wtf/text/StringImpl.h

   1 /*
   2  * Copyright (C) 1999 Lars Knoll (knoll@kde.org)
   3  * Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010 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 #ifndef StringImpl_h
  24 #define StringImpl_h
  25
  26 #include <limits.h>
  27 #include <wtf/ASCIICType.h>
  28 #include <wtf/CrossThreadRefCounted.h>
  29 #include <wtf/Forward.h>
  30 #include <wtf/OwnFastMallocPtr.h>
  31 #include <wtf/StdLibExtras.h>
  32 #include <wtf/StringHasher.h>
  33 #include <wtf/Vector.h>
  34 #include <wtf/text/StringImplBase.h>
  35 #include <wtf/unicode/Unicode.h>
  36
  37 #if USE(CF)
  38 typedef const struct __CFString * CFStringRef;
  39 #endif
  40
  41 #ifdef __OBJC__
  42 @class NSString;
  43 #endif
  44
  45 // FIXME: This is a temporary layering violation while we move string code to WTF.
  46 // Landing the file moves in one patch, will follow on with patches to change the namespaces.
  47 namespace JSC {
  48 struct IdentifierCStringTranslator;
  49 struct IdentifierUCharBufferTranslator;
  50 }
  51
  52 namespace WTF {
  53
  54 struct CStringTranslator;
  55 struct HashAndCharactersTranslator;
  56 struct HashAndUTF8CharactersTranslator;
  57 struct UCharBufferTranslator;
  58
  59 enum TextCaseSensitivity { TextCaseSensitive, TextCaseInsensitive };
  60
  61 typedef OwnFastMallocPtr<const UChar> SharableUChar;
  62 typedef CrossThreadRefCounted<SharableUChar> SharedUChar;
  63 typedef bool (*CharacterMatchFunctionPtr)(UChar);
  64
  65 class StringImpl : public StringImplBase {
  66     friend struct JSC::IdentifierCStringTranslator;
  67     friend struct JSC::IdentifierUCharBufferTranslator;
  68     friend struct WTF::CStringTranslator;
  69     friend struct WTF::HashAndCharactersTranslator;
  70     friend struct WTF::HashAndUTF8CharactersTranslator;
  71     friend struct WTF::UCharBufferTranslator;
  72     friend class AtomicStringImpl;
  73 private:
  74     // Used to construct static strings, which have an special refCount that can never hit zero.
  75     // This means that the static string will never be destroyed, which is important because
  76     // static strings will be shared across threads & ref-counted in a non-threadsafe manner.
  77     StringImpl(const UChar* characters, unsigned length, StaticStringConstructType)
  78         : StringImplBase(length, ConstructStaticString)
  79         , m_data(characters)
  80         , m_buffer(0)
  81         , m_hash(0)
  82     {
  83         // Ensure that the hash is computed so that AtomicStringHash can call existingHash()
  84         // with impunity. The empty string is special because it is never entered into
  85         // AtomicString's HashKey, but still needs to compare correctly.
  86         hash();
  87     }
  88
  89     // Create a normal string with internal storage (BufferInternal)
  90     StringImpl(unsigned length)
  91         : StringImplBase(length, BufferInternal)
  92         , m_data(reinterpret_cast<const UChar*>(this + 1))
  93         , m_buffer(0)
  94         , m_hash(0)
  95     {
  96         ASSERT(m_data);
  97         ASSERT(m_length);
  98     }
  99
 100     // Create a StringImpl adopting ownership of the provided buffer (BufferOwned)
 101     StringImpl(const UChar* characters, unsigned length)
 102         : StringImplBase(length, BufferOwned)
 103         , m_data(characters)
 104         , m_buffer(0)
 105         , m_hash(0)
 106     {
 107         ASSERT(m_data);
 108         ASSERT(m_length);
 109     }
 110
 111     // Used to create new strings that are a substring of an existing StringImpl (BufferSubstring)
 112     StringImpl(const UChar* characters, unsigned length, PassRefPtr<StringImpl> base)
 113         : StringImplBase(length, BufferSubstring)
 114         , m_data(characters)
 115         , m_substringBuffer(base.leakRef())
 116         , m_hash(0)
 117     {
 118         ASSERT(m_data);
 119         ASSERT(m_length);
 120         ASSERT(m_substringBuffer->bufferOwnership() != BufferSubstring);
 121     }
 122
 123     // Used to construct new strings sharing an existing SharedUChar (BufferShared)
 124     StringImpl(const UChar* characters, unsigned length, PassRefPtr<SharedUChar> sharedBuffer)
 125         : StringImplBase(length, BufferShared)
 126         , m_data(characters)
 127         , m_sharedBuffer(sharedBuffer.leakRef())
 128         , m_hash(0)
 129     {
 130         ASSERT(m_data);
 131         ASSERT(m_length);
 132     }
 133
 134     // For use only by AtomicString's XXXTranslator helpers.
 135     void setHash(unsigned hash)
 136     {
 137         ASSERT(!isStatic());
 138         ASSERT(!m_hash);
 139         ASSERT(hash == StringHasher::computeHash(m_data, m_length));
 140         m_hash = hash;
 141     }
 142
 143 public:
 144     ~StringImpl();
 145
 146     static PassRefPtr<StringImpl> create(const UChar*, unsigned length);
 147     static PassRefPtr<StringImpl> create(const char*, unsigned length);
 148     static PassRefPtr<StringImpl> create(const char*);
 149     static PassRefPtr<StringImpl> create(const UChar*, unsigned length, PassRefPtr<SharedUChar> sharedBuffer);
 150     static ALWAYS_INLINE PassRefPtr<StringImpl> create(PassRefPtr<StringImpl> rep, unsigned offset, unsigned length)
 151     {
 152         ASSERT(rep);
 153         ASSERT(length <= rep->length());
 154
 155         if (!length)
 156             return empty();
 157
 158         StringImpl* ownerRep = (rep->bufferOwnership() == BufferSubstring) ? rep->m_substringBuffer : rep.get();
 159         return adoptRef(new StringImpl(rep->m_data + offset, length, ownerRep));
 160     }
 161
 162     static PassRefPtr<StringImpl> createUninitialized(unsigned length, UChar*& data);
 163     static ALWAYS_INLINE PassRefPtr<StringImpl> tryCreateUninitialized(unsigned length, UChar*& output)
 164     {
 165         if (!length) {
 166             output = 0;
 167             return empty();
 168         }
 169
 170         if (length > ((std::numeric_limits<unsigned>::max() - sizeof(StringImpl)) / sizeof(UChar))) {
 171             output = 0;
 172             return 0;
 173         }
 174         StringImpl* resultImpl;
 175         if (!tryFastMalloc(sizeof(UChar) * length + sizeof(StringImpl)).getValue(resultImpl)) {
 176             output = 0;
 177             return 0;
 178         }
 179         output = reinterpret_cast<UChar*>(resultImpl + 1);
 180         return adoptRef(new(resultImpl) StringImpl(length));
 181     }
 182
 183     static unsigned dataOffset() { return OBJECT_OFFSETOF(StringImpl, m_data); }
 184     static PassRefPtr<StringImpl> createWithTerminatingNullCharacter(const StringImpl&);
 185     static PassRefPtr<StringImpl> createStrippingNullCharacters(const UChar*, unsigned length);
 186
 187     template<size_t inlineCapacity>
 188     static PassRefPtr<StringImpl> adopt(Vector<UChar, inlineCapacity>& vector)
 189     {
 190         if (size_t size = vector.size()) {
 191             ASSERT(vector.data());
 192             if (size > std::numeric_limits<unsigned>::max())
 193                 CRASH();
 194             return adoptRef(new StringImpl(vector.releaseBuffer(), size));
 195         }
 196         return empty();
 197     }
 198     static PassRefPtr<StringImpl> adopt(StringBuffer&);
 199
 200     SharedUChar* sharedBuffer();
 201     const UChar* characters() const { return m_data; }
 202
 203     size_t cost()
 204     {
 205         // For substrings, return the cost of the base string.
 206         if (bufferOwnership() == BufferSubstring)
 207             return m_substringBuffer->cost();
 208
 209         if (m_refCountAndFlags & s_refCountFlagShouldReportedCost) {
 210             m_refCountAndFlags &= ~s_refCountFlagShouldReportedCost;
 211             return m_length;
 212         }
 213         return 0;
 214     }
 215
 216     bool isIdentifier() const { return m_refCountAndFlags & s_refCountFlagIsIdentifier; }
 217     void setIsIdentifier(bool isIdentifier)
 218     {
 219         ASSERT(!isStatic());
 220         if (isIdentifier)
 221             m_refCountAndFlags |= s_refCountFlagIsIdentifier;
 222         else
 223             m_refCountAndFlags &= ~s_refCountFlagIsIdentifier;
 224     }
 225
 226     bool hasTerminatingNullCharacter() const { return m_refCountAndFlags & s_refCountFlagHasTerminatingNullCharacter; }
 227
 228     bool isAtomic() const { return m_refCountAndFlags & s_refCountFlagIsAtomic; }
 229     void setIsAtomic(bool isIdentifier)
 230     {
 231         ASSERT(!isStatic());
 232         if (isIdentifier)
 233             m_refCountAndFlags |= s_refCountFlagIsAtomic;
 234         else
 235             m_refCountAndFlags &= ~s_refCountFlagIsAtomic;
 236     }
 237
 238     unsigned hash() const { if (!m_hash) m_hash = StringHasher::computeHash(m_data, m_length); return m_hash; }
 239     unsigned existingHash() const { ASSERT(m_hash); return m_hash; }
 240
 241     ALWAYS_INLINE void deref() { m_refCountAndFlags -= s_refCountIncrement; if (!(m_refCountAndFlags & (s_refCountMask | s_refCountFlagStatic))) delete this; }
 242     ALWAYS_INLINE bool hasOneRef() const { return (m_refCountAndFlags & (s_refCountMask | s_refCountFlagStatic)) == s_refCountIncrement; }
 243
 244     static StringImpl* empty();
 245
 246     static void copyChars(UChar* destination, const UChar* source, unsigned numCharacters)
 247     {
 248         if (numCharacters <= s_copyCharsInlineCutOff) {
 249             for (unsigned i = 0; i < numCharacters; ++i)
 250                 destination[i] = source[i];
 251         } else
 252             memcpy(destination, source, numCharacters * sizeof(UChar));
 253     }
 254
 255     // Returns a StringImpl suitable for use on another thread.
 256     PassRefPtr<StringImpl> crossThreadString();
 257     // Makes a deep copy. Helpful only if you need to use a String on another thread
 258     // (use crossThreadString if the method call doesn't need to be threadsafe).
 259     // Since StringImpl objects are immutable, there's no other reason to make a copy.
 260     PassRefPtr<StringImpl> threadsafeCopy() const;
 261
 262     PassRefPtr<StringImpl> substring(unsigned pos, unsigned len = UINT_MAX);
 263
 264     UChar operator[](unsigned i) { ASSERT(i < m_length); return m_data[i]; }
 265     UChar32 characterStartingAt(unsigned);
 266
 267     bool containsOnlyWhitespace();
 268
 269     int toIntStrict(bool* ok = 0, int base = 10);
 270     unsigned toUIntStrict(bool* ok = 0, int base = 10);
 271     int64_t toInt64Strict(bool* ok = 0, int base = 10);
 272     uint64_t toUInt64Strict(bool* ok = 0, int base = 10);
 273     intptr_t toIntPtrStrict(bool* ok = 0, int base = 10);
 274
 275     int toInt(bool* ok = 0); // ignores trailing garbage
 276     unsigned toUInt(bool* ok = 0); // ignores trailing garbage
 277     int64_t toInt64(bool* ok = 0); // ignores trailing garbage
 278     uint64_t toUInt64(bool* ok = 0); // ignores trailing garbage
 279     intptr_t toIntPtr(bool* ok = 0); // ignores trailing garbage
 280
 281     double toDouble(bool* ok = 0, bool* didReadNumber = 0);
 282     float toFloat(bool* ok = 0, bool* didReadNumber = 0);
 283
 284     PassRefPtr<StringImpl> lower();
 285     PassRefPtr<StringImpl> upper();
 286
 287     enum LastCharacterBehavior { ObscureLastCharacter, DisplayLastCharacter };
 288
 289     PassRefPtr<StringImpl> secure(UChar, LastCharacterBehavior = ObscureLastCharacter);
 290     PassRefPtr<StringImpl> foldCase();
 291
 292     PassRefPtr<StringImpl> stripWhiteSpace();
 293     PassRefPtr<StringImpl> simplifyWhiteSpace();
 294
 295     PassRefPtr<StringImpl> removeCharacters(CharacterMatchFunctionPtr);
 296
 297     size_t find(UChar, unsigned index = 0);
 298     size_t find(CharacterMatchFunctionPtr, unsigned index = 0);
 299     size_t find(const char*, unsigned index = 0);
 300     size_t find(StringImpl*, unsigned index = 0);
 301     size_t findIgnoringCase(const char*, unsigned index = 0);
 302     size_t findIgnoringCase(StringImpl*, unsigned index = 0);
 303
 304     size_t reverseFind(UChar, unsigned index = UINT_MAX);
 305     size_t reverseFind(StringImpl*, unsigned index = UINT_MAX);
 306     size_t reverseFindIgnoringCase(StringImpl*, unsigned index = UINT_MAX);
 307
 308     bool startsWith(StringImpl* str, bool caseSensitive = true) { return (caseSensitive ? reverseFind(str, 0) : reverseFindIgnoringCase(str, 0)) == 0; }
 309     bool endsWith(StringImpl*, bool caseSensitive = true);
 310
 311     PassRefPtr<StringImpl> replace(UChar, UChar);
 312     PassRefPtr<StringImpl> replace(UChar, StringImpl*);
 313     PassRefPtr<StringImpl> replace(StringImpl*, StringImpl*);
 314     PassRefPtr<StringImpl> replace(unsigned index, unsigned len, StringImpl*);
 315
 316     int wordCount(int maxWordsToCount = INT_MAX);
 317
 318     WTF::Unicode::Direction defaultWritingDirection(bool* hasStrongDirectionality = 0);
 319
 320 #if USE(CF)
 321     CFStringRef createCFString();
 322 #endif
 323 #ifdef __OBJC__
 324     operator NSString*();
 325 #endif
 326
 327 private:
 328     // This number must be at least 2 to avoid sharing empty, null as well as 1 character strings from SmallStrings.
 329     static const unsigned s_copyCharsInlineCutOff = 20;
 330
 331     static PassRefPtr<StringImpl> createStrippingNullCharactersSlowCase(const UChar*, unsigned length);
 332
 333     BufferOwnership bufferOwnership() const { return static_cast<BufferOwnership>(m_refCountAndFlags & s_refCountMaskBufferOwnership); }
 334     bool isStatic() const { return m_refCountAndFlags & s_refCountFlagStatic; }
 335     const UChar* m_data;
 336     union {
 337         void* m_buffer;
 338         StringImpl* m_substringBuffer;
 339         SharedUChar* m_sharedBuffer;
 340     };
 341     mutable unsigned m_hash;
 342 };
 343
 344 bool equal(const StringImpl*, const StringImpl*);
 345 bool equal(const StringImpl*, const char*);
 346 inline bool equal(const char* a, StringImpl* b) { return equal(b, a); }
 347
 348 bool equalIgnoringCase(StringImpl*, StringImpl*);
 349 bool equalIgnoringCase(StringImpl*, const char*);
 350 inline bool equalIgnoringCase(const char* a, StringImpl* b) { return equalIgnoringCase(b, a); }
 351 bool equalIgnoringCase(const UChar* a, const char* b, unsigned length);
 352 inline bool equalIgnoringCase(const char* a, const UChar* b, unsigned length) { return equalIgnoringCase(b, a, length); }
 353
 354 bool equalIgnoringNullity(StringImpl*, StringImpl*);
 355
 356 template<size_t inlineCapacity>
 357 bool equalIgnoringNullity(const Vector<UChar, inlineCapacity>& a, StringImpl* b)
 358 {
 359     if (!b)
 360         return !a.size();
 361     if (a.size() != b->length())
 362         return false;
 363     return !memcmp(a.data(), b->characters(), b->length());
 364 }
 365
 366 int codePointCompare(const StringImpl*, const StringImpl*);
 367
 368 static inline bool isSpaceOrNewline(UChar c)
 369 {
 370     // Use isASCIISpace() for basic Latin-1.
 371     // This will include newlines, which aren't included in Unicode DirWS.
 372     return c <= 0x7F ? WTF::isASCIISpace(c) : WTF::Unicode::direction(c) == WTF::Unicode::WhiteSpaceNeutral;
 373 }
 374
 375 // This is a hot function because it's used when parsing HTML.
 376 inline PassRefPtr<StringImpl> StringImpl::createStrippingNullCharacters(const UChar* characters, unsigned length)
 377 {
 378     ASSERT(characters);
 379     ASSERT(length);
 380
 381     // Optimize for the case where there are no Null characters by quickly
 382     // searching for nulls, and then using StringImpl::create, which will
 383     // memcpy the whole buffer.  This is faster than assigning character by
 384     // character during the loop.
 385
 386     // Fast case.
 387     int foundNull = 0;
 388     for (unsigned i = 0; !foundNull && i < length; i++) {
 389         int c = characters[i]; // more efficient than using UChar here (at least on Intel Mac OS)
 390         foundNull |= !c;
 391     }
 392     if (!foundNull)
 393         return StringImpl::create(characters, length);
 394
 395     return StringImpl::createStrippingNullCharactersSlowCase(characters, length);
 396 }
 397
 398 struct StringHash;
 399
 400 // StringHash is the default hash for StringImpl* and RefPtr<StringImpl>
 401 template<typename T> struct DefaultHash;
 402 template<> struct DefaultHash<StringImpl*> {
 403     typedef StringHash Hash;
 404 };
 405 template<> struct DefaultHash<RefPtr<StringImpl> > {
 406     typedef StringHash Hash;
 407 };
 408
 409 }
 410
 411 using WTF::StringImpl;
 412 using WTF::equal;
 413 using WTF::TextCaseSensitivity;
 414 using WTF::TextCaseSensitive;
 415 using WTF::TextCaseInsensitive;
 416
 417 #endif