { |
| 370 | UChar* destination = extendBufferForAppending<UChar>(length); |
| 371 | if (!destination) { |
| 372 | ASSERT(hasOverflowed()); |
| 373 | return; |
| 374 | } |
| 375 | const LChar* end = characters + length; |
| 376 | while (characters < end) |
| 377 | *destination++ = *characters++; |
| 378 | } |
| 379 | } |
| 380 | |
| 381 | #if USE(CF) |
| 382 | |
| 383 | void StringBuilder::append(CFStringRef string) |
| 384 | { |
| 385 | // Fast path: avoid constructing a temporary String when possible. |
| 386 | if (auto* characters = CFStringGetCStringPtr(string, kCFStringEncodingISOLatin1)) { |
| 387 | appendCharacters(reinterpret_cast<const LChar*>(characters), CFStringGetLength(string)); |
| 388 | return; |
| 389 | } |
| 390 | append(String(string)); |
| 391 | } |
| 392 | |
| 393 | #endif |
| 394 | |
| 395 | void StringBuilder::appendNumber(int number) |
| 396 | { |
| 397 | numberToStringSigned<StringBuilder>(number, this); |
| 398 | } |
| 399 | |
| 400 | void StringBuilder::appendNumber(unsigned number) |
| 401 | { |
| 402 | numberToStringUnsigned<StringBuilder>(number, this); |
| 403 | } |
| 404 | |
| 405 | void StringBuilder::appendNumber(long number) |
| 406 | { |
| 407 | numberToStringSigned<StringBuilder>(number, this); |
| 408 | } |
| 409 | |
| 410 | void StringBuilder::appendNumber(unsigned long number) |
| 411 | { |
| 412 | numberToStringUnsigned<StringBuilder>(number, this); |
| 413 | } |
| 414 | |
| 415 | void StringBuilder::appendNumber(long long number) |
| 416 | { |
| 417 | numberToStringSigned<StringBuilder>(number, this); |
| 418 | } |
| 419 | |
| 420 | void StringBuilder::appendNumber(unsigned long long number) |
| 421 | { |
| 422 | numberToStringUnsigned<StringBuilder>(number, this); |
| 423 | } |
| 424 | |
| 425 | void StringBuilder::appendFixedPrecisionNumber(float number, unsigned precision, TrailingZerosTruncatingPolicy policy) |
| 426 | { |
| 427 | NumberToStringBuffer buffer; |
| 428 | append(numberToFixedPrecisionString(number, precision, buffer, policy == TruncateTrailingZeros)); |
| 429 | } |
| 430 | |
| 431 | void StringBuilder::appendFixedPrecisionNumber(double number, unsigned precision, TrailingZerosTruncatingPolicy policy) |
| 432 | { |
| 433 | NumberToStringBuffer buffer; |
| 434 | append(numberToFixedPrecisionString(number, precision, buffer, policy == TruncateTrailingZeros)); |
| 435 | } |
| 436 | |
| 437 | void StringBuilder::appendNumber(float number) |
| 438 | { |
| 439 | NumberToStringBuffer buffer; |
| 440 | append(numberToString(number, buffer)); |
| 441 | } |
| 442 | |
| 443 | void StringBuilder::appendNumber(double number) |
| 444 | { |
| 445 | NumberToStringBuffer buffer; |
| 446 | append(numberToString(number, buffer)); |
| 447 | } |
| 448 | |
| 449 | void StringBuilder::appendFixedWidthNumber(float number, unsigned decimalPlaces) |
| 450 | { |
| 451 | NumberToStringBuffer buffer; |
| 452 | append(numberToFixedWidthString(number, decimalPlaces, buffer)); |
| 453 | } |
| 454 | |
| 455 | void StringBuilder::appendFixedWidthNumber(double number, unsigned decimalPlaces) |
| 456 | { |
| 457 | NumberToStringBuffer buffer; |
| 458 | append(numberToFixedWidthString(number, decimalPlaces, buffer)); |
| 459 | } |
| 460 | |
| 461 | bool StringBuilder::canShrink() const |
| 462 | { |
| 463 | if (hasOverflowed()) |
| 464 | return false; |
| 465 | // Only shrink the buffer if it's less than 80% full. |
| 466 | // FIXME: We should tune this heuristic based some actual test case measurements. |
| 467 | unsigned length = m_length.unsafeGet(); |
| 468 | return m_buffer && m_buffer->length() > (length + (length >> 2)); |
| 469 | } |
| 470 | |
| 471 | void StringBuilder::shrinkToFit() |
| 472 | { |
| 473 | if (canShrink()) { |
| 474 | if (m_is8Bit) |
| 475 | reallocateBuffer<LChar>(m_length.unsafeGet()); |
| 476 | else |
| 477 | reallocateBuffer<UChar>(m_length.unsafeGet()); |
| 478 | ASSERT(!hasOverflowed()); |
| 479 | m_string = WTFMove(m_buffer); |
| 480 | } |
| 481 | } |
| 482 | |
| 483 | } // namespace WTF |
| 484 | |