| 1 | /* |
| 2 | * Copyright (C) 1999 Lars Knoll (knoll@kde.org) |
| 3 | * Copyright (C) 2005-2018 Apple Inc. All rights reserved. |
| 4 | * Copyright (C) 2009 Google Inc. All rights reserved. |
| 5 | * |
| 6 | * This library is free software; you can redistribute it and/or |
| 7 | * modify it under the terms of the GNU Library General Public |
| 8 | * License as published by the Free Software Foundation; either |
| 9 | * version 2 of the License, or (at your option) any later version. |
| 10 | * |
| 11 | * This library is distributed in the hope that it will be useful, |
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 14 | * Library General Public License for more details. |
| 15 | * |
| 16 | * You should have received a copy of the GNU Library General Public License |
| 17 | * along with this library; see the file COPYING.LIB. If not, write to |
| 18 | * the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
| 19 | * Boston, MA 02110-1301, USA. |
| 20 | * |
| 21 | */ |
| 22 | |
| 23 | #pragma once |
| 24 | |
| 25 | #include <limits.h> |
| 26 | #include <unicode/ustring.h> |
| 27 | #include <wtf/ASCIICType.h> |
| 28 | #include <wtf/CheckedArithmetic.h> |
| 29 | #include <wtf/Expected.h> |
| 30 | #include <wtf/MathExtras.h> |
| 31 | #include <wtf/StdLibExtras.h> |
| 32 | #include <wtf/Vector.h> |
| 33 | #include <wtf/text/ASCIIFastPath.h> |
| 34 | #include <wtf/text/ConversionMode.h> |
| 35 | #include <wtf/text/StringCommon.h> |
| 36 | #include <wtf/text/StringHasher.h> |
| 37 | #include <wtf/text/UTF8ConversionError.h> |
| 38 | |
| 39 | #if USE(CF) |
| 40 | typedef const struct __CFString * CFStringRef; |
| 41 | #endif |
| 42 | |
| 43 | #ifdef __OBJC__ |
| 44 | @class NSString; |
| 45 | #endif |
| 46 | |
| 47 | namespace JSC { |
| 48 | namespace LLInt { class Data; } |
| 49 | class ; |
| 50 | } |
| 51 | |
| 52 | namespace WTF { |
| 53 | |
| 54 | class SymbolImpl; |
| 55 | class SymbolRegistry; |
| 56 | |
| 57 | struct CStringTranslator; |
| 58 | struct HashAndUTF8CharactersTranslator; |
| 59 | struct LCharBufferTranslator; |
| 60 | struct StringHash; |
| 61 | struct SubstringTranslator; |
| 62 | struct UCharBufferTranslator; |
| 63 | |
| 64 | template<typename> class RetainPtr; |
| 65 | |
| 66 | template<typename> struct BufferFromStaticDataTranslator; |
| 67 | template<typename> struct HashAndCharactersTranslator; |
| 68 | |
| 69 | // Define STRING_STATS to 1 turn on runtime statistics of string sizes and memory usage. |
| 70 | #define STRING_STATS 0 |
| 71 | |
| 72 | template<bool isSpecialCharacter(UChar), typename CharacterType> bool isAllSpecialCharacters(const CharacterType*, size_t length); |
| 73 | |
| 74 | #if STRING_STATS |
| 75 | |
| 76 | struct StringStats { |
| 77 | void add8BitString(unsigned length, bool isSubString = false) |
| 78 | { |
| 79 | ++m_totalNumberStrings; |
| 80 | ++m_number8BitStrings; |
| 81 | if (!isSubString) |
| 82 | m_total8BitData += length; |
| 83 | } |
| 84 | |
| 85 | void add16BitString(unsigned length, bool isSubString = false) |
| 86 | { |
| 87 | ++m_totalNumberStrings; |
| 88 | ++m_number16BitStrings; |
| 89 | if (!isSubString) |
| 90 | m_total16BitData += length; |
| 91 | } |
| 92 | |
| 93 | void removeString(StringImpl&); |
| 94 | void printStats(); |
| 95 | |
| 96 | static const unsigned s_printStringStatsFrequency = 5000; |
| 97 | static std::atomic<unsigned> s_stringRemovesTillPrintStats; |
| 98 | |
| 99 | std::atomic<unsigned> m_refCalls; |
| 100 | std::atomic<unsigned> m_derefCalls; |
| 101 | |
| 102 | std::atomic<unsigned> m_totalNumberStrings; |
| 103 | std::atomic<unsigned> m_number8BitStrings; |
| 104 | std::atomic<unsigned> m_number16BitStrings; |
| 105 | std::atomic<unsigned long long> m_total8BitData; |
| 106 | std::atomic<unsigned long long> m_total16BitData; |
| 107 | }; |
| 108 | |
| 109 | #define STRING_STATS_ADD_8BIT_STRING(length) StringImpl::stringStats().add8BitString(length) |
| 110 | #define STRING_STATS_ADD_8BIT_STRING2(length, isSubString) StringImpl::stringStats().add8BitString(length, isSubString) |
| 111 | #define STRING_STATS_ADD_16BIT_STRING(length) StringImpl::stringStats().add16BitString(length) |
| 112 | #define STRING_STATS_ADD_16BIT_STRING2(length, isSubString) StringImpl::stringStats().add16BitString(length, isSubString) |
| 113 | #define STRING_STATS_REMOVE_STRING(string) StringImpl::stringStats().removeString(string) |
| 114 | #define STRING_STATS_REF_STRING(string) ++StringImpl::stringStats().m_refCalls; |
| 115 | #define STRING_STATS_DEREF_STRING(string) ++StringImpl::stringStats().m_derefCalls; |
| 116 | |
| 117 | #else |
| 118 | |
| 119 | #define STRING_STATS_ADD_8BIT_STRING(length) ((void)0) |
| 120 | #define STRING_STATS_ADD_8BIT_STRING2(length, isSubString) ((void)0) |
| 121 | #define STRING_STATS_ADD_16BIT_STRING(length) ((void)0) |
| 122 | #define STRING_STATS_ADD_16BIT_STRING2(length, isSubString) ((void)0) |
| 123 | #define STRING_STATS_ADD_UPCONVERTED_STRING(length) ((void)0) |
| 124 | #define STRING_STATS_REMOVE_STRING(string) ((void)0) |
| 125 | #define STRING_STATS_REF_STRING(string) ((void)0) |
| 126 | #define STRING_STATS_DEREF_STRING(string) ((void)0) |
| 127 | |
| 128 | #endif |
| 129 | |
| 130 | template<typename CharacterType> inline bool isLatin1(CharacterType character) |
| 131 | { |
| 132 | using UnsignedCharacterType = typename std::make_unsigned<CharacterType>::type; |
| 133 | return static_cast<UnsignedCharacterType>(character) <= static_cast<UnsignedCharacterType>(0xFF); |
| 134 | } |
| 135 | |
| 136 | class StringImplShape { |
| 137 | WTF_MAKE_NONCOPYABLE(StringImplShape); |
| 138 | public: |
| 139 | static constexpr unsigned MaxLength = std::numeric_limits<int32_t>::max(); |
| 140 | |
| 141 | protected: |
| 142 | StringImplShape(unsigned refCount, unsigned length, const LChar*, unsigned hashAndFlags); |
| 143 | StringImplShape(unsigned refCount, unsigned length, const UChar*, unsigned hashAndFlags); |
| 144 | |
| 145 | enum ConstructWithConstExprTag { ConstructWithConstExpr }; |
| 146 | template<unsigned characterCount> constexpr StringImplShape(unsigned refCount, unsigned length, const char (&characters)[characterCount], unsigned hashAndFlags, ConstructWithConstExprTag); |
| 147 | template<unsigned characterCount> constexpr StringImplShape(unsigned refCount, unsigned length, const char16_t (&characters)[characterCount], unsigned hashAndFlags, ConstructWithConstExprTag); |
| 148 | |
| 149 | unsigned m_refCount; |
| 150 | unsigned m_length; |
| 151 | union { |
| 152 | const LChar* m_data8; |
| 153 | const UChar* m_data16; |
| 154 | // It seems that reinterpret_cast prevents constexpr's compile time initialization in VC++. |
| 155 | // These are needed to avoid reinterpret_cast. |
| 156 | const char* m_data8Char; |
| 157 | const char16_t* m_data16Char; |
| 158 | }; |
| 159 | mutable unsigned m_hashAndFlags; |
| 160 | }; |
| 161 | |
| 162 | // FIXME: Use of StringImpl and const is rather confused. |
| 163 | // The actual string inside a StringImpl is immutable, so you can't modify a string using a StringImpl&. |
| 164 | // We could mark every member function const and always use "const StringImpl&" and "const StringImpl*". |
| 165 | // Or we could say that "const" doesn't make sense at all and use "StringImpl&" and "StringImpl*" everywhere. |
| 166 | // Right now we use a mix of both, which makes code more confusing and has no benefit. |
| 167 | |
| 168 | class StringImpl : private StringImplShape { |
| 169 | WTF_MAKE_NONCOPYABLE(StringImpl); WTF_MAKE_FAST_ALLOCATED; |
| 170 | |
| 171 | friend class AtomicStringImpl; |
| 172 | friend class JSC::LLInt::Data; |
| 173 | friend class JSC::LLIntOffsetsExtractor; |
| 174 | friend class PrivateSymbolImpl; |
| 175 | friend class RegisteredSymbolImpl; |
| 176 | friend class SymbolImpl; |
| 177 | friend class ExternalStringImpl; |
| 178 | |
| 179 | friend struct WTF::CStringTranslator; |
| 180 | friend struct WTF::HashAndUTF8CharactersTranslator; |
| 181 | friend struct WTF::LCharBufferTranslator; |
| 182 | friend struct WTF::SubstringTranslator; |
| 183 | friend struct WTF::UCharBufferTranslator; |
| 184 | |
| 185 | template<typename> friend struct WTF::BufferFromStaticDataTranslator; |
| 186 | template<typename> friend struct WTF::HashAndCharactersTranslator; |
| 187 | |
| 188 | public: |
| 189 | enum BufferOwnership { BufferInternal, BufferOwned, BufferSubstring, BufferExternal }; |
| 190 | |
| 191 | static constexpr unsigned MaxLength = StringImplShape::MaxLength; |
| 192 | |
| 193 | // The bottom 6 bits in the hash are flags. |
| 194 | static constexpr const unsigned s_flagCount = 6; |
| 195 | private: |
| 196 | static constexpr const unsigned s_flagMask = (1u << s_flagCount) - 1; |
| 197 | static_assert(s_flagCount <= StringHasher::flagCount, "StringHasher reserves enough bits for StringImpl flags" ); |
| 198 | static constexpr const unsigned s_flagStringKindCount = 4; |
| 199 | |
| 200 | static constexpr const unsigned s_hashFlagStringKindIsAtomic = 1u << (s_flagStringKindCount); |
| 201 | static constexpr const unsigned s_hashFlagStringKindIsSymbol = 1u << (s_flagStringKindCount + 1); |
| 202 | static constexpr const unsigned s_hashMaskStringKind = s_hashFlagStringKindIsAtomic | s_hashFlagStringKindIsSymbol; |
| 203 | static constexpr const unsigned s_hashFlagDidReportCost = 1u << 3; |
| 204 | static constexpr const unsigned s_hashFlag8BitBuffer = 1u << 2; |
| 205 | static constexpr const unsigned s_hashMaskBufferOwnership = (1u << 0) | (1u << 1); |
| 206 | |
| 207 | enum StringKind { |
| 208 | StringNormal = 0u, // non-symbol, non-atomic |
| 209 | StringAtomic = s_hashFlagStringKindIsAtomic, // non-symbol, atomic |
| 210 | StringSymbol = s_hashFlagStringKindIsSymbol, // symbol, non-atomic |
| 211 | }; |
| 212 | |
| 213 | // Create a normal 8-bit string with internal storage (BufferInternal). |
| 214 | enum Force8Bit { Force8BitConstructor }; |
| 215 | StringImpl(unsigned length, Force8Bit); |
| 216 | |
| 217 | // Create a normal 16-bit string with internal storage (BufferInternal). |
| 218 | explicit StringImpl(unsigned length); |
| 219 | |
| 220 | // Create a StringImpl adopting ownership of the provided buffer (BufferOwned). |
| 221 | StringImpl(MallocPtr<LChar>, unsigned length); |
| 222 | StringImpl(MallocPtr<UChar>, unsigned length); |
| 223 | enum ConstructWithoutCopyingTag { ConstructWithoutCopying }; |
| 224 | StringImpl(const UChar*, unsigned length, ConstructWithoutCopyingTag); |
| 225 | StringImpl(const LChar*, unsigned length, ConstructWithoutCopyingTag); |
| 226 | |
| 227 | // Used to create new strings that are a substring of an existing StringImpl (BufferSubstring). |
| 228 | StringImpl(const LChar*, unsigned length, Ref<StringImpl>&&); |
| 229 | StringImpl(const UChar*, unsigned length, Ref<StringImpl>&&); |
| 230 | |
| 231 | public: |
| 232 | WTF_EXPORT_PRIVATE static void destroy(StringImpl*); |
| 233 | |
| 234 | WTF_EXPORT_PRIVATE static Ref<StringImpl> create(const UChar*, unsigned length); |
| 235 | WTF_EXPORT_PRIVATE static Ref<StringImpl> create(const LChar*, unsigned length); |
| 236 | WTF_EXPORT_PRIVATE static Ref<StringImpl> create8BitIfPossible(const UChar*, unsigned length); |
| 237 | template<size_t inlineCapacity> static Ref<StringImpl> create8BitIfPossible(const Vector<UChar, inlineCapacity>&); |
| 238 | WTF_EXPORT_PRIVATE static Ref<StringImpl> create8BitIfPossible(const UChar*); |
| 239 | |
| 240 | ALWAYS_INLINE static Ref<StringImpl> create(const char* characters, unsigned length) { return create(reinterpret_cast<const LChar*>(characters), length); } |
| 241 | WTF_EXPORT_PRIVATE static Ref<StringImpl> create(const LChar*); |
| 242 | ALWAYS_INLINE static Ref<StringImpl> create(const char* string) { return create(reinterpret_cast<const LChar*>(string)); } |
| 243 | |
| 244 | static Ref<StringImpl> createSubstringSharingImpl(StringImpl&, unsigned offset, unsigned length); |
| 245 | |
| 246 | template<unsigned characterCount> static Ref<StringImpl> createFromLiteral(const char (&)[characterCount]); |
| 247 | |
| 248 | // FIXME: Replace calls to these overloads of createFromLiteral to createWithoutCopying instead. |
| 249 | WTF_EXPORT_PRIVATE static Ref<StringImpl> createFromLiteral(const char*, unsigned length); |
| 250 | WTF_EXPORT_PRIVATE static Ref<StringImpl> createFromLiteral(const char*); |
| 251 | |
| 252 | WTF_EXPORT_PRIVATE static Ref<StringImpl> createWithoutCopying(const UChar*, unsigned length); |
| 253 | WTF_EXPORT_PRIVATE static Ref<StringImpl> createWithoutCopying(const LChar*, unsigned length); |
| 254 | WTF_EXPORT_PRIVATE static Ref<StringImpl> createUninitialized(unsigned length, LChar*&); |
| 255 | WTF_EXPORT_PRIVATE static Ref<StringImpl> createUninitialized(unsigned length, UChar*&); |
| 256 | template<typename CharacterType> static RefPtr<StringImpl> tryCreateUninitialized(unsigned length, CharacterType*&); |
| 257 | |
| 258 | // Reallocate the StringImpl. The originalString must be only owned by the Ref, |
| 259 | // and the buffer ownership must be BufferInternal. Just like the input pointer of realloc(), |
| 260 | // the originalString can't be used after this function. |
| 261 | static Ref<StringImpl> reallocate(Ref<StringImpl>&& originalString, unsigned length, LChar*& data); |
| 262 | static Ref<StringImpl> reallocate(Ref<StringImpl>&& originalString, unsigned length, UChar*& data); |
| 263 | static Expected<Ref<StringImpl>, UTF8ConversionError> tryReallocate(Ref<StringImpl>&& originalString, unsigned length, LChar*& data); |
| 264 | static Expected<Ref<StringImpl>, UTF8ConversionError> tryReallocate(Ref<StringImpl>&& originalString, unsigned length, UChar*& data); |
| 265 | |
| 266 | static unsigned flagsOffset() { return OBJECT_OFFSETOF(StringImpl, m_hashAndFlags); } |
| 267 | static constexpr unsigned flagIs8Bit() { return s_hashFlag8BitBuffer; } |
| 268 | static constexpr unsigned flagIsAtomic() { return s_hashFlagStringKindIsAtomic; } |
| 269 | static constexpr unsigned flagIsSymbol() { return s_hashFlagStringKindIsSymbol; } |
| 270 | static constexpr unsigned maskStringKind() { return s_hashMaskStringKind; } |
| 271 | static unsigned dataOffset() { return OBJECT_OFFSETOF(StringImpl, m_data8); } |
| 272 | |
| 273 | template<typename CharacterType, size_t inlineCapacity, typename OverflowHandler, size_t minCapacity> |
| 274 | static Ref<StringImpl> adopt(Vector<CharacterType, inlineCapacity, OverflowHandler, minCapacity>&&); |
| 275 | |
| 276 | WTF_EXPORT_PRIVATE static Ref<StringImpl> adopt(StringBuffer<UChar>&&); |
| 277 | WTF_EXPORT_PRIVATE static Ref<StringImpl> adopt(StringBuffer<LChar>&&); |
| 278 | |
| 279 | unsigned length() const { return m_length; } |
| 280 | static ptrdiff_t lengthMemoryOffset() { return OBJECT_OFFSETOF(StringImpl, m_length); } |
| 281 | bool isEmpty() const { return !m_length; } |
| 282 | |
| 283 | bool is8Bit() const { return m_hashAndFlags & s_hashFlag8BitBuffer; } |
| 284 | ALWAYS_INLINE const LChar* characters8() const { ASSERT(is8Bit()); return m_data8; } |
| 285 | ALWAYS_INLINE const UChar* characters16() const { ASSERT(!is8Bit()); return m_data16; } |
| 286 | |
| 287 | template<typename CharacterType> const CharacterType* characters() const; |
| 288 | |
| 289 | size_t cost() const; |
| 290 | size_t costDuringGC(); |
| 291 | |
| 292 | WTF_EXPORT_PRIVATE size_t sizeInBytes() const; |
| 293 | |
| 294 | bool isSymbol() const { return m_hashAndFlags & s_hashFlagStringKindIsSymbol; } |
| 295 | bool isAtomic() const { return m_hashAndFlags & s_hashFlagStringKindIsAtomic; } |
| 296 | void setIsAtomic(bool); |
| 297 | |
| 298 | bool isExternal() const { return bufferOwnership() == BufferExternal; } |
| 299 | |
| 300 | bool isSubString() const { return bufferOwnership() == BufferSubstring; } |
| 301 | |
| 302 | static WTF_EXPORT_PRIVATE Expected<CString, UTF8ConversionError> utf8ForCharacters(const LChar* characters, unsigned length); |
| 303 | static WTF_EXPORT_PRIVATE Expected<CString, UTF8ConversionError> utf8ForCharacters(const UChar* characters, unsigned length, ConversionMode = LenientConversion); |
| 304 | |
| 305 | WTF_EXPORT_PRIVATE Expected<CString, UTF8ConversionError> tryGetUtf8ForRange(unsigned offset, unsigned length, ConversionMode = LenientConversion) const; |
| 306 | WTF_EXPORT_PRIVATE Expected<CString, UTF8ConversionError> tryGetUtf8(ConversionMode = LenientConversion) const; |
| 307 | WTF_EXPORT_PRIVATE CString utf8(ConversionMode = LenientConversion) const; |
| 308 | |
| 309 | private: |
| 310 | static WTF_EXPORT_PRIVATE UTF8ConversionError utf8Impl(const UChar* characters, unsigned length, char*& buffer, size_t bufferSize, ConversionMode); |
| 311 | |
| 312 | // The high bits of 'hash' are always empty, but we prefer to store our flags |
| 313 | // in the low bits because it makes them slightly more efficient to access. |
| 314 | // So, we shift left and right when setting and getting our hash code. |
| 315 | void setHash(unsigned) const; |
| 316 | |
| 317 | unsigned rawHash() const { return m_hashAndFlags >> s_flagCount; } |
| 318 | |
| 319 | public: |
| 320 | bool hasHash() const { return !!rawHash(); } |
| 321 | |
| 322 | unsigned existingHash() const { ASSERT(hasHash()); return rawHash(); } |
| 323 | unsigned hash() const { return hasHash() ? rawHash() : hashSlowCase(); } |
| 324 | |
| 325 | WTF_EXPORT_PRIVATE unsigned concurrentHash() const; |
| 326 | |
| 327 | unsigned symbolAwareHash() const; |
| 328 | unsigned existingSymbolAwareHash() const; |
| 329 | |
| 330 | bool isStatic() const { return m_refCount & s_refCountFlagIsStaticString; } |
| 331 | |
| 332 | size_t refCount() const { return m_refCount / s_refCountIncrement; } |
| 333 | bool hasOneRef() const { return m_refCount == s_refCountIncrement; } |
| 334 | bool hasAtLeastOneRef() const { return m_refCount; } // For assertions. |
| 335 | |
| 336 | void ref(); |
| 337 | void deref(); |
| 338 | |
| 339 | class StaticStringImpl : private StringImplShape { |
| 340 | WTF_MAKE_NONCOPYABLE(StaticStringImpl); |
| 341 | public: |
| 342 | // Used to construct static strings, which have an special refCount that can never hit zero. |
| 343 | // This means that the static string will never be destroyed, which is important because |
| 344 | // static strings will be shared across threads & ref-counted in a non-threadsafe manner. |
| 345 | // |
| 346 | // In order to make StaticStringImpl thread safe, we also need to ensure that the rest of |
| 347 | // the fields are never mutated by threads. We have this guarantee because: |
| 348 | // |
| 349 | // 1. m_length is only set on construction and never mutated thereafter. |
| 350 | // |
| 351 | // 2. m_data8 and m_data16 are only set on construction and never mutated thereafter. |
| 352 | // We also know that a StringImpl never changes from 8 bit to 16 bit because there |
| 353 | // is no way to set/clear the s_hashFlag8BitBuffer flag other than at construction. |
| 354 | // |
| 355 | // 3. m_hashAndFlags will not be mutated by different threads because: |
| 356 | // |
| 357 | // a. StaticStringImpl's constructor sets the s_hashFlagDidReportCost flag to ensure |
| 358 | // that StringImpl::cost() returns early. |
| 359 | // This means StaticStringImpl costs are not counted. But since there should only |
| 360 | // be a finite set of StaticStringImpls, their cost can be aggregated into a single |
| 361 | // system cost if needed. |
| 362 | // b. setIsAtomic() is never called on a StaticStringImpl. |
| 363 | // setIsAtomic() asserts !isStatic(). |
| 364 | // c. setHash() is never called on a StaticStringImpl. |
| 365 | // StaticStringImpl's constructor sets the hash on construction. |
| 366 | // StringImpl::hash() only sets a new hash iff !hasHash(). |
| 367 | // Additionally, StringImpl::setHash() asserts hasHash() and !isStatic(). |
| 368 | |
| 369 | template<unsigned characterCount> constexpr StaticStringImpl(const char (&characters)[characterCount], StringKind = StringNormal); |
| 370 | template<unsigned characterCount> constexpr StaticStringImpl(const char16_t (&characters)[characterCount], StringKind = StringNormal); |
| 371 | operator StringImpl&(); |
| 372 | }; |
| 373 | |
| 374 | WTF_EXPORT_PRIVATE static StaticStringImpl s_atomicEmptyString; |
| 375 | ALWAYS_INLINE static StringImpl* empty() { return reinterpret_cast<StringImpl*>(&s_atomicEmptyString); } |
| 376 | |
| 377 | // FIXME: Does this really belong in StringImpl? |
| 378 | template<typename CharacterType> static void copyCharacters(CharacterType* destination, const CharacterType* source, unsigned numCharacters); |
| 379 | static void copyCharacters(UChar* destination, const LChar* source, unsigned numCharacters); |
| 380 | |
| 381 | // Some string features, like reference counting and the atomicity flag, are not |
| 382 | // thread-safe. We achieve thread safety by isolation, giving each thread |
| 383 | // its own copy of the string. |
| 384 | Ref<StringImpl> isolatedCopy() const; |
| 385 | |
| 386 | WTF_EXPORT_PRIVATE Ref<StringImpl> substring(unsigned position, unsigned length = MaxLength); |
| 387 | |
| 388 | UChar at(unsigned) const; |
| 389 | UChar operator[](unsigned i) const { return at(i); } |
| 390 | WTF_EXPORT_PRIVATE UChar32 characterStartingAt(unsigned); |
| 391 | |
| 392 | int toIntStrict(bool* ok = 0, int base = 10); |
| 393 | unsigned toUIntStrict(bool* ok = 0, int base = 10); |
| 394 | int64_t toInt64Strict(bool* ok = 0, int base = 10); |
| 395 | uint64_t toUInt64Strict(bool* ok = 0, int base = 10); |
| 396 | intptr_t toIntPtrStrict(bool* ok = 0, int base = 10); |
| 397 | |
| 398 | WTF_EXPORT_PRIVATE int toInt(bool* ok = 0); // ignores trailing garbage |
| 399 | unsigned toUInt(bool* ok = 0); // ignores trailing garbage |
| 400 | int64_t toInt64(bool* ok = 0); // ignores trailing garbage |
| 401 | uint64_t toUInt64(bool* ok = 0); // ignores trailing garbage |
| 402 | intptr_t toIntPtr(bool* ok = 0); // ignores trailing garbage |
| 403 | |
| 404 | // FIXME: Like the strict functions above, these give false for "ok" when there is trailing garbage. |
| 405 | // Like the non-strict functions above, these return the value when there is trailing garbage. |
| 406 | // It would be better if these were more consistent with the above functions instead. |
| 407 | double toDouble(bool* ok = 0); |
| 408 | float toFloat(bool* ok = 0); |
| 409 | |
| 410 | WTF_EXPORT_PRIVATE Ref<StringImpl> convertToASCIILowercase(); |
| 411 | WTF_EXPORT_PRIVATE Ref<StringImpl> convertToASCIIUppercase(); |
| 412 | WTF_EXPORT_PRIVATE Ref<StringImpl> convertToLowercaseWithoutLocale(); |
| 413 | WTF_EXPORT_PRIVATE Ref<StringImpl> convertToLowercaseWithoutLocaleStartingAtFailingIndex8Bit(unsigned); |
| 414 | WTF_EXPORT_PRIVATE Ref<StringImpl> convertToUppercaseWithoutLocale(); |
| 415 | WTF_EXPORT_PRIVATE Ref<StringImpl> convertToLowercaseWithLocale(const AtomicString& localeIdentifier); |
| 416 | WTF_EXPORT_PRIVATE Ref<StringImpl> convertToUppercaseWithLocale(const AtomicString& localeIdentifier); |
| 417 | |
| 418 | Ref<StringImpl> foldCase(); |
| 419 | |
| 420 | Ref<StringImpl> stripWhiteSpace(); |
| 421 | WTF_EXPORT_PRIVATE Ref<StringImpl> simplifyWhiteSpace(); |
| 422 | Ref<StringImpl> simplifyWhiteSpace(CodeUnitMatchFunction); |
| 423 | |
| 424 | Ref<StringImpl> stripLeadingAndTrailingCharacters(CodeUnitMatchFunction); |
| 425 | Ref<StringImpl> removeCharacters(CodeUnitMatchFunction); |
| 426 | |
| 427 | bool isAllASCII() const; |
| 428 | bool isAllLatin1() const; |
| 429 | template<bool isSpecialCharacter(UChar)> bool isAllSpecialCharacters() const; |
| 430 | |
| 431 | size_t find(LChar character, unsigned start = 0); |
| 432 | size_t find(char character, unsigned start = 0); |
| 433 | size_t find(UChar character, unsigned start = 0); |
| 434 | WTF_EXPORT_PRIVATE size_t find(CodeUnitMatchFunction, unsigned index = 0); |
| 435 | size_t find(const LChar*, unsigned index = 0); |
| 436 | ALWAYS_INLINE size_t find(const char* string, unsigned index = 0) { return find(reinterpret_cast<const LChar*>(string), index); } |
| 437 | WTF_EXPORT_PRIVATE size_t find(StringImpl*); |
| 438 | WTF_EXPORT_PRIVATE size_t find(StringImpl*, unsigned index); |
| 439 | WTF_EXPORT_PRIVATE size_t findIgnoringASCIICase(const StringImpl&) const; |
| 440 | WTF_EXPORT_PRIVATE size_t findIgnoringASCIICase(const StringImpl&, unsigned startOffset) const; |
| 441 | WTF_EXPORT_PRIVATE size_t findIgnoringASCIICase(const StringImpl*) const; |
| 442 | WTF_EXPORT_PRIVATE size_t findIgnoringASCIICase(const StringImpl*, unsigned startOffset) const; |
| 443 | |
| 444 | WTF_EXPORT_PRIVATE size_t reverseFind(UChar, unsigned index = MaxLength); |
| 445 | WTF_EXPORT_PRIVATE size_t reverseFind(StringImpl*, unsigned index = MaxLength); |
| 446 | |
| 447 | WTF_EXPORT_PRIVATE bool startsWith(const StringImpl*) const; |
| 448 | WTF_EXPORT_PRIVATE bool startsWith(const StringImpl&) const; |
| 449 | WTF_EXPORT_PRIVATE bool startsWithIgnoringASCIICase(const StringImpl*) const; |
| 450 | WTF_EXPORT_PRIVATE bool startsWithIgnoringASCIICase(const StringImpl&) const; |
| 451 | WTF_EXPORT_PRIVATE bool startsWith(UChar) const; |
| 452 | WTF_EXPORT_PRIVATE bool startsWith(const char*, unsigned matchLength) const; |
| 453 | template<unsigned matchLength> bool startsWith(const char (&prefix)[matchLength]) const { return startsWith(prefix, matchLength - 1); } |
| 454 | WTF_EXPORT_PRIVATE bool hasInfixStartingAt(const StringImpl&, unsigned startOffset) const; |
| 455 | |
| 456 | WTF_EXPORT_PRIVATE bool endsWith(StringImpl*); |
| 457 | WTF_EXPORT_PRIVATE bool endsWith(StringImpl&); |
| 458 | WTF_EXPORT_PRIVATE bool endsWithIgnoringASCIICase(const StringImpl*) const; |
| 459 | WTF_EXPORT_PRIVATE bool endsWithIgnoringASCIICase(const StringImpl&) const; |
| 460 | WTF_EXPORT_PRIVATE bool endsWith(UChar) const; |
| 461 | WTF_EXPORT_PRIVATE bool endsWith(const char*, unsigned matchLength) const; |
| 462 | template<unsigned matchLength> bool endsWith(const char (&prefix)[matchLength]) const { return endsWith(prefix, matchLength - 1); } |
| 463 | WTF_EXPORT_PRIVATE bool hasInfixEndingAt(const StringImpl&, unsigned endOffset) const; |
| 464 | |
| 465 | WTF_EXPORT_PRIVATE Ref<StringImpl> replace(UChar, UChar); |
| 466 | WTF_EXPORT_PRIVATE Ref<StringImpl> replace(UChar, StringImpl*); |
| 467 | ALWAYS_INLINE Ref<StringImpl> replace(UChar pattern, const char* replacement, unsigned replacementLength) { return replace(pattern, reinterpret_cast<const LChar*>(replacement), replacementLength); } |
| 468 | WTF_EXPORT_PRIVATE Ref<StringImpl> replace(UChar, const LChar*, unsigned replacementLength); |
| 469 | Ref<StringImpl> replace(UChar, const UChar*, unsigned replacementLength); |
| 470 | WTF_EXPORT_PRIVATE Ref<StringImpl> replace(StringImpl*, StringImpl*); |
| 471 | WTF_EXPORT_PRIVATE Ref<StringImpl> replace(unsigned index, unsigned length, StringImpl*); |
| 472 | |
| 473 | WTF_EXPORT_PRIVATE UCharDirection defaultWritingDirection(bool* hasStrongDirectionality = nullptr); |
| 474 | |
| 475 | #if USE(CF) |
| 476 | RetainPtr<CFStringRef> createCFString(); |
| 477 | #endif |
| 478 | |
| 479 | #ifdef __OBJC__ |
| 480 | WTF_EXPORT_PRIVATE operator NSString *(); |
| 481 | #endif |
| 482 | |
| 483 | #if STRING_STATS |
| 484 | ALWAYS_INLINE static StringStats& stringStats() { return m_stringStats; } |
| 485 | #endif |
| 486 | |
| 487 | BufferOwnership bufferOwnership() const { return static_cast<BufferOwnership>(m_hashAndFlags & s_hashMaskBufferOwnership); } |
| 488 | |
| 489 | template<typename T> static size_t () { return tailOffset<T>(); } |
| 490 | |
| 491 | protected: |
| 492 | ~StringImpl(); |
| 493 | |
| 494 | // Used to create new symbol string that holds an existing [[Description]] string as a substring buffer (BufferSubstring). |
| 495 | enum CreateSymbolTag { CreateSymbol }; |
| 496 | StringImpl(CreateSymbolTag, const LChar*, unsigned length); |
| 497 | StringImpl(CreateSymbolTag, const UChar*, unsigned length); |
| 498 | |
| 499 | // Null symbol. |
| 500 | explicit StringImpl(CreateSymbolTag); |
| 501 | |
| 502 | private: |
| 503 | template<typename> static size_t allocationSize(Checked<size_t> tailElementCount); |
| 504 | template<typename> static size_t maxInternalLength(); |
| 505 | template<typename> static size_t tailOffset(); |
| 506 | |
| 507 | bool requiresCopy() const; |
| 508 | template<typename T> const T* tailPointer() const; |
| 509 | template<typename T> T* tailPointer(); |
| 510 | StringImpl* const& substringBuffer() const; |
| 511 | StringImpl*& substringBuffer(); |
| 512 | |
| 513 | enum class CaseConvertType { Upper, Lower }; |
| 514 | template<CaseConvertType, typename CharacterType> static Ref<StringImpl> convertASCIICase(StringImpl&, const CharacterType*, unsigned); |
| 515 | |
| 516 | template<class CodeUnitPredicate> Ref<StringImpl> stripMatchedCharacters(CodeUnitPredicate); |
| 517 | template<typename CharacterType> ALWAYS_INLINE Ref<StringImpl> removeCharacters(const CharacterType* characters, CodeUnitMatchFunction); |
| 518 | template<typename CharacterType, class CodeUnitPredicate> Ref<StringImpl> simplifyMatchedCharactersToSpace(CodeUnitPredicate); |
| 519 | template<typename CharacterType> static Ref<StringImpl> constructInternal(StringImpl&, unsigned); |
| 520 | template<typename CharacterType> static Ref<StringImpl> createUninitializedInternal(unsigned, CharacterType*&); |
| 521 | template<typename CharacterType> static Ref<StringImpl> createUninitializedInternalNonEmpty(unsigned, CharacterType*&); |
| 522 | template<typename CharacterType> static Expected<Ref<StringImpl>, UTF8ConversionError> reallocateInternal(Ref<StringImpl>&&, unsigned, CharacterType*&); |
| 523 | template<typename CharacterType> static Ref<StringImpl> createInternal(const CharacterType*, unsigned); |
| 524 | WTF_EXPORT_PRIVATE NEVER_INLINE unsigned hashSlowCase() const; |
| 525 | |
| 526 | // The bottom bit in the ref count indicates a static (immortal) string. |
| 527 | static const unsigned s_refCountFlagIsStaticString = 0x1; |
| 528 | static const unsigned s_refCountIncrement = 0x2; // This allows us to ref / deref without disturbing the static string flag. |
| 529 | |
| 530 | #if STRING_STATS |
| 531 | WTF_EXPORT_PRIVATE static StringStats m_stringStats; |
| 532 | #endif |
| 533 | |
| 534 | public: |
| 535 | void assertHashIsCorrect() const; |
| 536 | }; |
| 537 | |
| 538 | using StaticStringImpl = StringImpl::StaticStringImpl; |
| 539 | |
| 540 | static_assert(sizeof(StringImpl) == sizeof(StaticStringImpl), "" ); |
| 541 | |
| 542 | #if !ASSERT_DISABLED |
| 543 | |
| 544 | // StringImpls created from StaticStringImpl will ASSERT in the generic ValueCheck<T>::checkConsistency |
| 545 | // as they are not allocated by fastMalloc. We don't currently have any way to detect that case |
| 546 | // so we ignore the consistency check for all StringImpl*. |
| 547 | template<> struct ValueCheck<StringImpl*> { |
| 548 | static void checkConsistency(const StringImpl*) { } |
| 549 | }; |
| 550 | |
| 551 | #endif |
| 552 | |
| 553 | WTF_EXPORT_PRIVATE bool equal(const StringImpl*, const StringImpl*); |
| 554 | WTF_EXPORT_PRIVATE bool equal(const StringImpl*, const LChar*); |
| 555 | inline bool equal(const StringImpl* a, const char* b) { return equal(a, reinterpret_cast<const LChar*>(b)); } |
| 556 | WTF_EXPORT_PRIVATE bool equal(const StringImpl*, const LChar*, unsigned); |
| 557 | WTF_EXPORT_PRIVATE bool equal(const StringImpl*, const UChar*, unsigned); |
| 558 | inline bool equal(const StringImpl* a, const char* b, unsigned length) { return equal(a, reinterpret_cast<const LChar*>(b), length); } |
| 559 | inline bool equal(const LChar* a, StringImpl* b) { return equal(b, a); } |
| 560 | inline bool equal(const char* a, StringImpl* b) { return equal(b, reinterpret_cast<const LChar*>(a)); } |
| 561 | WTF_EXPORT_PRIVATE bool equal(const StringImpl& a, const StringImpl& b); |
| 562 | |
| 563 | WTF_EXPORT_PRIVATE bool equalIgnoringNullity(StringImpl*, StringImpl*); |
| 564 | WTF_EXPORT_PRIVATE bool equalIgnoringNullity(const UChar*, size_t length, StringImpl*); |
| 565 | |
| 566 | bool equalIgnoringASCIICase(const StringImpl&, const StringImpl&); |
| 567 | WTF_EXPORT_PRIVATE bool equalIgnoringASCIICase(const StringImpl*, const StringImpl*); |
| 568 | bool equalIgnoringASCIICase(const StringImpl&, const char*); |
| 569 | bool equalIgnoringASCIICase(const StringImpl*, const char*); |
| 570 | |
| 571 | WTF_EXPORT_PRIVATE bool equalIgnoringASCIICaseNonNull(const StringImpl*, const StringImpl*); |
| 572 | |
| 573 | template<unsigned length> bool equalLettersIgnoringASCIICase(const StringImpl&, const char (&lowercaseLetters)[length]); |
| 574 | template<unsigned length> bool equalLettersIgnoringASCIICase(const StringImpl*, const char (&lowercaseLetters)[length]); |
| 575 | |
| 576 | size_t find(const LChar*, unsigned length, CodeUnitMatchFunction, unsigned index = 0); |
| 577 | size_t find(const UChar*, unsigned length, CodeUnitMatchFunction, unsigned index = 0); |
| 578 | |
| 579 | template<typename CharacterType> size_t reverseFindLineTerminator(const CharacterType*, unsigned length, unsigned index = StringImpl::MaxLength); |
| 580 | template<typename CharacterType> size_t reverseFind(const CharacterType*, unsigned length, CharacterType matchCharacter, unsigned index = StringImpl::MaxLength); |
| 581 | size_t reverseFind(const UChar*, unsigned length, LChar matchCharacter, unsigned index = StringImpl::MaxLength); |
| 582 | size_t reverseFind(const LChar*, unsigned length, UChar matchCharacter, unsigned index = StringImpl::MaxLength); |
| 583 | |
| 584 | template<size_t inlineCapacity> bool equalIgnoringNullity(const Vector<UChar, inlineCapacity>&, StringImpl*); |
| 585 | |
| 586 | template<typename CharacterType1, typename CharacterType2> int codePointCompare(const CharacterType1*, unsigned length1, const CharacterType2*, unsigned length2); |
| 587 | int codePointCompare(const StringImpl*, const StringImpl*); |
| 588 | |
| 589 | // FIXME: Should rename this to make clear it uses the Unicode definition of whitespace. |
| 590 | // Most WebKit callers don't want that would use isASCIISpace or isHTMLSpace instead. |
| 591 | bool isSpaceOrNewline(UChar32); |
| 592 | |
| 593 | template<typename CharacterType> unsigned lengthOfNullTerminatedString(const CharacterType*); |
| 594 | |
| 595 | // StringHash is the default hash for StringImpl* and RefPtr<StringImpl> |
| 596 | template<typename T> struct DefaultHash; |
| 597 | template<> struct DefaultHash<StringImpl*> { |
| 598 | typedef StringHash Hash; |
| 599 | }; |
| 600 | template<> struct DefaultHash<RefPtr<StringImpl>> { |
| 601 | typedef StringHash Hash; |
| 602 | }; |
| 603 | |
| 604 | #define MAKE_STATIC_STRING_IMPL(characters) ([] { \ |
| 605 | static StaticStringImpl impl(characters); \ |
| 606 | return &impl; \ |
| 607 | }()) |
| 608 | |
| 609 | template<> ALWAYS_INLINE Ref<StringImpl> StringImpl::constructInternal<LChar>(StringImpl& string, unsigned length) |
| 610 | { |
| 611 | return adoptRef(*new (NotNull, &string) StringImpl { length, Force8BitConstructor }); |
| 612 | } |
| 613 | |
| 614 | template<> ALWAYS_INLINE Ref<StringImpl> StringImpl::constructInternal<UChar>(StringImpl& string, unsigned length) |
| 615 | { |
| 616 | return adoptRef(*new (NotNull, &string) StringImpl { length }); |
| 617 | } |
| 618 | |
| 619 | template<> ALWAYS_INLINE const LChar* StringImpl::characters<LChar>() const |
| 620 | { |
| 621 | return characters8(); |
| 622 | } |
| 623 | |
| 624 | template<> ALWAYS_INLINE const UChar* StringImpl::characters<UChar>() const |
| 625 | { |
| 626 | return characters16(); |
| 627 | } |
| 628 | |
| 629 | inline size_t find(const LChar* characters, unsigned length, CodeUnitMatchFunction matchFunction, unsigned index) |
| 630 | { |
| 631 | while (index < length) { |
| 632 | if (matchFunction(characters[index])) |
| 633 | return index; |
| 634 | ++index; |
| 635 | } |
| 636 | return notFound; |
| 637 | } |
| 638 | |
| 639 | inline size_t find(const UChar* characters, unsigned length, CodeUnitMatchFunction matchFunction, unsigned index) |
| 640 | { |
| 641 | while (index < length) { |
| 642 | if (matchFunction(characters[index])) |
| 643 | return index; |
| 644 | ++index; |
| 645 | } |
| 646 | return notFound; |
| 647 | } |
| 648 | |
| 649 | template<typename CharacterType> inline size_t reverseFindLineTerminator(const CharacterType* characters, unsigned length, unsigned index) |
| 650 | { |
| 651 | if (!length) |
| 652 | return notFound; |
| 653 | if (index >= length) |
| 654 | index = length - 1; |
| 655 | auto character = characters[index]; |
| 656 | while (character != '\n' && character != '\r') { |
| 657 | if (!index--) |
| 658 | return notFound; |
| 659 | character = characters[index]; |
| 660 | } |
| 661 | return index; |
| 662 | } |
| 663 | |
| 664 | template<typename CharacterType> inline size_t reverseFind(const CharacterType* characters, unsigned length, CharacterType matchCharacter, unsigned index) |
| 665 | { |
| 666 | if (!length) |
| 667 | return notFound; |
| 668 | if (index >= length) |
| 669 | index = length - 1; |
| 670 | while (characters[index] != matchCharacter) { |
| 671 | if (!index--) |
| 672 | return notFound; |
| 673 | } |
| 674 | return index; |
| 675 | } |
| 676 | |
| 677 | ALWAYS_INLINE size_t reverseFind(const UChar* characters, unsigned length, LChar matchCharacter, unsigned index) |
| 678 | { |
| 679 | return reverseFind(characters, length, static_cast<UChar>(matchCharacter), index); |
| 680 | } |
| 681 | |
| 682 | inline size_t reverseFind(const LChar* characters, unsigned length, UChar matchCharacter, unsigned index) |
| 683 | { |
| 684 | if (matchCharacter & ~0xFF) |
| 685 | return notFound; |
| 686 | return reverseFind(characters, length, static_cast<LChar>(matchCharacter), index); |
| 687 | } |
| 688 | |
| 689 | inline size_t StringImpl::find(LChar character, unsigned start) |
| 690 | { |
| 691 | if (is8Bit()) |
| 692 | return WTF::find(characters8(), m_length, character, start); |
| 693 | return WTF::find(characters16(), m_length, character, start); |
| 694 | } |
| 695 | |
| 696 | ALWAYS_INLINE size_t StringImpl::find(char character, unsigned start) |
| 697 | { |
| 698 | return find(static_cast<LChar>(character), start); |
| 699 | } |
| 700 | |
| 701 | inline size_t StringImpl::find(UChar character, unsigned start) |
| 702 | { |
| 703 | if (is8Bit()) |
| 704 | return WTF::find(characters8(), m_length, character, start); |
| 705 | return WTF::find(characters16(), m_length, character, start); |
| 706 | } |
| 707 | |
| 708 | template<size_t inlineCapacity> inline bool equalIgnoringNullity(const Vector<UChar, inlineCapacity>& a, StringImpl* b) |
| 709 | { |
| 710 | return equalIgnoringNullity(a.data(), a.size(), b); |
| 711 | } |
| 712 | |
| 713 | template<typename CharacterType1, typename CharacterType2> inline int codePointCompare(const CharacterType1* characters1, unsigned length1, const CharacterType2* characters2, unsigned length2) |
| 714 | { |
| 715 | unsigned commonLength = std::min(length1, length2); |
| 716 | |
| 717 | unsigned position = 0; |
| 718 | while (position < commonLength && *characters1 == *characters2) { |
| 719 | ++characters1; |
| 720 | ++characters2; |
| 721 | ++position; |
| 722 | } |
| 723 | |
| 724 | if (position < commonLength) |
| 725 | return (characters1[0] > characters2[0]) ? 1 : -1; |
| 726 | |
| 727 | if (length1 == length2) |
| 728 | return 0; |
| 729 | return (length1 > length2) ? 1 : -1; |
| 730 | } |
| 731 | |
| 732 | inline int codePointCompare(const StringImpl* string1, const StringImpl* string2) |
| 733 | { |
| 734 | // FIXME: Should null strings compare as less than empty strings rather than equal to them? |
| 735 | if (!string1) |
| 736 | return (string2 && string2->length()) ? -1 : 0; |
| 737 | if (!string2) |
| 738 | return string1->length() ? 1 : 0; |
| 739 | |
| 740 | bool string1Is8Bit = string1->is8Bit(); |
| 741 | bool string2Is8Bit = string2->is8Bit(); |
| 742 | if (string1Is8Bit) { |
| 743 | if (string2Is8Bit) |
| 744 | return codePointCompare(string1->characters8(), string1->length(), string2->characters8(), string2->length()); |
| 745 | return codePointCompare(string1->characters8(), string1->length(), string2->characters16(), string2->length()); |
| 746 | } |
| 747 | if (string2Is8Bit) |
| 748 | return codePointCompare(string1->characters16(), string1->length(), string2->characters8(), string2->length()); |
| 749 | return codePointCompare(string1->characters16(), string1->length(), string2->characters16(), string2->length()); |
| 750 | } |
| 751 | |
| 752 | inline bool isSpaceOrNewline(UChar32 character) |
| 753 | { |
| 754 | // Use isASCIISpace() for all Latin-1 characters. This will include newlines, which aren't included in Unicode DirWS. |
| 755 | return character <= 0xFF ? isASCIISpace(character) : u_charDirection(character) == U_WHITE_SPACE_NEUTRAL; |
| 756 | } |
| 757 | |
| 758 | template<typename CharacterType> inline unsigned lengthOfNullTerminatedString(const CharacterType* string) |
| 759 | { |
| 760 | ASSERT(string); |
| 761 | size_t length = 0; |
| 762 | while (string[length]) |
| 763 | ++length; |
| 764 | |
| 765 | RELEASE_ASSERT(length < StringImpl::MaxLength); |
| 766 | return static_cast<unsigned>(length); |
| 767 | } |
| 768 | |
| 769 | inline StringImplShape::StringImplShape(unsigned refCount, unsigned length, const LChar* data8, unsigned hashAndFlags) |
| 770 | : m_refCount(refCount) |
| 771 | , m_length(length) |
| 772 | , m_data8(data8) |
| 773 | , m_hashAndFlags(hashAndFlags) |
| 774 | { |
| 775 | } |
| 776 | |
| 777 | inline StringImplShape::StringImplShape(unsigned refCount, unsigned length, const UChar* data16, unsigned hashAndFlags) |
| 778 | : m_refCount(refCount) |
| 779 | , m_length(length) |
| 780 | , m_data16(data16) |
| 781 | , m_hashAndFlags(hashAndFlags) |
| 782 | { |
| 783 | } |
| 784 | |
| 785 | template<unsigned characterCount> constexpr StringImplShape::StringImplShape(unsigned refCount, unsigned length, const char (&characters)[characterCount], unsigned hashAndFlags, ConstructWithConstExprTag) |
| 786 | : m_refCount(refCount) |
| 787 | , m_length(length) |
| 788 | , m_data8Char(characters) |
| 789 | , m_hashAndFlags(hashAndFlags) |
| 790 | { |
| 791 | } |
| 792 | |
| 793 | template<unsigned characterCount> constexpr StringImplShape::StringImplShape(unsigned refCount, unsigned length, const char16_t (&characters)[characterCount], unsigned hashAndFlags, ConstructWithConstExprTag) |
| 794 | : m_refCount(refCount) |
| 795 | , m_length(length) |
| 796 | , m_data16Char(characters) |
| 797 | , m_hashAndFlags(hashAndFlags) |
| 798 | { |
| 799 | } |
| 800 | |
| 801 | inline Ref<StringImpl> StringImpl::isolatedCopy() const |
| 802 | { |
| 803 | if (!requiresCopy()) { |
| 804 | if (is8Bit()) |
| 805 | return StringImpl::createWithoutCopying(m_data8, m_length); |
| 806 | return StringImpl::createWithoutCopying(m_data16, m_length); |
| 807 | } |
| 808 | |
| 809 | if (is8Bit()) |
| 810 | return create(m_data8, m_length); |
| 811 | return create(m_data16, m_length); |
| 812 | } |
| 813 | |
| 814 | inline bool StringImpl::isAllASCII() const |
| 815 | { |
| 816 | if (is8Bit()) |
| 817 | return charactersAreAllASCII(characters8(), length()); |
| 818 | return charactersAreAllASCII(characters16(), length()); |
| 819 | } |
| 820 | |
| 821 | inline bool StringImpl::isAllLatin1() const |
| 822 | { |
| 823 | if (is8Bit()) |
| 824 | return true; |
| 825 | auto* characters = characters16(); |
| 826 | UChar ored = 0; |
| 827 | for (size_t i = 0; i < length(); ++i) |
| 828 | ored |= characters[i]; |
| 829 | return !(ored & 0xFF00); |
| 830 | } |
| 831 | |
| 832 | template<bool isSpecialCharacter(UChar), typename CharacterType> inline bool isAllSpecialCharacters(const CharacterType* characters, size_t length) |
| 833 | { |
| 834 | for (size_t i = 0; i < length; ++i) { |
| 835 | if (!isSpecialCharacter(characters[i])) |
| 836 | return false; |
| 837 | } |
| 838 | return true; |
| 839 | } |
| 840 | |
| 841 | template<bool isSpecialCharacter(UChar)> inline bool StringImpl::isAllSpecialCharacters() const |
| 842 | { |
| 843 | if (is8Bit()) |
| 844 | return WTF::isAllSpecialCharacters<isSpecialCharacter>(characters8(), length()); |
| 845 | return WTF::isAllSpecialCharacters<isSpecialCharacter>(characters16(), length()); |
| 846 | } |
| 847 | |
| 848 | inline StringImpl::StringImpl(unsigned length, Force8Bit) |
| 849 | : StringImplShape(s_refCountIncrement, length, tailPointer<LChar>(), s_hashFlag8BitBuffer | StringNormal | BufferInternal) |
| 850 | { |
| 851 | ASSERT(m_data8); |
| 852 | ASSERT(m_length); |
| 853 | |
| 854 | STRING_STATS_ADD_8BIT_STRING(m_length); |
| 855 | } |
| 856 | |
| 857 | inline StringImpl::StringImpl(unsigned length) |
| 858 | : StringImplShape(s_refCountIncrement, length, tailPointer<UChar>(), StringNormal | BufferInternal) |
| 859 | { |
| 860 | ASSERT(m_data16); |
| 861 | ASSERT(m_length); |
| 862 | |
| 863 | STRING_STATS_ADD_16BIT_STRING(m_length); |
| 864 | } |
| 865 | |
| 866 | inline StringImpl::StringImpl(MallocPtr<LChar> characters, unsigned length) |
| 867 | : StringImplShape(s_refCountIncrement, length, characters.leakPtr(), s_hashFlag8BitBuffer | StringNormal | BufferOwned) |
| 868 | { |
| 869 | ASSERT(m_data8); |
| 870 | ASSERT(m_length); |
| 871 | |
| 872 | STRING_STATS_ADD_8BIT_STRING(m_length); |
| 873 | } |
| 874 | |
| 875 | inline StringImpl::StringImpl(const UChar* characters, unsigned length, ConstructWithoutCopyingTag) |
| 876 | : StringImplShape(s_refCountIncrement, length, characters, StringNormal | BufferInternal) |
| 877 | { |
| 878 | ASSERT(m_data16); |
| 879 | ASSERT(m_length); |
| 880 | |
| 881 | STRING_STATS_ADD_16BIT_STRING(m_length); |
| 882 | } |
| 883 | |
| 884 | inline StringImpl::StringImpl(const LChar* characters, unsigned length, ConstructWithoutCopyingTag) |
| 885 | : StringImplShape(s_refCountIncrement, length, characters, s_hashFlag8BitBuffer | StringNormal | BufferInternal) |
| 886 | { |
| 887 | ASSERT(m_data8); |
| 888 | ASSERT(m_length); |
| 889 | |
| 890 | STRING_STATS_ADD_8BIT_STRING(m_length); |
| 891 | } |
| 892 | |
| 893 | inline StringImpl::StringImpl(MallocPtr<UChar> characters, unsigned length) |
| 894 | : StringImplShape(s_refCountIncrement, length, characters.leakPtr(), StringNormal | BufferOwned) |
| 895 | { |
| 896 | ASSERT(m_data16); |
| 897 | ASSERT(m_length); |
| 898 | |
| 899 | STRING_STATS_ADD_16BIT_STRING(m_length); |
| 900 | } |
| 901 | |
| 902 | inline StringImpl::StringImpl(const LChar* characters, unsigned length, Ref<StringImpl>&& base) |
| 903 | : StringImplShape(s_refCountIncrement, length, characters, s_hashFlag8BitBuffer | StringNormal | BufferSubstring) |
| 904 | { |
| 905 | ASSERT(is8Bit()); |
| 906 | ASSERT(m_data8); |
| 907 | ASSERT(m_length); |
| 908 | ASSERT(base->bufferOwnership() != BufferSubstring); |
| 909 | |
| 910 | substringBuffer() = &base.leakRef(); |
| 911 | |
| 912 | STRING_STATS_ADD_8BIT_STRING2(m_length, true); |
| 913 | } |
| 914 | |
| 915 | inline StringImpl::StringImpl(const UChar* characters, unsigned length, Ref<StringImpl>&& base) |
| 916 | : StringImplShape(s_refCountIncrement, length, characters, StringNormal | BufferSubstring) |
| 917 | { |
| 918 | ASSERT(!is8Bit()); |
| 919 | ASSERT(m_data16); |
| 920 | ASSERT(m_length); |
| 921 | ASSERT(base->bufferOwnership() != BufferSubstring); |
| 922 | |
| 923 | substringBuffer() = &base.leakRef(); |
| 924 | |
| 925 | STRING_STATS_ADD_16BIT_STRING2(m_length, true); |
| 926 | } |
| 927 | |
| 928 | template<size_t inlineCapacity> inline Ref<StringImpl> StringImpl::create8BitIfPossible(const Vector<UChar, inlineCapacity>& vector) |
| 929 | { |
| 930 | return create8BitIfPossible(vector.data(), vector.size()); |
| 931 | } |
| 932 | |
| 933 | ALWAYS_INLINE Ref<StringImpl> StringImpl::createSubstringSharingImpl(StringImpl& rep, unsigned offset, unsigned length) |
| 934 | { |
| 935 | ASSERT(length <= rep.length()); |
| 936 | |
| 937 | if (!length) |
| 938 | return *empty(); |
| 939 | |
| 940 | // Coyping the thing would save more memory sometimes, largely due to the size of pointer. |
| 941 | size_t substringSize = allocationSize<StringImpl*>(1); |
| 942 | if (rep.is8Bit()) { |
| 943 | if (substringSize >= allocationSize<LChar>(length)) |
| 944 | return create(rep.m_data8 + offset, length); |
| 945 | } else { |
| 946 | if (substringSize >= allocationSize<UChar>(length)) |
| 947 | return create(rep.m_data16 + offset, length); |
| 948 | } |
| 949 | |
| 950 | auto* ownerRep = ((rep.bufferOwnership() == BufferSubstring) ? rep.substringBuffer() : &rep); |
| 951 | |
| 952 | // We allocate a buffer that contains both the StringImpl struct as well as the pointer to the owner string. |
| 953 | auto* stringImpl = static_cast<StringImpl*>(fastMalloc(substringSize)); |
| 954 | if (rep.is8Bit()) |
| 955 | return adoptRef(*new (NotNull, stringImpl) StringImpl(rep.m_data8 + offset, length, *ownerRep)); |
| 956 | return adoptRef(*new (NotNull, stringImpl) StringImpl(rep.m_data16 + offset, length, *ownerRep)); |
| 957 | } |
| 958 | |
| 959 | template<unsigned characterCount> ALWAYS_INLINE Ref<StringImpl> StringImpl::createFromLiteral(const char (&characters)[characterCount]) |
| 960 | { |
| 961 | COMPILE_ASSERT(characterCount > 1, StringImplFromLiteralNotEmpty); |
| 962 | COMPILE_ASSERT((characterCount - 1 <= ((unsigned(~0) - sizeof(StringImpl)) / sizeof(LChar))), StringImplFromLiteralCannotOverflow); |
| 963 | |
| 964 | return createWithoutCopying(reinterpret_cast<const LChar*>(characters), characterCount - 1); |
| 965 | } |
| 966 | |
| 967 | template<typename CharacterType> ALWAYS_INLINE RefPtr<StringImpl> StringImpl::tryCreateUninitialized(unsigned length, CharacterType*& output) |
| 968 | { |
| 969 | if (!length) { |
| 970 | output = nullptr; |
| 971 | return empty(); |
| 972 | } |
| 973 | |
| 974 | if (length > maxInternalLength<CharacterType>()) { |
| 975 | output = nullptr; |
| 976 | return nullptr; |
| 977 | } |
| 978 | StringImpl* result; |
| 979 | if (!tryFastMalloc(allocationSize<CharacterType>(length)).getValue(result)) { |
| 980 | output = nullptr; |
| 981 | return nullptr; |
| 982 | } |
| 983 | output = result->tailPointer<CharacterType>(); |
| 984 | |
| 985 | return constructInternal<CharacterType>(*result, length); |
| 986 | } |
| 987 | |
| 988 | template<typename CharacterType, size_t inlineCapacity, typename OverflowHandler, size_t minCapacity> |
| 989 | inline Ref<StringImpl> StringImpl::adopt(Vector<CharacterType, inlineCapacity, OverflowHandler, minCapacity>&& vector) |
| 990 | { |
| 991 | if (size_t size = vector.size()) { |
| 992 | ASSERT(vector.data()); |
| 993 | if (size > MaxLength) |
| 994 | CRASH(); |
| 995 | return adoptRef(*new StringImpl(vector.releaseBuffer(), size)); |
| 996 | } |
| 997 | return *empty(); |
| 998 | } |
| 999 | |
| 1000 | inline size_t StringImpl::cost() const |
| 1001 | { |
| 1002 | // For substrings, return the cost of the base string. |
| 1003 | if (bufferOwnership() == BufferSubstring) |
| 1004 | return substringBuffer()->cost(); |
| 1005 | |
| 1006 | // Note: we must not alter the m_hashAndFlags field in instances of StaticStringImpl. |
| 1007 | // We ensure this by pre-setting the s_hashFlagDidReportCost bit in all instances of |
| 1008 | // StaticStringImpl. As a result, StaticStringImpl instances will always return a cost of |
| 1009 | // 0 here and avoid modifying m_hashAndFlags. |
| 1010 | if (m_hashAndFlags & s_hashFlagDidReportCost) |
| 1011 | return 0; |
| 1012 | |
| 1013 | m_hashAndFlags |= s_hashFlagDidReportCost; |
| 1014 | size_t result = m_length; |
| 1015 | if (!is8Bit()) |
| 1016 | result <<= 1; |
| 1017 | return result; |
| 1018 | } |
| 1019 | |
| 1020 | inline size_t StringImpl::costDuringGC() |
| 1021 | { |
| 1022 | if (isStatic()) |
| 1023 | return 0; |
| 1024 | |
| 1025 | if (bufferOwnership() == BufferSubstring) |
| 1026 | return divideRoundedUp(substringBuffer()->costDuringGC(), refCount()); |
| 1027 | |
| 1028 | size_t result = m_length; |
| 1029 | if (!is8Bit()) |
| 1030 | result <<= 1; |
| 1031 | return divideRoundedUp(result, refCount()); |
| 1032 | } |
| 1033 | |
| 1034 | inline void StringImpl::setIsAtomic(bool isAtomic) |
| 1035 | { |
| 1036 | ASSERT(!isStatic()); |
| 1037 | ASSERT(!isSymbol()); |
| 1038 | if (isAtomic) |
| 1039 | m_hashAndFlags |= s_hashFlagStringKindIsAtomic; |
| 1040 | else |
| 1041 | m_hashAndFlags &= ~s_hashFlagStringKindIsAtomic; |
| 1042 | } |
| 1043 | |
| 1044 | inline void StringImpl::setHash(unsigned hash) const |
| 1045 | { |
| 1046 | // The high bits of 'hash' are always empty, but we prefer to store our flags |
| 1047 | // in the low bits because it makes them slightly more efficient to access. |
| 1048 | // So, we shift left and right when setting and getting our hash code. |
| 1049 | |
| 1050 | ASSERT(!hasHash()); |
| 1051 | ASSERT(!isStatic()); |
| 1052 | // Multiple clients assume that StringHasher is the canonical string hash function. |
| 1053 | ASSERT(hash == (is8Bit() ? StringHasher::computeHashAndMaskTop8Bits(m_data8, m_length) : StringHasher::computeHashAndMaskTop8Bits(m_data16, m_length))); |
| 1054 | ASSERT(!(hash & (s_flagMask << (8 * sizeof(hash) - s_flagCount)))); // Verify that enough high bits are empty. |
| 1055 | |
| 1056 | hash <<= s_flagCount; |
| 1057 | ASSERT(!(hash & m_hashAndFlags)); // Verify that enough low bits are empty after shift. |
| 1058 | ASSERT(hash); // Verify that 0 is a valid sentinel hash value. |
| 1059 | |
| 1060 | m_hashAndFlags |= hash; // Store hash with flags in low bits. |
| 1061 | } |
| 1062 | |
| 1063 | inline void StringImpl::ref() |
| 1064 | { |
| 1065 | STRING_STATS_REF_STRING(*this); |
| 1066 | |
| 1067 | m_refCount += s_refCountIncrement; |
| 1068 | } |
| 1069 | |
| 1070 | inline void StringImpl::deref() |
| 1071 | { |
| 1072 | STRING_STATS_DEREF_STRING(*this); |
| 1073 | |
| 1074 | unsigned tempRefCount = m_refCount - s_refCountIncrement; |
| 1075 | if (!tempRefCount) { |
| 1076 | StringImpl::destroy(this); |
| 1077 | return; |
| 1078 | } |
| 1079 | m_refCount = tempRefCount; |
| 1080 | } |
| 1081 | |
| 1082 | template<typename CharacterType> inline void StringImpl::copyCharacters(CharacterType* destination, const CharacterType* source, unsigned numCharacters) |
| 1083 | { |
| 1084 | if (numCharacters == 1) { |
| 1085 | *destination = *source; |
| 1086 | return; |
| 1087 | } |
| 1088 | memcpy(destination, source, numCharacters * sizeof(CharacterType)); |
| 1089 | } |
| 1090 | |
| 1091 | ALWAYS_INLINE void StringImpl::copyCharacters(UChar* destination, const LChar* source, unsigned numCharacters) |
| 1092 | { |
| 1093 | for (unsigned i = 0; i < numCharacters; ++i) |
| 1094 | destination[i] = source[i]; |
| 1095 | } |
| 1096 | |
| 1097 | inline UChar StringImpl::at(unsigned i) const |
| 1098 | { |
| 1099 | ASSERT_WITH_SECURITY_IMPLICATION(i < m_length); |
| 1100 | return is8Bit() ? m_data8[i] : m_data16[i]; |
| 1101 | } |
| 1102 | |
| 1103 | inline StringImpl::StringImpl(CreateSymbolTag, const LChar* characters, unsigned length) |
| 1104 | : StringImplShape(s_refCountIncrement, length, characters, s_hashFlag8BitBuffer | StringSymbol | BufferSubstring) |
| 1105 | { |
| 1106 | ASSERT(is8Bit()); |
| 1107 | ASSERT(m_data8); |
| 1108 | STRING_STATS_ADD_8BIT_STRING2(m_length, true); |
| 1109 | } |
| 1110 | |
| 1111 | inline StringImpl::StringImpl(CreateSymbolTag, const UChar* characters, unsigned length) |
| 1112 | : StringImplShape(s_refCountIncrement, length, characters, StringSymbol | BufferSubstring) |
| 1113 | { |
| 1114 | ASSERT(!is8Bit()); |
| 1115 | ASSERT(m_data16); |
| 1116 | STRING_STATS_ADD_16BIT_STRING2(m_length, true); |
| 1117 | } |
| 1118 | |
| 1119 | inline StringImpl::StringImpl(CreateSymbolTag) |
| 1120 | : StringImplShape(s_refCountIncrement, 0, empty()->characters8(), s_hashFlag8BitBuffer | StringSymbol | BufferSubstring) |
| 1121 | { |
| 1122 | ASSERT(is8Bit()); |
| 1123 | ASSERT(m_data8); |
| 1124 | STRING_STATS_ADD_8BIT_STRING2(m_length, true); |
| 1125 | } |
| 1126 | |
| 1127 | template<typename T> inline size_t StringImpl::allocationSize(Checked<size_t> tailElementCount) |
| 1128 | { |
| 1129 | return (tailOffset<T>() + tailElementCount * sizeof(T)).unsafeGet(); |
| 1130 | } |
| 1131 | |
| 1132 | template<typename CharacterType> |
| 1133 | inline size_t StringImpl::maxInternalLength() |
| 1134 | { |
| 1135 | // In order to not overflow the unsigned length, the check for (std::numeric_limits<unsigned>::max() - sizeof(StringImpl)) is needed when sizeof(CharacterType) == 2. |
| 1136 | return std::min(static_cast<size_t>(MaxLength), (std::numeric_limits<unsigned>::max() - sizeof(StringImpl)) / sizeof(CharacterType)); |
| 1137 | } |
| 1138 | |
| 1139 | template<typename T> inline size_t StringImpl::tailOffset() |
| 1140 | { |
| 1141 | #if COMPILER(MSVC) |
| 1142 | // MSVC doesn't support alignof yet. |
| 1143 | return roundUpToMultipleOf<sizeof(T)>(sizeof(StringImpl)); |
| 1144 | #else |
| 1145 | return roundUpToMultipleOf<alignof(T)>(offsetof(StringImpl, m_hashAndFlags) + sizeof(StringImpl::m_hashAndFlags)); |
| 1146 | #endif |
| 1147 | } |
| 1148 | |
| 1149 | inline bool StringImpl::requiresCopy() const |
| 1150 | { |
| 1151 | if (bufferOwnership() != BufferInternal) |
| 1152 | return true; |
| 1153 | |
| 1154 | if (is8Bit()) |
| 1155 | return m_data8 == tailPointer<LChar>(); |
| 1156 | return m_data16 == tailPointer<UChar>(); |
| 1157 | } |
| 1158 | |
| 1159 | template<typename T> inline const T* StringImpl::tailPointer() const |
| 1160 | { |
| 1161 | return reinterpret_cast_ptr<const T*>(reinterpret_cast<const uint8_t*>(this) + tailOffset<T>()); |
| 1162 | } |
| 1163 | |
| 1164 | template<typename T> inline T* StringImpl::tailPointer() |
| 1165 | { |
| 1166 | return reinterpret_cast_ptr<T*>(reinterpret_cast<uint8_t*>(this) + tailOffset<T>()); |
| 1167 | } |
| 1168 | |
| 1169 | inline StringImpl* const& StringImpl::substringBuffer() const |
| 1170 | { |
| 1171 | ASSERT(bufferOwnership() == BufferSubstring); |
| 1172 | |
| 1173 | return *tailPointer<StringImpl*>(); |
| 1174 | } |
| 1175 | |
| 1176 | inline StringImpl*& StringImpl::substringBuffer() |
| 1177 | { |
| 1178 | ASSERT(bufferOwnership() == BufferSubstring); |
| 1179 | |
| 1180 | return *tailPointer<StringImpl*>(); |
| 1181 | } |
| 1182 | |
| 1183 | inline void StringImpl::assertHashIsCorrect() const |
| 1184 | { |
| 1185 | ASSERT(existingHash() == StringHasher::computeHashAndMaskTop8Bits(characters8(), length())); |
| 1186 | } |
| 1187 | |
| 1188 | template<unsigned characterCount> constexpr StringImpl::StaticStringImpl::StaticStringImpl(const char (&characters)[characterCount], StringKind stringKind) |
| 1189 | : StringImplShape(s_refCountFlagIsStaticString, characterCount - 1, characters, |
| 1190 | s_hashFlag8BitBuffer | s_hashFlagDidReportCost | stringKind | BufferInternal | (StringHasher::computeLiteralHashAndMaskTop8Bits(characters) << s_flagCount), ConstructWithConstExpr) |
| 1191 | { |
| 1192 | } |
| 1193 | |
| 1194 | template<unsigned characterCount> constexpr StringImpl::StaticStringImpl::StaticStringImpl(const char16_t (&characters)[characterCount], StringKind stringKind) |
| 1195 | : StringImplShape(s_refCountFlagIsStaticString, characterCount - 1, characters, |
| 1196 | s_hashFlagDidReportCost | stringKind | BufferInternal | (StringHasher::computeLiteralHashAndMaskTop8Bits(characters) << s_flagCount), ConstructWithConstExpr) |
| 1197 | { |
| 1198 | } |
| 1199 | |
| 1200 | inline StringImpl::StaticStringImpl::operator StringImpl&() |
| 1201 | { |
| 1202 | return *reinterpret_cast<StringImpl*>(this); |
| 1203 | } |
| 1204 | |
| 1205 | inline bool equalIgnoringASCIICase(const StringImpl& a, const StringImpl& b) |
| 1206 | { |
| 1207 | return equalIgnoringASCIICaseCommon(a, b); |
| 1208 | } |
| 1209 | |
| 1210 | inline bool equalIgnoringASCIICase(const StringImpl& a, const char* b) |
| 1211 | { |
| 1212 | return equalIgnoringASCIICaseCommon(a, b); |
| 1213 | } |
| 1214 | |
| 1215 | inline bool equalIgnoringASCIICase(const StringImpl* a, const char* b) |
| 1216 | { |
| 1217 | return a && equalIgnoringASCIICase(*a, b); |
| 1218 | } |
| 1219 | |
| 1220 | template<unsigned length> inline bool startsWithLettersIgnoringASCIICase(const StringImpl& string, const char (&lowercaseLetters)[length]) |
| 1221 | { |
| 1222 | return startsWithLettersIgnoringASCIICaseCommon(string, lowercaseLetters); |
| 1223 | } |
| 1224 | |
| 1225 | template<unsigned length> inline bool startsWithLettersIgnoringASCIICase(const StringImpl* string, const char (&lowercaseLetters)[length]) |
| 1226 | { |
| 1227 | return string && startsWithLettersIgnoringASCIICase(*string, lowercaseLetters); |
| 1228 | } |
| 1229 | |
| 1230 | template<unsigned length> inline bool equalLettersIgnoringASCIICase(const StringImpl& string, const char (&lowercaseLetters)[length]) |
| 1231 | { |
| 1232 | return equalLettersIgnoringASCIICaseCommon(string, lowercaseLetters); |
| 1233 | } |
| 1234 | |
| 1235 | template<unsigned length> inline bool equalLettersIgnoringASCIICase(const StringImpl* string, const char (&lowercaseLetters)[length]) |
| 1236 | { |
| 1237 | return string && equalLettersIgnoringASCIICase(*string, lowercaseLetters); |
| 1238 | } |
| 1239 | |
| 1240 | } // namespace WTF |
| 1241 | |
| 1242 | using WTF::StaticStringImpl; |
| 1243 | using WTF::StringImpl; |
| 1244 | using WTF::equal; |
| 1245 | using WTF::isLatin1; |
| 1246 | |