1/*
2 * Copyright (C) 1999 Lars Knoll (knoll@kde.org)
3 * Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2013 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/Forward.h"
29#include "wtf/HashMap.h"
30#include "wtf/StringHasher.h"
31#include "wtf/Vector.h"
32#include "wtf/WTFExport.h"
33#include "wtf/unicode/Unicode.h"
34
35#if USE(CF)
36typedef const struct __CFString * CFStringRef;
37#endif
38
39#ifdef __OBJC__
40@class NSString;
41#endif
42
43namespace WTF {
44
45struct AlreadyHashed;
46struct CStringTranslator;
47template<typename CharacterType> struct HashAndCharactersTranslator;
48struct HashAndUTF8CharactersTranslator;
49struct LCharBufferTranslator;
50struct CharBufferFromLiteralDataTranslator;
51struct SubstringTranslator;
52struct UCharBufferTranslator;
53template<typename> class RetainPtr;
54
55enum TextCaseSensitivity { TextCaseSensitive, TextCaseInsensitive };
56
57enum StripBehavior { StripExtraWhiteSpace, DoNotStripWhiteSpace };
58
59typedef bool (*CharacterMatchFunctionPtr)(UChar);
60typedef bool (*IsWhiteSpaceFunctionPtr)(UChar);
61typedef HashMap<unsigned, StringImpl*, AlreadyHashed> StaticStringsTable;
62
63// Define STRING_STATS to turn on run time statistics of string sizes and memory usage
64#undef STRING_STATS
65
66#ifdef STRING_STATS
67struct StringStats {
68    inline void add8BitString(unsigned length)
69    {
70        ++m_totalNumberStrings;
71        ++m_number8BitStrings;
72        m_total8BitData += length;
73    }
74
75    inline void add16BitString(unsigned length)
76    {
77        ++m_totalNumberStrings;
78        ++m_number16BitStrings;
79        m_total16BitData += length;
80    }
81
82    void removeString(StringImpl*);
83    void printStats();
84
85    static const unsigned s_printStringStatsFrequency = 5000;
86    static unsigned s_stringRemovesTillPrintStats;
87
88    unsigned m_totalNumberStrings;
89    unsigned m_number8BitStrings;
90    unsigned m_number16BitStrings;
91    unsigned long long m_total8BitData;
92    unsigned long long m_total16BitData;
93};
94
95void addStringForStats(StringImpl*);
96void removeStringForStats(StringImpl*);
97
98#define STRING_STATS_ADD_8BIT_STRING(length) StringImpl::stringStats().add8BitString(length); addStringForStats(this)
99#define STRING_STATS_ADD_16BIT_STRING(length) StringImpl::stringStats().add16BitString(length); addStringForStats(this)
100#define STRING_STATS_REMOVE_STRING(string) StringImpl::stringStats().removeString(string); removeStringForStats(this)
101#else
102#define STRING_STATS_ADD_8BIT_STRING(length) ((void)0)
103#define STRING_STATS_ADD_16BIT_STRING(length) ((void)0)
104#define STRING_STATS_REMOVE_STRING(string) ((void)0)
105#endif
106
107// You can find documentation about this class in this doc:
108// https://docs.google.com/document/d/1kOCUlJdh2WJMJGDf-WoEQhmnjKLaOYRbiHz5TiGJl14/edit?usp=sharing
109class WTF_EXPORT StringImpl {
110    WTF_MAKE_NONCOPYABLE(StringImpl);
111    friend struct WTF::CStringTranslator;
112    template<typename CharacterType> friend struct WTF::HashAndCharactersTranslator;
113    friend struct WTF::HashAndUTF8CharactersTranslator;
114    friend struct WTF::CharBufferFromLiteralDataTranslator;
115    friend struct WTF::LCharBufferTranslator;
116    friend struct WTF::SubstringTranslator;
117    friend struct WTF::UCharBufferTranslator;
118
119private:
120    // StringImpls are allocated out of the WTF buffer partition.
121    void* operator new(size_t);
122    void* operator new(size_t, void* ptr) { return ptr; };
123    void operator delete(void*);
124
125    // Used to construct static strings, which have an special refCount that can never hit zero.
126    // This means that the static string will never be destroyed, which is important because
127    // static strings will be shared across threads & ref-counted in a non-threadsafe manner.
128    enum ConstructEmptyStringTag { ConstructEmptyString };
129    explicit StringImpl(ConstructEmptyStringTag)
130        : m_refCount(1)
131        , m_length(0)
132        , m_hash(0)
133        , m_isAtomic(false)
134        , m_is8Bit(true)
135        , m_isStatic(true)
136    {
137        // Ensure that the hash is computed so that AtomicStringHash can call existingHash()
138        // with impunity. The empty string is special because it is never entered into
139        // AtomicString's HashKey, but still needs to compare correctly.
140        STRING_STATS_ADD_8BIT_STRING(m_length);
141        hash();
142    }
143
144    // FIXME: there has to be a less hacky way to do this.
145    enum Force8Bit { Force8BitConstructor };
146    StringImpl(unsigned length, Force8Bit)
147        : m_refCount(1)
148        , m_length(length)
149        , m_hash(0)
150        , m_isAtomic(false)
151        , m_is8Bit(true)
152        , m_isStatic(false)
153    {
154        ASSERT(m_length);
155        STRING_STATS_ADD_8BIT_STRING(m_length);
156    }
157
158    StringImpl(unsigned length)
159        : m_refCount(1)
160        , m_length(length)
161        , m_hash(0)
162        , m_isAtomic(false)
163        , m_is8Bit(false)
164        , m_isStatic(false)
165    {
166        ASSERT(m_length);
167        STRING_STATS_ADD_16BIT_STRING(m_length);
168    }
169
170    enum StaticStringTag { StaticString };
171    StringImpl(unsigned length, unsigned hash, StaticStringTag)
172        : m_refCount(1)
173        , m_length(length)
174        , m_hash(hash)
175        , m_isAtomic(false)
176        , m_is8Bit(true)
177        , m_isStatic(true)
178    {
179    }
180
181public:
182    ~StringImpl();
183
184    static StringImpl* createStatic(const char* string, unsigned length, unsigned hash);
185    static void freezeStaticStrings();
186    static const StaticStringsTable& allStaticStrings();
187    static unsigned highestStaticStringLength() { return m_highestStaticStringLength; }
188
189    static PassRefPtr<StringImpl> create(const UChar*, unsigned length);
190    static PassRefPtr<StringImpl> create(const LChar*, unsigned length);
191    static PassRefPtr<StringImpl> create8BitIfPossible(const UChar*, unsigned length);
192    template<size_t inlineCapacity>
193    static PassRefPtr<StringImpl> create8BitIfPossible(const Vector<UChar, inlineCapacity>& vector)
194    {
195        return create8BitIfPossible(vector.data(), vector.size());
196    }
197
198    ALWAYS_INLINE static PassRefPtr<StringImpl> create(const char* s, unsigned length) { return create(reinterpret_cast<const LChar*>(s), length); }
199    static PassRefPtr<StringImpl> create(const LChar*);
200    ALWAYS_INLINE static PassRefPtr<StringImpl> create(const char* s) { return create(reinterpret_cast<const LChar*>(s)); }
201
202    static PassRefPtr<StringImpl> createUninitialized(unsigned length, LChar*& data);
203    static PassRefPtr<StringImpl> createUninitialized(unsigned length, UChar*& data);
204
205    // Reallocate the StringImpl. The originalString must be only owned by the PassRefPtr.
206    // Just like the input pointer of realloc(), the originalString can't be used after this function.
207    static PassRefPtr<StringImpl> reallocate(PassRefPtr<StringImpl> originalString, unsigned length, LChar*& data);
208    static PassRefPtr<StringImpl> reallocate(PassRefPtr<StringImpl> originalString, unsigned length, UChar*& data);
209
210    // If this StringImpl has only one reference, we can truncate the string by updating
211    // its m_length property without actually re-allocating its buffer.
212    void truncateAssumingIsolated(unsigned length)
213    {
214        ASSERT(hasOneRef());
215        ASSERT(length <= m_length);
216        m_length = length;
217    }
218
219    unsigned length() const { return m_length; }
220    bool is8Bit() const { return m_is8Bit; }
221
222    ALWAYS_INLINE const LChar* characters8() const { ASSERT(is8Bit()); return reinterpret_cast<const LChar*>(this + 1); }
223    ALWAYS_INLINE const UChar* characters16() const { ASSERT(!is8Bit()); return reinterpret_cast<const UChar*>(this + 1); }
224
225    template <typename CharType>
226    ALWAYS_INLINE const CharType * getCharacters() const;
227
228    size_t sizeInBytes() const;
229
230    bool isAtomic() const { return m_isAtomic; }
231    void setIsAtomic(bool isAtomic) { m_isAtomic = isAtomic; }
232
233    bool isStatic() const { return m_isStatic; }
234
235private:
236    // The high bits of 'hash' are always empty, but we prefer to store our flags
237    // in the low bits because it makes them slightly more efficient to access.
238    // So, we shift left and right when setting and getting our hash code.
239    void setHash(unsigned hash) const
240    {
241        ASSERT(!hasHash());
242        // Multiple clients assume that StringHasher is the canonical string hash function.
243        ASSERT(hash == (is8Bit() ? StringHasher::computeHashAndMaskTop8Bits(characters8(), m_length) : StringHasher::computeHashAndMaskTop8Bits(characters16(), m_length)));
244        m_hash = hash;
245        ASSERT(hash); // Verify that 0 is a valid sentinel hash value.
246    }
247
248    unsigned rawHash() const
249    {
250        return m_hash;
251    }
252
253    void destroyIfNotStatic();
254
255public:
256    bool hasHash() const
257    {
258        return rawHash() != 0;
259    }
260
261    unsigned existingHash() const
262    {
263        ASSERT(hasHash());
264        return rawHash();
265    }
266
267    unsigned hash() const
268    {
269        if (hasHash())
270            return existingHash();
271        return hashSlowCase();
272    }
273
274    inline bool hasOneRef() const
275    {
276        return m_refCount == 1;
277    }
278
279    inline void ref()
280    {
281        ++m_refCount;
282    }
283
284    inline void deref()
285    {
286        if (hasOneRef()) {
287            destroyIfNotStatic();
288            return;
289        }
290
291        --m_refCount;
292    }
293
294    static StringImpl* empty();
295
296    // FIXME: Does this really belong in StringImpl?
297    template <typename T> static void copyChars(T* destination, const T* source, unsigned numCharacters)
298    {
299        if (numCharacters == 1) {
300            *destination = *source;
301            return;
302        }
303
304        // FIXME: Is this implementation really faster than memcpy?
305        if (numCharacters <= s_copyCharsInlineCutOff) {
306            unsigned i = 0;
307#if (CPU(X86) || CPU(X86_64))
308            const unsigned charsPerInt = sizeof(uint32_t) / sizeof(T);
309
310            if (numCharacters > charsPerInt) {
311                unsigned stopCount = numCharacters & ~(charsPerInt - 1);
312
313                const uint32_t* srcCharacters = reinterpret_cast<const uint32_t*>(source);
314                uint32_t* destCharacters = reinterpret_cast<uint32_t*>(destination);
315                for (unsigned j = 0; i < stopCount; i += charsPerInt, ++j)
316                    destCharacters[j] = srcCharacters[j];
317            }
318#endif
319            for (; i < numCharacters; ++i)
320                destination[i] = source[i];
321        } else
322            memcpy(destination, source, numCharacters * sizeof(T));
323    }
324
325    ALWAYS_INLINE static void copyChars(UChar* destination, const LChar* source, unsigned numCharacters)
326    {
327        for (unsigned i = 0; i < numCharacters; ++i)
328            destination[i] = source[i];
329    }
330
331    // Some string features, like refcounting and the atomicity flag, are not
332    // thread-safe. We achieve thread safety by isolation, giving each thread
333    // its own copy of the string.
334    PassRefPtr<StringImpl> isolatedCopy() const;
335
336    PassRefPtr<StringImpl> substring(unsigned pos, unsigned len = UINT_MAX);
337
338    UChar operator[](unsigned i) const
339    {
340        ASSERT_WITH_SECURITY_IMPLICATION(i < m_length);
341        if (is8Bit())
342            return characters8()[i];
343        return characters16()[i];
344    }
345    UChar32 characterStartingAt(unsigned);
346
347    bool containsOnlyWhitespace();
348
349    int toIntStrict(bool* ok = 0, int base = 10);
350    unsigned toUIntStrict(bool* ok = 0, int base = 10);
351    int64_t toInt64Strict(bool* ok = 0, int base = 10);
352    uint64_t toUInt64Strict(bool* ok = 0, int base = 10);
353    intptr_t toIntPtrStrict(bool* ok = 0, int base = 10);
354
355    int toInt(bool* ok = 0); // ignores trailing garbage
356    unsigned toUInt(bool* ok = 0); // ignores trailing garbage
357    int64_t toInt64(bool* ok = 0); // ignores trailing garbage
358    uint64_t toUInt64(bool* ok = 0); // ignores trailing garbage
359    intptr_t toIntPtr(bool* ok = 0); // ignores trailing garbage
360
361    // FIXME: Like the strict functions above, these give false for "ok" when there is trailing garbage.
362    // Like the non-strict functions above, these return the value when there is trailing garbage.
363    // It would be better if these were more consistent with the above functions instead.
364    double toDouble(bool* ok = 0);
365    float toFloat(bool* ok = 0);
366
367    PassRefPtr<StringImpl> lower();
368    PassRefPtr<StringImpl> upper();
369    PassRefPtr<StringImpl> lower(const AtomicString& localeIdentifier);
370    PassRefPtr<StringImpl> upper(const AtomicString& localeIdentifier);
371
372    PassRefPtr<StringImpl> fill(UChar);
373    // FIXME: Do we need fill(char) or can we just do the right thing if UChar is ASCII?
374    PassRefPtr<StringImpl> foldCase();
375
376    PassRefPtr<StringImpl> stripWhiteSpace();
377    PassRefPtr<StringImpl> stripWhiteSpace(IsWhiteSpaceFunctionPtr);
378    PassRefPtr<StringImpl> simplifyWhiteSpace(StripBehavior stripBehavior = StripExtraWhiteSpace);
379    PassRefPtr<StringImpl> simplifyWhiteSpace(IsWhiteSpaceFunctionPtr, StripBehavior stripBehavior = StripExtraWhiteSpace);
380
381    PassRefPtr<StringImpl> removeCharacters(CharacterMatchFunctionPtr);
382    template <typename CharType>
383    ALWAYS_INLINE PassRefPtr<StringImpl> removeCharacters(const CharType* characters, CharacterMatchFunctionPtr);
384
385    size_t find(LChar character, unsigned start = 0);
386    size_t find(char character, unsigned start = 0);
387    size_t find(UChar character, unsigned start = 0);
388    size_t find(CharacterMatchFunctionPtr, unsigned index = 0);
389    size_t find(const LChar*, unsigned index = 0);
390    ALWAYS_INLINE size_t find(const char* s, unsigned index = 0) { return find(reinterpret_cast<const LChar*>(s), index); }
391    size_t find(StringImpl*);
392    size_t find(StringImpl*, unsigned index);
393    size_t findIgnoringCase(const LChar*, unsigned index = 0);
394    ALWAYS_INLINE size_t findIgnoringCase(const char* s, unsigned index = 0) { return findIgnoringCase(reinterpret_cast<const LChar*>(s), index); }
395    size_t findIgnoringCase(StringImpl*, unsigned index = 0);
396
397    size_t findNextLineStart(unsigned index = UINT_MAX);
398
399    size_t reverseFind(UChar, unsigned index = UINT_MAX);
400    size_t reverseFind(StringImpl*, unsigned index = UINT_MAX);
401    size_t reverseFindIgnoringCase(StringImpl*, unsigned index = UINT_MAX);
402
403    size_t count(LChar) const;
404
405    bool startsWith(StringImpl* str, bool caseSensitive = true) { return (caseSensitive ? reverseFind(str, 0) : reverseFindIgnoringCase(str, 0)) == 0; }
406    bool startsWith(UChar) const;
407    bool startsWith(const char*, unsigned matchLength, bool caseSensitive) const;
408    template<unsigned matchLength>
409    bool startsWith(const char (&prefix)[matchLength], bool caseSensitive = true) const { return startsWith(prefix, matchLength - 1, caseSensitive); }
410
411    bool endsWith(StringImpl*, bool caseSensitive = true);
412    bool endsWith(UChar) const;
413    bool endsWith(const char*, unsigned matchLength, bool caseSensitive) const;
414    template<unsigned matchLength>
415    bool endsWith(const char (&prefix)[matchLength], bool caseSensitive = true) const { return endsWith(prefix, matchLength - 1, caseSensitive); }
416
417    PassRefPtr<StringImpl> replace(UChar, UChar);
418    PassRefPtr<StringImpl> replace(UChar, StringImpl*);
419    ALWAYS_INLINE PassRefPtr<StringImpl> replace(UChar pattern, const char* replacement, unsigned replacementLength) { return replace(pattern, reinterpret_cast<const LChar*>(replacement), replacementLength); }
420    PassRefPtr<StringImpl> replace(UChar, const LChar*, unsigned replacementLength);
421    PassRefPtr<StringImpl> replace(UChar, const UChar*, unsigned replacementLength);
422    PassRefPtr<StringImpl> replace(StringImpl*, StringImpl*);
423    PassRefPtr<StringImpl> replace(unsigned index, unsigned len, StringImpl*);
424    PassRefPtr<StringImpl> upconvertedString();
425
426#if USE(CF)
427    RetainPtr<CFStringRef> createCFString();
428#endif
429#ifdef __OBJC__
430    operator NSString*();
431#endif
432
433#ifdef STRING_STATS
434    ALWAYS_INLINE static StringStats& stringStats() { return m_stringStats; }
435#endif
436
437private:
438    // This number must be at least 2 to avoid sharing empty, null as well as 1 character strings from SmallStrings.
439    static const unsigned s_copyCharsInlineCutOff = 20;
440
441    template <class UCharPredicate> PassRefPtr<StringImpl> stripMatchedCharacters(UCharPredicate);
442    template <typename CharType, class UCharPredicate> PassRefPtr<StringImpl> simplifyMatchedCharactersToSpace(UCharPredicate, StripBehavior);
443    NEVER_INLINE unsigned hashSlowCase() const;
444
445#ifdef STRING_STATS
446    static StringStats m_stringStats;
447#endif
448
449    static unsigned m_highestStaticStringLength;
450
451#ifndef NDEBUG
452    void assertHashIsCorrect()
453    {
454        ASSERT(hasHash());
455        ASSERT(existingHash() == StringHasher::computeHashAndMaskTop8Bits(characters8(), length()));
456    }
457#endif
458
459private:
460    unsigned m_refCount;
461    unsigned m_length;
462    mutable unsigned m_hash : 24;
463    unsigned m_isAtomic : 1;
464    unsigned m_is8Bit : 1;
465    unsigned m_isStatic : 1;
466};
467
468template <>
469ALWAYS_INLINE const LChar* StringImpl::getCharacters<LChar>() const { return characters8(); }
470
471template <>
472ALWAYS_INLINE const UChar* StringImpl::getCharacters<UChar>() const { return characters16(); }
473
474WTF_EXPORT bool equal(const StringImpl*, const StringImpl*);
475WTF_EXPORT bool equal(const StringImpl*, const LChar*);
476inline bool equal(const StringImpl* a, const char* b) { return equal(a, reinterpret_cast<const LChar*>(b)); }
477WTF_EXPORT bool equal(const StringImpl*, const LChar*, unsigned);
478WTF_EXPORT bool equal(const StringImpl*, const UChar*, unsigned);
479inline bool equal(const StringImpl* a, const char* b, unsigned length) { return equal(a, reinterpret_cast<const LChar*>(b), length); }
480inline bool equal(const LChar* a, StringImpl* b) { return equal(b, a); }
481inline bool equal(const char* a, StringImpl* b) { return equal(b, reinterpret_cast<const LChar*>(a)); }
482WTF_EXPORT bool equalNonNull(const StringImpl* a, const StringImpl* b);
483
484template<typename CharType>
485ALWAYS_INLINE bool equal(const CharType* a, const CharType* b, unsigned length) { return !memcmp(a, b, length * sizeof(CharType)); }
486
487ALWAYS_INLINE bool equal(const LChar* a, const UChar* b, unsigned length)
488{
489    for (unsigned i = 0; i < length; ++i) {
490        if (a[i] != b[i])
491            return false;
492    }
493    return true;
494}
495
496ALWAYS_INLINE bool equal(const UChar* a, const LChar* b, unsigned length) { return equal(b, a, length); }
497
498WTF_EXPORT bool equalIgnoringCase(const StringImpl*, const StringImpl*);
499WTF_EXPORT bool equalIgnoringCase(const StringImpl*, const LChar*);
500inline bool equalIgnoringCase(const LChar* a, const StringImpl* b) { return equalIgnoringCase(b, a); }
501WTF_EXPORT bool equalIgnoringCase(const LChar*, const LChar*, unsigned);
502WTF_EXPORT bool equalIgnoringCase(const UChar*, const LChar*, unsigned);
503inline bool equalIgnoringCase(const UChar* a, const char* b, unsigned length) { return equalIgnoringCase(a, reinterpret_cast<const LChar*>(b), length); }
504inline bool equalIgnoringCase(const LChar* a, const UChar* b, unsigned length) { return equalIgnoringCase(b, a, length); }
505inline bool equalIgnoringCase(const char* a, const UChar* b, unsigned length) { return equalIgnoringCase(b, reinterpret_cast<const LChar*>(a), length); }
506inline bool equalIgnoringCase(const char* a, const LChar* b, unsigned length) { return equalIgnoringCase(b, reinterpret_cast<const LChar*>(a), length); }
507inline bool equalIgnoringCase(const UChar* a, const UChar* b, int length)
508{
509    ASSERT(length >= 0);
510    return !Unicode::umemcasecmp(a, b, length);
511}
512WTF_EXPORT bool equalIgnoringCaseNonNull(const StringImpl*, const StringImpl*);
513
514WTF_EXPORT bool equalIgnoringNullity(StringImpl*, StringImpl*);
515
516template<typename CharacterType>
517inline size_t find(const CharacterType* characters, unsigned length, CharacterType matchCharacter, unsigned index = 0)
518{
519    while (index < length) {
520        if (characters[index] == matchCharacter)
521            return index;
522        ++index;
523    }
524    return kNotFound;
525}
526
527ALWAYS_INLINE size_t find(const UChar* characters, unsigned length, LChar matchCharacter, unsigned index = 0)
528{
529    return find(characters, length, static_cast<UChar>(matchCharacter), index);
530}
531
532inline size_t find(const LChar* characters, unsigned length, UChar matchCharacter, unsigned index = 0)
533{
534    if (matchCharacter & ~0xFF)
535        return kNotFound;
536    return find(characters, length, static_cast<LChar>(matchCharacter), index);
537}
538
539inline size_t find(const LChar* characters, unsigned length, CharacterMatchFunctionPtr matchFunction, unsigned index = 0)
540{
541    while (index < length) {
542        if (matchFunction(characters[index]))
543            return index;
544        ++index;
545    }
546    return kNotFound;
547}
548
549inline size_t find(const UChar* characters, unsigned length, CharacterMatchFunctionPtr matchFunction, unsigned index = 0)
550{
551    while (index < length) {
552        if (matchFunction(characters[index]))
553            return index;
554        ++index;
555    }
556    return kNotFound;
557}
558
559template<typename CharacterType>
560inline size_t findNextLineStart(const CharacterType* characters, unsigned length, unsigned index = 0)
561{
562    while (index < length) {
563        CharacterType c = characters[index++];
564        if ((c != '\n') && (c != '\r'))
565            continue;
566
567        // There can only be a start of a new line if there are more characters
568        // beyond the current character.
569        if (index < length) {
570            // The 3 common types of line terminators are 1. \r\n (Windows),
571            // 2. \r (old MacOS) and 3. \n (Unix'es).
572
573            if (c == '\n')
574                return index; // Case 3: just \n.
575
576            CharacterType c2 = characters[index];
577            if (c2 != '\n')
578                return index; // Case 2: just \r.
579
580            // Case 1: \r\n.
581            // But, there's only a start of a new line if there are more
582            // characters beyond the \r\n.
583            if (++index < length)
584                return index;
585        }
586    }
587    return kNotFound;
588}
589
590template<typename CharacterType>
591inline size_t reverseFindLineTerminator(const CharacterType* characters, unsigned length, unsigned index = UINT_MAX)
592{
593    if (!length)
594        return kNotFound;
595    if (index >= length)
596        index = length - 1;
597    CharacterType c = characters[index];
598    while ((c != '\n') && (c != '\r')) {
599        if (!index--)
600            return kNotFound;
601        c = characters[index];
602    }
603    return index;
604}
605
606template<typename CharacterType>
607inline size_t reverseFind(const CharacterType* characters, unsigned length, CharacterType matchCharacter, unsigned index = UINT_MAX)
608{
609    if (!length)
610        return kNotFound;
611    if (index >= length)
612        index = length - 1;
613    while (characters[index] != matchCharacter) {
614        if (!index--)
615            return kNotFound;
616    }
617    return index;
618}
619
620ALWAYS_INLINE size_t reverseFind(const UChar* characters, unsigned length, LChar matchCharacter, unsigned index = UINT_MAX)
621{
622    return reverseFind(characters, length, static_cast<UChar>(matchCharacter), index);
623}
624
625inline size_t reverseFind(const LChar* characters, unsigned length, UChar matchCharacter, unsigned index = UINT_MAX)
626{
627    if (matchCharacter & ~0xFF)
628        return kNotFound;
629    return reverseFind(characters, length, static_cast<LChar>(matchCharacter), index);
630}
631
632inline size_t StringImpl::find(LChar character, unsigned start)
633{
634    if (is8Bit())
635        return WTF::find(characters8(), m_length, character, start);
636    return WTF::find(characters16(), m_length, character, start);
637}
638
639ALWAYS_INLINE size_t StringImpl::find(char character, unsigned start)
640{
641    return find(static_cast<LChar>(character), start);
642}
643
644inline size_t StringImpl::find(UChar character, unsigned start)
645{
646    if (is8Bit())
647        return WTF::find(characters8(), m_length, character, start);
648    return WTF::find(characters16(), m_length, character, start);
649}
650
651inline unsigned lengthOfNullTerminatedString(const UChar* string)
652{
653    size_t length = 0;
654    while (string[length] != UChar(0))
655        ++length;
656    RELEASE_ASSERT(length <= std::numeric_limits<unsigned>::max());
657    return static_cast<unsigned>(length);
658}
659
660template<size_t inlineCapacity>
661bool equalIgnoringNullity(const Vector<UChar, inlineCapacity>& a, StringImpl* b)
662{
663    if (!b)
664        return !a.size();
665    if (a.size() != b->length())
666        return false;
667    if (b->is8Bit())
668        return equal(a.data(), b->characters8(), b->length());
669    return equal(a.data(), b->characters16(), b->length());
670}
671
672template<typename CharacterType1, typename CharacterType2>
673static inline int codePointCompare(unsigned l1, unsigned l2, const CharacterType1* c1, const CharacterType2* c2)
674{
675    const unsigned lmin = l1 < l2 ? l1 : l2;
676    unsigned pos = 0;
677    while (pos < lmin && *c1 == *c2) {
678        ++c1;
679        ++c2;
680        ++pos;
681    }
682
683    if (pos < lmin)
684        return (c1[0] > c2[0]) ? 1 : -1;
685
686    if (l1 == l2)
687        return 0;
688
689    return (l1 > l2) ? 1 : -1;
690}
691
692static inline int codePointCompare8(const StringImpl* string1, const StringImpl* string2)
693{
694    return codePointCompare(string1->length(), string2->length(), string1->characters8(), string2->characters8());
695}
696
697static inline int codePointCompare16(const StringImpl* string1, const StringImpl* string2)
698{
699    return codePointCompare(string1->length(), string2->length(), string1->characters16(), string2->characters16());
700}
701
702static inline int codePointCompare8To16(const StringImpl* string1, const StringImpl* string2)
703{
704    return codePointCompare(string1->length(), string2->length(), string1->characters8(), string2->characters16());
705}
706
707static inline int codePointCompare(const StringImpl* string1, const StringImpl* string2)
708{
709    if (!string1)
710        return (string2 && string2->length()) ? -1 : 0;
711
712    if (!string2)
713        return string1->length() ? 1 : 0;
714
715    bool string1Is8Bit = string1->is8Bit();
716    bool string2Is8Bit = string2->is8Bit();
717    if (string1Is8Bit) {
718        if (string2Is8Bit)
719            return codePointCompare8(string1, string2);
720        return codePointCompare8To16(string1, string2);
721    }
722    if (string2Is8Bit)
723        return -codePointCompare8To16(string2, string1);
724    return codePointCompare16(string1, string2);
725}
726
727static inline bool isSpaceOrNewline(UChar c)
728{
729    // Use isASCIISpace() for basic Latin-1.
730    // This will include newlines, which aren't included in Unicode DirWS.
731    return c <= 0x7F ? WTF::isASCIISpace(c) : WTF::Unicode::direction(c) == WTF::Unicode::WhiteSpaceNeutral;
732}
733
734inline PassRefPtr<StringImpl> StringImpl::isolatedCopy() const
735{
736    if (is8Bit())
737        return create(characters8(), m_length);
738    return create(characters16(), m_length);
739}
740
741struct StringHash;
742
743// StringHash is the default hash for StringImpl* and RefPtr<StringImpl>
744template<typename T> struct DefaultHash;
745template<> struct DefaultHash<StringImpl*> {
746    typedef StringHash Hash;
747};
748template<> struct DefaultHash<RefPtr<StringImpl> > {
749    typedef StringHash Hash;
750};
751
752}
753
754using WTF::StringImpl;
755using WTF::equal;
756using WTF::equalNonNull;
757using WTF::TextCaseSensitivity;
758using WTF::TextCaseSensitive;
759using WTF::TextCaseInsensitive;
760
761#endif
762