/*
* Copyright (C) 1999-2000 Harri Porten (porten@kde.org)
- * Copyright (C) 2004, 2005, 2006, 2007, 2008 Apple Inc. All rights reserved.
+ * Copyright (C) 2004, 2005, 2006, 2007, 2008, 2009 Apple Inc. All rights reserved.
* Copyright (C) 2007 Cameron Zwarich (cwzwarich@uwaterloo.ca)
- * Copyright (c) 2009, Google Inc. All rights reserved.
+ * Copyright (C) 2009 Google Inc. All rights reserved.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
#include "UString.h"
#include "JSGlobalObjectFunctions.h"
-#include "Collector.h"
-#include "dtoa.h"
+#include "Heap.h"
#include "Identifier.h"
#include "Operations.h"
#include <ctype.h>
-#include <float.h>
#include <limits.h>
-#include <math.h>
+#include <limits>
#include <stdio.h>
#include <stdlib.h>
#include <wtf/ASCIICType.h>
#include <wtf/Assertions.h>
#include <wtf/MathExtras.h>
+#include <wtf/StringExtras.h>
#include <wtf/Vector.h>
+#include <wtf/dtoa.h>
#include <wtf/unicode/UTF8.h>
-#if HAVE(STRING_H)
-#include <string.h>
-#endif
#if HAVE(STRINGS_H)
#include <strings.h>
#endif
using namespace WTF::Unicode;
using namespace std;
-// This can be tuned differently per platform by putting platform #ifs right here.
-// If you don't define this macro at all, then copyChars will just call directly
-// to memcpy.
-#define USTRING_COPY_CHARS_INLINE_CUTOFF 20
-
namespace JSC {
-
-extern const double NaN;
-extern const double Inf;
-static inline size_t overflowIndicator() { return std::numeric_limits<size_t>::max(); }
-static inline size_t maxUChars() { return std::numeric_limits<size_t>::max() / sizeof(UChar); }
+COMPILE_ASSERT(sizeof(UString) == sizeof(void*), UString_should_stay_small);
-static inline UChar* allocChars(size_t length)
+// Construct a string with UTF-16 data.
+UString::UString(const UChar* characters, unsigned length)
+ : m_impl(characters ? StringImpl::create(characters, length) : 0)
{
- ASSERT(length);
- if (length > maxUChars())
- return 0;
- return static_cast<UChar*>(tryFastMalloc(sizeof(UChar) * length));
}
-static inline UChar* reallocChars(UChar* buffer, size_t length)
+// Construct a string with UTF-16 data, from a null-terminated source.
+UString::UString(const UChar* characters)
{
- ASSERT(length);
- if (length > maxUChars())
- return 0;
- return static_cast<UChar*>(tryFastRealloc(buffer, sizeof(UChar) * length));
-}
-
-static inline void copyChars(UChar* destination, const UChar* source, unsigned numCharacters)
-{
-#ifdef USTRING_COPY_CHARS_INLINE_CUTOFF
- if (numCharacters <= USTRING_COPY_CHARS_INLINE_CUTOFF) {
- for (unsigned i = 0; i < numCharacters; ++i)
- destination[i] = source[i];
+ if (!characters)
return;
- }
-#endif
- memcpy(destination, source, numCharacters * sizeof(UChar));
-}
-
-COMPILE_ASSERT(sizeof(UChar) == 2, uchar_is_2_bytes)
-
-CString::CString(const char* c)
- : m_length(strlen(c))
- , m_data(new char[m_length + 1])
-{
- memcpy(m_data, c, m_length + 1);
-}
-
-CString::CString(const char* c, size_t length)
- : m_length(length)
- , m_data(new char[length + 1])
-{
- memcpy(m_data, c, m_length);
- m_data[m_length] = 0;
-}
-
-CString::CString(const CString& b)
-{
- m_length = b.m_length;
- if (b.m_data) {
- m_data = new char[m_length + 1];
- memcpy(m_data, b.m_data, m_length + 1);
- } else
- m_data = 0;
-}
-
-CString::~CString()
-{
- delete [] m_data;
-}
-
-CString CString::adopt(char* c, size_t length)
-{
- CString s;
- s.m_data = c;
- s.m_length = length;
- return s;
-}
-
-CString& CString::append(const CString& t)
-{
- char* n;
- n = new char[m_length + t.m_length + 1];
- if (m_length)
- memcpy(n, m_data, m_length);
- if (t.m_length)
- memcpy(n + m_length, t.m_data, t.m_length);
- m_length += t.m_length;
- n[m_length] = 0;
-
- delete [] m_data;
- m_data = n;
-
- return *this;
-}
-
-CString& CString::operator=(const char* c)
-{
- if (m_data)
- delete [] m_data;
- m_length = strlen(c);
- m_data = new char[m_length + 1];
- memcpy(m_data, c, m_length + 1);
-
- return *this;
-}
-
-CString& CString::operator=(const CString& str)
-{
- if (this == &str)
- return *this;
-
- if (m_data)
- delete [] m_data;
- m_length = str.m_length;
- if (str.m_data) {
- m_data = new char[m_length + 1];
- memcpy(m_data, str.m_data, m_length + 1);
- } else
- m_data = 0;
-
- return *this;
-}
-
-bool operator==(const CString& c1, const CString& c2)
-{
- size_t len = c1.size();
- return len == c2.size() && (len == 0 || memcmp(c1.c_str(), c2.c_str(), len) == 0);
-}
-
-// These static strings are immutable, except for rc, whose initial value is chosen to
-// reduce the possibility of it becoming zero due to ref/deref not being thread-safe.
-static UChar sharedEmptyChar;
-UString::BaseString* UString::Rep::nullBaseString;
-UString::BaseString* UString::Rep::emptyBaseString;
-UString* UString::nullUString;
-
-static void initializeStaticBaseString(int len, UChar* buf, UString::BaseString& base)
-{
- base.offset = 0;
- base.len = len;
- base.rc = INT_MAX / 2;
- base._hash = 0;
- base.m_identifierTableAndFlags.setFlag(UString::Rep::StaticFlag);
- base.m_baseString = 0;
- base.buf = buf;
- base.preCapacity = 0;
- base.usedPreCapacity = 0;
- base.capacity = 0;
- base.usedCapacity = 0;
- base.reportedCost = 0;
- base.checkConsistency();
-}
-
-void initializeUString()
-{
- UString::Rep::nullBaseString = new UString::BaseString;
- initializeStaticBaseString(0, 0, *UString::Rep::nullBaseString);
-
- UString::Rep::emptyBaseString = new UString::BaseString;
- initializeStaticBaseString(0, &sharedEmptyChar, *UString::Rep::emptyBaseString);
-
- UString::nullUString = new UString;
-}
-
-static char* statBuffer = 0; // Only used for debugging via UString::ascii().
-
-PassRefPtr<UString::Rep> UString::Rep::createCopying(const UChar* d, int l)
-{
- UChar* copyD = static_cast<UChar*>(fastMalloc(l * sizeof(UChar)));
- copyChars(copyD, d, l);
- return create(copyD, l);
-}
-
-PassRefPtr<UString::Rep> UString::Rep::create(UChar* d, int l)
-{
- BaseString* r = new BaseString;
- r->offset = 0;
- r->len = l;
- r->rc = 1;
- r->_hash = 0;
- r->m_baseString = 0;
- r->reportedCost = 0;
- r->buf = d;
- r->usedCapacity = l;
- r->capacity = l;
- r->usedPreCapacity = 0;
- r->preCapacity = 0;
-
- r->checkConsistency();
-
- // steal the single reference this Rep was created with
- return adoptRef(r);
-}
-
-PassRefPtr<UString::Rep> UString::Rep::create(PassRefPtr<Rep> rep, int offset, int length)
-{
- ASSERT(rep);
- rep->checkConsistency();
-
- int repOffset = rep->offset;
-
- PassRefPtr<BaseString> base = rep->baseString();
-
- ASSERT(-(offset + repOffset) <= base->usedPreCapacity);
- ASSERT(offset + repOffset + length <= base->usedCapacity);
-
- Rep* r = new Rep;
- r->offset = repOffset + offset;
- r->len = length;
- r->rc = 1;
- r->_hash = 0;
- r->setBaseString(base);
-
- r->checkConsistency();
-
- // steal the single reference this Rep was created with
- return adoptRef(r);
-}
-
-PassRefPtr<UString::Rep> UString::Rep::createFromUTF8(const char* string)
-{
- if (!string)
- return &UString::Rep::null();
-
- size_t length = strlen(string);
- Vector<UChar, 1024> buffer(length);
- UChar* p = buffer.data();
- if (conversionOK != convertUTF8ToUTF16(&string, string + length, &p, p + length))
- return &UString::Rep::null();
-
- return UString::Rep::createCopying(buffer.data(), p - buffer.data());
-}
-
-void UString::Rep::destroy()
-{
- checkConsistency();
-
- // Static null and empty strings can never be destroyed, but we cannot rely on
- // reference counting, because ref/deref are not thread-safe.
- if (!isStatic()) {
- if (identifierTable())
- Identifier::remove(this);
- UString::BaseString* base = baseString();
- if (base == this)
- fastFree(base->buf);
- else
- base->deref();
-
- delete this;
- }
-}
-
-// Golden ratio - arbitrary start value to avoid mapping all 0's to all 0's
-// or anything like that.
-const unsigned PHI = 0x9e3779b9U;
-
-// Paul Hsieh's SuperFastHash
-// http://www.azillionmonkeys.com/qed/hash.html
-unsigned UString::Rep::computeHash(const UChar* s, int len)
-{
- unsigned l = len;
- uint32_t hash = PHI;
- uint32_t tmp;
-
- int rem = l & 1;
- l >>= 1;
-
- // Main loop
- for (; l > 0; l--) {
- hash += s[0];
- tmp = (s[1] << 11) ^ hash;
- hash = (hash << 16) ^ tmp;
- s += 2;
- hash += hash >> 11;
- }
-
- // Handle end case
- if (rem) {
- hash += s[0];
- hash ^= hash << 11;
- hash += hash >> 17;
- }
-
- // Force "avalanching" of final 127 bits
- hash ^= hash << 3;
- hash += hash >> 5;
- hash ^= hash << 2;
- hash += hash >> 15;
- hash ^= hash << 10;
-
- // this avoids ever returning a hash code of 0, since that is used to
- // signal "hash not computed yet", using a value that is likely to be
- // effectively the same as 0 when the low bits are masked
- if (hash == 0)
- hash = 0x80000000;
-
- return hash;
-}
-
-// Paul Hsieh's SuperFastHash
-// http://www.azillionmonkeys.com/qed/hash.html
-unsigned UString::Rep::computeHash(const char* s, int l)
-{
- // This hash is designed to work on 16-bit chunks at a time. But since the normal case
- // (above) is to hash UTF-16 characters, we just treat the 8-bit chars as if they
- // were 16-bit chunks, which should give matching results
-
- uint32_t hash = PHI;
- uint32_t tmp;
-
- size_t rem = l & 1;
- l >>= 1;
-
- // Main loop
- for (; l > 0; l--) {
- hash += static_cast<unsigned char>(s[0]);
- tmp = (static_cast<unsigned char>(s[1]) << 11) ^ hash;
- hash = (hash << 16) ^ tmp;
- s += 2;
- hash += hash >> 11;
- }
-
- // Handle end case
- if (rem) {
- hash += static_cast<unsigned char>(s[0]);
- hash ^= hash << 11;
- hash += hash >> 17;
- }
-
- // Force "avalanching" of final 127 bits
- hash ^= hash << 3;
- hash += hash >> 5;
- hash ^= hash << 2;
- hash += hash >> 15;
- hash ^= hash << 10;
-
- // this avoids ever returning a hash code of 0, since that is used to
- // signal "hash not computed yet", using a value that is likely to be
- // effectively the same as 0 when the low bits are masked
- if (hash == 0)
- hash = 0x80000000;
-
- return hash;
-}
-
-#ifndef NDEBUG
-void UString::Rep::checkConsistency() const
-{
- const UString::BaseString* base = baseString();
-
- // There is no recursion for base strings.
- ASSERT(base == base->baseString());
- if (isStatic()) {
- // There are only two static strings: null and empty.
- ASSERT(!len);
-
- // Static strings cannot get in identifier tables, because they are globally shared.
- ASSERT(!identifierTable());
- }
+ int length = 0;
+ while (characters[length] != UChar(0))
+ ++length;
- // The string fits in buffer.
- ASSERT(base->usedPreCapacity <= base->preCapacity);
- ASSERT(base->usedCapacity <= base->capacity);
- ASSERT(-offset <= base->usedPreCapacity);
- ASSERT(offset + len <= base->usedCapacity);
+ m_impl = StringImpl::create(characters, length);
}
-#endif
-// put these early so they can be inlined
-static inline size_t expandedSize(size_t size, size_t otherSize)
+// Construct a string with latin1 data.
+UString::UString(const LChar* characters, unsigned length)
+ : m_impl(characters ? StringImpl::create(characters, length) : 0)
{
- // Do the size calculation in two parts, returning overflowIndicator if
- // we overflow the maximum value that we can handle.
-
- if (size > maxUChars())
- return overflowIndicator();
-
- size_t expandedSize = ((size + 10) / 10 * 11) + 1;
- if (maxUChars() - expandedSize < otherSize)
- return overflowIndicator();
-
- return expandedSize + otherSize;
}
-static inline bool expandCapacity(UString::Rep* rep, int requiredLength)
+UString::UString(const char* characters, unsigned length)
+ : m_impl(characters ? StringImpl::create(reinterpret_cast<const LChar*>(characters), length) : 0)
{
- rep->checkConsistency();
-
- UString::BaseString* base = rep->baseString();
-
- if (requiredLength > base->capacity) {
- size_t newCapacity = expandedSize(requiredLength, base->preCapacity);
- UChar* oldBuf = base->buf;
- base->buf = reallocChars(base->buf, newCapacity);
- if (!base->buf) {
- base->buf = oldBuf;
- return false;
- }
- base->capacity = newCapacity - base->preCapacity;
- }
- if (requiredLength > base->usedCapacity)
- base->usedCapacity = requiredLength;
-
- rep->checkConsistency();
- return true;
}
-void UString::expandCapacity(int requiredLength)
+// Construct a string with latin1 data, from a null-terminated source.
+UString::UString(const LChar* characters)
+ : m_impl(characters ? StringImpl::create(characters) : 0)
{
- if (!JSC::expandCapacity(m_rep.get(), requiredLength))
- makeNull();
}
-void UString::expandPreCapacity(int requiredPreCap)
+UString::UString(const char* characters)
+ : m_impl(characters ? StringImpl::create(reinterpret_cast<const LChar*>(characters)) : 0)
{
- m_rep->checkConsistency();
-
- BaseString* base = m_rep->baseString();
-
- if (requiredPreCap > base->preCapacity) {
- size_t newCapacity = expandedSize(requiredPreCap, base->capacity);
- int delta = newCapacity - base->capacity - base->preCapacity;
-
- UChar* newBuf = allocChars(newCapacity);
- if (!newBuf) {
- makeNull();
- return;
- }
- copyChars(newBuf + delta, base->buf, base->capacity + base->preCapacity);
- fastFree(base->buf);
- base->buf = newBuf;
-
- base->preCapacity = newCapacity - base->capacity;
- }
- if (requiredPreCap > base->usedPreCapacity)
- base->usedPreCapacity = requiredPreCap;
-
- m_rep->checkConsistency();
}
-static PassRefPtr<UString::Rep> createRep(const char* c)
+UString UString::number(int i)
{
- if (!c)
- return &UString::Rep::null();
-
- if (!c[0])
- return &UString::Rep::empty();
-
- size_t length = strlen(c);
- UChar* d = allocChars(length);
- if (!d)
- return &UString::Rep::null();
- else {
- for (size_t i = 0; i < length; i++)
- d[i] = static_cast<unsigned char>(c[i]); // use unsigned char to zero-extend instead of sign-extend
- return UString::Rep::create(d, static_cast<int>(length));
- }
-
-}
-
-UString::UString(const char* c)
- : m_rep(createRep(c))
-{
-}
-
-UString::UString(const UChar* c, int length)
-{
- if (length == 0)
- m_rep = &Rep::empty();
- else
- m_rep = Rep::createCopying(c, length);
-}
-
-UString::UString(UChar* c, int length, bool copy)
-{
- if (length == 0)
- m_rep = &Rep::empty();
- else if (copy)
- m_rep = Rep::createCopying(c, length);
- else
- m_rep = Rep::create(c, length);
-}
-
-UString::UString(const Vector<UChar>& buffer)
-{
- if (!buffer.size())
- m_rep = &Rep::empty();
- else
- m_rep = Rep::createCopying(buffer.data(), buffer.size());
-}
-
-static ALWAYS_INLINE int newCapacityWithOverflowCheck(const int currentCapacity, const int extendLength, const bool plusOne = false)
-{
- ASSERT_WITH_MESSAGE(extendLength >= 0, "extendedLength = %d", extendLength);
-
- const int plusLength = plusOne ? 1 : 0;
- if (currentCapacity > std::numeric_limits<int>::max() - extendLength - plusLength)
- CRASH();
-
- return currentCapacity + extendLength + plusLength;
-}
-
-static ALWAYS_INLINE PassRefPtr<UString::Rep> concatenate(PassRefPtr<UString::Rep> r, const UChar* tData, int tSize)
-{
- RefPtr<UString::Rep> rep = r;
-
- rep->checkConsistency();
-
- int thisSize = rep->size();
- int thisOffset = rep->offset;
- int length = thisSize + tSize;
- UString::BaseString* base = rep->baseString();
-
- // possible cases:
- if (tSize == 0) {
- // t is empty
- } else if (thisSize == 0) {
- // this is empty
- rep = UString::Rep::createCopying(tData, tSize);
- } else if (rep == base && rep->rc == 1) {
- // this is direct and has refcount of 1 (so we can just alter it directly)
- if (!expandCapacity(rep.get(), newCapacityWithOverflowCheck(thisOffset, length)))
- rep = &UString::Rep::null();
- if (rep->data()) {
- copyChars(rep->data() + thisSize, tData, tSize);
- rep->len = length;
- rep->_hash = 0;
- }
- } else if (thisOffset + thisSize == base->usedCapacity && thisSize >= minShareSize) {
- // this reaches the end of the buffer - extend it if it's long enough to append to
- if (!expandCapacity(rep.get(), newCapacityWithOverflowCheck(thisOffset, length)))
- rep = &UString::Rep::null();
- if (rep->data()) {
- copyChars(rep->data() + thisSize, tData, tSize);
- rep = UString::Rep::create(rep, 0, length);
- }
- } else {
- // this is shared with someone using more capacity, gotta make a whole new string
- size_t newCapacity = expandedSize(length, 0);
- UChar* d = allocChars(newCapacity);
- if (!d)
- rep = &UString::Rep::null();
- else {
- copyChars(d, rep->data(), thisSize);
- copyChars(d + thisSize, tData, tSize);
- rep = UString::Rep::create(d, length);
- rep->baseString()->capacity = newCapacity;
- }
- }
+ LChar buf[1 + sizeof(i) * 3];
+ LChar* end = buf + WTF_ARRAY_LENGTH(buf);
+ LChar* p = end;
- rep->checkConsistency();
-
- return rep.release();
-}
-
-static ALWAYS_INLINE PassRefPtr<UString::Rep> concatenate(PassRefPtr<UString::Rep> r, const char* t)
-{
- RefPtr<UString::Rep> rep = r;
-
- rep->checkConsistency();
-
- int thisSize = rep->size();
- int thisOffset = rep->offset;
- int tSize = static_cast<int>(strlen(t));
- int length = thisSize + tSize;
- UString::BaseString* base = rep->baseString();
-
- // possible cases:
- if (thisSize == 0) {
- // this is empty
- rep = createRep(t);
- } else if (tSize == 0) {
- // t is empty, we'll just return *this below.
- } else if (rep == base && rep->rc == 1) {
- // this is direct and has refcount of 1 (so we can just alter it directly)
- expandCapacity(rep.get(), newCapacityWithOverflowCheck(thisOffset, length));
- UChar* d = rep->data();
- if (d) {
- for (int i = 0; i < tSize; ++i)
- d[thisSize + i] = static_cast<unsigned char>(t[i]); // use unsigned char to zero-extend instead of sign-extend
- rep->len = length;
- rep->_hash = 0;
- }
- } else if (thisOffset + thisSize == base->usedCapacity && thisSize >= minShareSize) {
- // this string reaches the end of the buffer - extend it
- expandCapacity(rep.get(), newCapacityWithOverflowCheck(thisOffset, length));
- UChar* d = rep->data();
- if (d) {
- for (int i = 0; i < tSize; ++i)
- d[thisSize + i] = static_cast<unsigned char>(t[i]); // use unsigned char to zero-extend instead of sign-extend
- rep = UString::Rep::create(rep, 0, length);
- }
- } else {
- // this is shared with someone using more capacity, gotta make a whole new string
- size_t newCapacity = expandedSize(length, 0);
- UChar* d = allocChars(newCapacity);
- if (!d)
- rep = &UString::Rep::null();
- else {
- copyChars(d, rep->data(), thisSize);
- for (int i = 0; i < tSize; ++i)
- d[thisSize + i] = static_cast<unsigned char>(t[i]); // use unsigned char to zero-extend instead of sign-extend
- rep = UString::Rep::create(d, length);
- rep->baseString()->capacity = newCapacity;
- }
- }
-
- rep->checkConsistency();
-
- return rep.release();
-}
-
-PassRefPtr<UString::Rep> concatenate(UString::Rep* a, UString::Rep* b)
-{
- a->checkConsistency();
- b->checkConsistency();
-
- int aSize = a->size();
- int aOffset = a->offset;
- int bSize = b->size();
- int bOffset = b->offset;
- int length = aSize + bSize;
-
- // possible cases:
-
- // a is empty
- if (aSize == 0)
- return b;
- // b is empty
- if (bSize == 0)
- return a;
-
- UString::BaseString* aBase = a->baseString();
- if (bSize == 1 && aOffset + aSize == aBase->usedCapacity && aOffset + length <= aBase->capacity) {
- // b is a single character (common fast case)
- aBase->usedCapacity = aOffset + length;
- a->data()[aSize] = b->data()[0];
- return UString::Rep::create(a, 0, length);
- }
-
- UString::BaseString* bBase = b->baseString();
- if (aOffset + aSize == aBase->usedCapacity && aSize >= minShareSize && 4 * aSize >= bSize
- && (-bOffset != bBase->usedPreCapacity || aSize >= bSize)) {
- // - a reaches the end of its buffer so it qualifies for shared append
- // - also, it's at least a quarter the length of b - appending to a much shorter
- // string does more harm than good
- // - however, if b qualifies for prepend and is longer than a, we'd rather prepend
-
- UString x(a);
- x.expandCapacity(newCapacityWithOverflowCheck(aOffset, length));
- if (!a->data() || !x.data())
- return 0;
- copyChars(a->data() + aSize, b->data(), bSize);
- PassRefPtr<UString::Rep> result = UString::Rep::create(a, 0, length);
-
- a->checkConsistency();
- b->checkConsistency();
- result->checkConsistency();
-
- return result;
- }
-
- if (-bOffset == bBase->usedPreCapacity && bSize >= minShareSize && 4 * bSize >= aSize) {
- // - b reaches the beginning of its buffer so it qualifies for shared prepend
- // - also, it's at least a quarter the length of a - prepending to a much shorter
- // string does more harm than good
- UString y(b);
- y.expandPreCapacity(-bOffset + aSize);
- if (!b->data() || !y.data())
- return 0;
- copyChars(b->data() - aSize, a->data(), aSize);
- PassRefPtr<UString::Rep> result = UString::Rep::create(b, -aSize, length);
-
- a->checkConsistency();
- b->checkConsistency();
- result->checkConsistency();
-
- return result;
- }
-
- // a does not qualify for append, and b does not qualify for prepend, gotta make a whole new string
- size_t newCapacity = expandedSize(length, 0);
- UChar* d = allocChars(newCapacity);
- if (!d)
- return 0;
- copyChars(d, a->data(), aSize);
- copyChars(d + aSize, b->data(), bSize);
- PassRefPtr<UString::Rep> result = UString::Rep::create(d, length);
- result->baseString()->capacity = newCapacity;
-
- a->checkConsistency();
- b->checkConsistency();
- result->checkConsistency();
-
- return result;
-}
-
-PassRefPtr<UString::Rep> concatenate(UString::Rep* rep, int i)
-{
- UChar buf[1 + sizeof(i) * 3];
- UChar* end = buf + sizeof(buf) / sizeof(UChar);
- UChar* p = end;
-
if (i == 0)
*--p = '0';
else if (i == INT_MIN) {
char minBuf[1 + sizeof(i) * 3];
- sprintf(minBuf, "%d", INT_MIN);
- return concatenate(rep, minBuf);
+ snprintf(minBuf, sizeof(minBuf), "%d", INT_MIN);
+ return UString(minBuf);
} else {
bool negative = false;
if (i < 0) {
*--p = '-';
}
- return concatenate(rep, p, static_cast<int>(end - p));
-
+ return UString(p, static_cast<unsigned>(end - p));
}
-PassRefPtr<UString::Rep> concatenate(UString::Rep* rep, double d)
+UString UString::number(long long i)
{
- // avoid ever printing -NaN, in JS conceptually there is only one NaN value
- if (isnan(d))
- return concatenate(rep, "NaN");
-
- if (d == 0.0) // stringify -0 as 0
- d = 0.0;
+ LChar buf[1 + sizeof(i) * 3];
+ LChar* end = buf + WTF_ARRAY_LENGTH(buf);
+ LChar* p = end;
- char buf[80];
- int decimalPoint;
- int sign;
-
- char* result = WTF::dtoa(d, 0, &decimalPoint, &sign, NULL);
- int length = static_cast<int>(strlen(result));
-
- int i = 0;
- if (sign)
- buf[i++] = '-';
-
- if (decimalPoint <= 0 && decimalPoint > -6) {
- buf[i++] = '0';
- buf[i++] = '.';
- for (int j = decimalPoint; j < 0; j++)
- buf[i++] = '0';
- strcpy(buf + i, result);
- } else if (decimalPoint <= 21 && decimalPoint > 0) {
- if (length <= decimalPoint) {
- strcpy(buf + i, result);
- i += length;
- for (int j = 0; j < decimalPoint - length; j++)
- buf[i++] = '0';
- buf[i] = '\0';
- } else {
- strncpy(buf + i, result, decimalPoint);
- i += decimalPoint;
- buf[i++] = '.';
- strcpy(buf + i, result + decimalPoint);
- }
- } else if (result[0] < '0' || result[0] > '9')
- strcpy(buf + i, result);
- else {
- buf[i++] = result[0];
- if (length > 1) {
- buf[i++] = '.';
- strcpy(buf + i, result + 1);
- i += length - 1;
- }
-
- buf[i++] = 'e';
- buf[i++] = (decimalPoint >= 0) ? '+' : '-';
- // decimalPoint can't be more than 3 digits decimal given the
- // nature of float representation
- int exponential = decimalPoint - 1;
- if (exponential < 0)
- exponential = -exponential;
- if (exponential >= 100)
- buf[i++] = static_cast<char>('0' + exponential / 100);
- if (exponential >= 10)
- buf[i++] = static_cast<char>('0' + (exponential % 100) / 10);
- buf[i++] = static_cast<char>('0' + exponential % 10);
- buf[i++] = '\0';
- }
-
- WTF::freedtoa(result);
-
- return concatenate(rep, buf);
-}
-
-UString UString::from(int i)
-{
- UChar buf[1 + sizeof(i) * 3];
- UChar* end = buf + sizeof(buf) / sizeof(UChar);
- UChar* p = end;
-
if (i == 0)
*--p = '0';
- else if (i == INT_MIN) {
+ else if (i == std::numeric_limits<long long>::min()) {
char minBuf[1 + sizeof(i) * 3];
- snprintf(minBuf, 1 + sizeof(i) * 3, "%d", INT_MIN);
+#if OS(WINDOWS)
+ snprintf(minBuf, sizeof(minBuf), "%I64d", std::numeric_limits<long long>::min());
+#else
+ snprintf(minBuf, sizeof(minBuf), "%lld", std::numeric_limits<long long>::min());
+#endif
return UString(minBuf);
} else {
bool negative = false;
*--p = '-';
}
- return UString(p, static_cast<int>(end - p));
+ return UString(p, static_cast<unsigned>(end - p));
}
-UString UString::from(unsigned int u)
+UString UString::number(unsigned u)
{
- UChar buf[sizeof(u) * 3];
- UChar* end = buf + sizeof(buf) / sizeof(UChar);
- UChar* p = end;
-
+ LChar buf[sizeof(u) * 3];
+ LChar* end = buf + WTF_ARRAY_LENGTH(buf);
+ LChar* p = end;
+
if (u == 0)
*--p = '0';
else {
u /= 10;
}
}
-
- return UString(p, static_cast<int>(end - p));
+
+ return UString(p, static_cast<unsigned>(end - p));
}
-UString UString::from(long l)
+UString UString::number(long l)
{
- UChar buf[1 + sizeof(l) * 3];
- UChar* end = buf + sizeof(buf) / sizeof(UChar);
- UChar* p = end;
+ LChar buf[1 + sizeof(l) * 3];
+ LChar* end = buf + WTF_ARRAY_LENGTH(buf);
+ LChar* p = end;
if (l == 0)
*--p = '0';
else if (l == LONG_MIN) {
char minBuf[1 + sizeof(l) * 3];
- snprintf(minBuf, 1 + sizeof(l) * 3, "%ld", LONG_MIN);
+ snprintf(minBuf, sizeof(minBuf), "%ld", LONG_MIN);
return UString(minBuf);
} else {
bool negative = false;
*--p = '-';
}
- return UString(p, static_cast<int>(end - p));
+ return UString(p, end - p);
}
-UString UString::from(double d)
+UString UString::number(double d)
{
- // avoid ever printing -NaN, in JS conceptually there is only one NaN value
- if (isnan(d))
- return "NaN";
-
- char buf[80];
- int decimalPoint;
- int sign;
-
- char* result = WTF::dtoa(d, 0, &decimalPoint, &sign, NULL);
- int length = static_cast<int>(strlen(result));
-
- int i = 0;
- if (sign)
- buf[i++] = '-';
-
- if (decimalPoint <= 0 && decimalPoint > -6) {
- buf[i++] = '0';
- buf[i++] = '.';
- for (int j = decimalPoint; j < 0; j++)
- buf[i++] = '0';
- strlcpy(buf + i, result, sizeof(buf) - i);
- } else if (decimalPoint <= 21 && decimalPoint > 0) {
- if (length <= decimalPoint) {
- strlcpy(buf + i, result, sizeof(buf) - i);
- i += length;
- for (int j = 0; j < decimalPoint - length; j++)
- buf[i++] = '0';
- buf[i] = '\0';
- } else {
- int len = (decimalPoint <= static_cast<int>(sizeof(buf)) - i ? decimalPoint : sizeof(buf) - i);
- strncpy(buf + i, result, len);
- i += len;
- buf[i++] = '.';
- strlcpy(buf + i, result + decimalPoint, sizeof(buf) - i);
- }
- } else if (result[0] < '0' || result[0] > '9')
- strlcpy(buf + i, result, sizeof(buf) - i);
- else {
- buf[i++] = result[0];
- if (length > 1) {
- buf[i++] = '.';
- strlcpy(buf + i, result + 1, sizeof(buf) - i);
- i += length - 1;
- }
-
- buf[i++] = 'e';
- buf[i++] = (decimalPoint >= 0) ? '+' : '-';
- // decimalPoint can't be more than 3 digits decimal given the
- // nature of float representation
- int exponential = decimalPoint - 1;
- if (exponential < 0)
- exponential = -exponential;
- if (exponential >= 100)
- buf[i++] = static_cast<char>('0' + exponential / 100);
- if (exponential >= 10)
- buf[i++] = static_cast<char>('0' + (exponential % 100) / 10);
- buf[i++] = static_cast<char>('0' + exponential % 10);
- buf[i++] = '\0';
- ASSERT(i <= static_cast<int>(sizeof(buf)));
- }
-
- WTF::freedtoa(result);
-
- return UString(buf);
+ NumberToStringBuffer buffer;
+ return UString(numberToString(d, buffer));
}
-UString UString::spliceSubstringsWithSeparators(const Range* substringRanges, int rangeCount, const UString* separators, int separatorCount) const
+UString UString::substringSharingImpl(unsigned offset, unsigned length) const
{
- m_rep->checkConsistency();
-
- if (rangeCount == 1 && separatorCount == 0) {
- int thisSize = size();
- int position = substringRanges[0].position;
- int length = substringRanges[0].length;
- if (position <= 0 && length >= thisSize)
- return *this;
- return UString::Rep::create(m_rep, max(0, position), min(thisSize, length));
- }
-
- int totalLength = 0;
- for (int i = 0; i < rangeCount; i++)
- totalLength += substringRanges[i].length;
- for (int i = 0; i < separatorCount; i++)
- totalLength += separators[i].size();
+ // FIXME: We used to check against a limit of Heap::minExtraCost / sizeof(UChar).
- if (totalLength == 0)
- return "";
-
- UChar* buffer = allocChars(totalLength);
- if (!buffer)
- return null();
-
- int maxCount = max(rangeCount, separatorCount);
- int bufferPos = 0;
- for (int i = 0; i < maxCount; i++) {
- if (i < rangeCount) {
- copyChars(buffer + bufferPos, data() + substringRanges[i].position, substringRanges[i].length);
- bufferPos += substringRanges[i].length;
- }
- if (i < separatorCount) {
- copyChars(buffer + bufferPos, separators[i].data(), separators[i].size());
- bufferPos += separators[i].size();
- }
- }
-
- return UString::Rep::create(buffer, totalLength);
-}
-
-UString& UString::append(const UString &t)
-{
- m_rep->checkConsistency();
- t.rep()->checkConsistency();
+ unsigned stringLength = this->length();
+ offset = min(offset, stringLength);
+ length = min(length, stringLength - offset);
- int thisSize = size();
- int thisOffset = m_rep->offset;
- int tSize = t.size();
- int length = thisSize + tSize;
- BaseString* base = m_rep->baseString();
-
- // possible cases:
- if (thisSize == 0) {
- // this is empty
- *this = t;
- } else if (tSize == 0) {
- // t is empty
- } else if (m_rep == base && m_rep->rc == 1) {
- // this is direct and has refcount of 1 (so we can just alter it directly)
- expandCapacity(newCapacityWithOverflowCheck(thisOffset, length));
- if (data()) {
- copyChars(m_rep->data() + thisSize, t.data(), tSize);
- m_rep->len = length;
- m_rep->_hash = 0;
- }
- } else if (thisOffset + thisSize == base->usedCapacity && thisSize >= minShareSize) {
- // this reaches the end of the buffer - extend it if it's long enough to append to
- expandCapacity(newCapacityWithOverflowCheck(thisOffset, length));
- if (data()) {
- copyChars(m_rep->data() + thisSize, t.data(), tSize);
- m_rep = Rep::create(m_rep, 0, length);
- }
- } else {
- // this is shared with someone using more capacity, gotta make a whole new string
- size_t newCapacity = expandedSize(length, 0);
- UChar* d = allocChars(newCapacity);
- if (!d)
- makeNull();
- else {
- copyChars(d, data(), thisSize);
- copyChars(d + thisSize, t.data(), tSize);
- m_rep = Rep::create(d, length);
- m_rep->baseString()->capacity = newCapacity;
- }
- }
-
- m_rep->checkConsistency();
- t.rep()->checkConsistency();
-
- return *this;
+ if (!offset && length == stringLength)
+ return *this;
+ return UString(StringImpl::create(m_impl, offset, length));
}
-UString& UString::append(const UChar* tData, int tSize)
+bool operator==(const UString& s1, const char *s2)
{
- m_rep = concatenate(m_rep.release(), tData, tSize);
- return *this;
-}
+ if (s1.isEmpty())
+ return !s2;
-UString& UString::append(const char* t)
-{
- m_rep = concatenate(m_rep.release(), t);
- return *this;
+ return equal(s1.impl(), s2);
}
-UString& UString::append(UChar c)
+// This method assumes that all simple checks have been performed by
+// the inlined operator==() in the header file.
+bool equalSlowCase(const UString& s1, const UString& s2)
{
- m_rep->checkConsistency();
-
- int thisOffset = m_rep->offset;
- int length = size();
- BaseString* base = m_rep->baseString();
+ StringImpl* rep1 = s1.impl();
+ StringImpl* rep2 = s2.impl();
+ unsigned size1 = rep1->length();
- // possible cases:
- if (length == 0) {
- // this is empty - must make a new m_rep because we don't want to pollute the shared empty one
- size_t newCapacity = expandedSize(1, 0);
- UChar* d = allocChars(newCapacity);
- if (!d)
- makeNull();
- else {
- d[0] = c;
- m_rep = Rep::create(d, 1);
- m_rep->baseString()->capacity = newCapacity;
- }
- } else if (m_rep == base && m_rep->rc == 1) {
- // this is direct and has refcount of 1 (so we can just alter it directly)
- expandCapacity(newCapacityWithOverflowCheck(thisOffset, length, true));
- UChar* d = m_rep->data();
- if (d) {
- d[length] = c;
- m_rep->len = length + 1;
- m_rep->_hash = 0;
- }
- } else if (thisOffset + length == base->usedCapacity && length >= minShareSize) {
- // this reaches the end of the string - extend it and share
- expandCapacity(newCapacityWithOverflowCheck(thisOffset, length, true));
- UChar* d = m_rep->data();
- if (d) {
- d[length] = c;
- m_rep = Rep::create(m_rep, 0, length + 1);
+ // At this point we know
+ // (a) that the strings are the same length and
+ // (b) that they are greater than zero length.
+ bool s1Is8Bit = rep1->is8Bit();
+ bool s2Is8Bit = rep2->is8Bit();
+
+ if (s1Is8Bit) {
+ const LChar* d1 = rep1->characters8();
+ if (s2Is8Bit) {
+ const LChar* d2 = rep2->characters8();
+
+ if (d1 == d2) // Check to see if the data pointers are the same.
+ return true;
+
+ // Do quick checks for sizes 1 and 2.
+ switch (size1) {
+ case 1:
+ return d1[0] == d2[0];
+ case 2:
+ return (d1[0] == d2[0]) & (d1[1] == d2[1]);
+ default:
+ return (!memcmp(d1, d2, size1 * sizeof(LChar)));
+ }
}
- } else {
- // this is shared with someone using more capacity, gotta make a whole new string
- size_t newCapacity = expandedSize(length + 1, 0);
- UChar* d = allocChars(newCapacity);
- if (!d)
- makeNull();
- else {
- copyChars(d, data(), length);
- d[length] = c;
- m_rep = Rep::create(d, length + 1);
- m_rep->baseString()->capacity = newCapacity;
+
+ const UChar* d2 = rep2->characters16();
+
+ for (unsigned i = 0; i < size1; i++) {
+ if (d1[i] != d2[i])
+ return false;
}
+ return true;
}
-
- m_rep->checkConsistency();
-
- return *this;
-}
-
-bool UString::getCString(CStringBuffer& buffer) const
-{
- int length = size();
- int neededSize = length + 1;
- buffer.resize(neededSize);
- char* buf = buffer.data();
-
- UChar ored = 0;
- const UChar* p = data();
- char* q = buf;
- const UChar* limit = p + length;
- while (p != limit) {
- UChar c = p[0];
- ored |= c;
- *q = static_cast<char>(c);
- ++p;
- ++q;
- }
- *q = '\0';
-
- return !(ored & 0xFF00);
-}
-
-char* UString::ascii() const
-{
- int length = size();
- int neededSize = length + 1;
- delete[] statBuffer;
- statBuffer = new char[neededSize];
-
- const UChar* p = data();
- char* q = statBuffer;
- const UChar* limit = p + length;
- while (p != limit) {
- *q = static_cast<char>(p[0]);
- ++p;
- ++q;
- }
- *q = '\0';
-
- return statBuffer;
-}
-
-UString& UString::operator=(const char* c)
-{
- if (!c) {
- m_rep = &Rep::null();
- return *this;
- }
-
- if (!c[0]) {
- m_rep = &Rep::empty();
- return *this;
- }
-
- int l = static_cast<int>(strlen(c));
- UChar* d;
- BaseString* base = m_rep->baseString();
- if (m_rep->rc == 1 && l <= base->capacity && m_rep == base && m_rep->offset == 0 && base->preCapacity == 0) {
- d = base->buf;
- m_rep->_hash = 0;
- m_rep->len = l;
- } else {
- d = allocChars(l);
- if (!d) {
- makeNull();
- return *this;
+
+ if (s2Is8Bit) {
+ const UChar* d1 = rep1->characters16();
+ const LChar* d2 = rep2->characters8();
+
+ for (unsigned i = 0; i < size1; i++) {
+ if (d1[i] != d2[i])
+ return false;
}
- m_rep = Rep::create(d, l);
- }
- for (int i = 0; i < l; i++)
- d[i] = static_cast<unsigned char>(c[i]); // use unsigned char to zero-extend instead of sign-extend
-
- return *this;
-}
-
-bool UString::is8Bit() const
-{
- const UChar* u = data();
- const UChar* limit = u + size();
- while (u < limit) {
- if (u[0] > 0xFF)
- return false;
- ++u;
- }
-
- return true;
-}
-
-UChar UString::operator[](int pos) const
-{
- if (pos >= size())
- return '\0';
- return data()[pos];
-}
-
-double UString::toDouble(bool tolerateTrailingJunk, bool tolerateEmptyString) const
-{
- if (size() == 1) {
- UChar c = data()[0];
- if (isASCIIDigit(c))
- return c - '0';
- if (isASCIISpace(c) && tolerateEmptyString)
- return 0;
- return NaN;
+ return true;
+
}
-
- // FIXME: If tolerateTrailingJunk is true, then we want to tolerate non-8-bit junk
- // after the number, so this is too strict a check.
- CStringBuffer s;
- if (!getCString(s))
- return NaN;
- const char* c = s.data();
-
- // skip leading white space
- while (isASCIISpace(*c))
- c++;
-
- // empty string ?
- if (*c == '\0')
- return tolerateEmptyString ? 0.0 : NaN;
-
- double d;
-
- // hex number ?
- if (*c == '0' && (*(c + 1) == 'x' || *(c + 1) == 'X')) {
- const char* firstDigitPosition = c + 2;
- c++;
- d = 0.0;
- while (*(++c)) {
- if (*c >= '0' && *c <= '9')
- d = d * 16.0 + *c - '0';
- else if ((*c >= 'A' && *c <= 'F') || (*c >= 'a' && *c <= 'f'))
- d = d * 16.0 + (*c & 0xdf) - 'A' + 10.0;
- else
- break;
- }
-
- if (d >= mantissaOverflowLowerBound)
- d = parseIntOverflow(firstDigitPosition, c - firstDigitPosition, 16);
- } else {
- // regular number ?
- char* end;
- d = WTF::strtod(c, &end);
- if ((d != 0.0 || end != c) && d != Inf && d != -Inf) {
- c = end;
- } else {
- double sign = 1.0;
-
- if (*c == '+')
- c++;
- else if (*c == '-') {
- sign = -1.0;
- c++;
- }
-
- // We used strtod() to do the conversion. However, strtod() handles
- // infinite values slightly differently than JavaScript in that it
- // converts the string "inf" with any capitalization to infinity,
- // whereas the ECMA spec requires that it be converted to NaN.
-
- if (c[0] == 'I' && c[1] == 'n' && c[2] == 'f' && c[3] == 'i' && c[4] == 'n' && c[5] == 'i' && c[6] == 't' && c[7] == 'y') {
- d = sign * Inf;
- c += 8;
- } else if ((d == Inf || d == -Inf) && *c != 'I' && *c != 'i')
- c = end;
- else
- return NaN;
- }
+
+ const UChar* d1 = rep1->characters16();
+ const UChar* d2 = rep2->characters16();
+
+ if (d1 == d2) // Check to see if the data pointers are the same.
+ return true;
+
+ // Do quick checks for sizes 1 and 2.
+ switch (size1) {
+ case 1:
+ return d1[0] == d2[0];
+ case 2:
+ return (d1[0] == d2[0]) & (d1[1] == d2[1]);
+ default:
+ return (!memcmp(d1, d2, size1 * sizeof(UChar)));
}
-
- // allow trailing white space
- while (isASCIISpace(*c))
- c++;
- // don't allow anything after - unless tolerant=true
- if (!tolerateTrailingJunk && *c != '\0')
- d = NaN;
-
- return d;
}
-double UString::toDouble(bool tolerateTrailingJunk) const
-{
- return toDouble(tolerateTrailingJunk, true);
-}
-
-double UString::toDouble() const
-{
- return toDouble(false, true);
-}
-
-uint32_t UString::toUInt32(bool* ok) const
+bool operator<(const UString& s1, const UString& s2)
{
- double d = toDouble();
- bool b = true;
-
- if (d != static_cast<uint32_t>(d)) {
- b = false;
- d = 0;
+ const unsigned l1 = s1.length();
+ const unsigned l2 = s2.length();
+ const unsigned lmin = l1 < l2 ? l1 : l2;
+ if (s1.is8Bit() && s2.is8Bit()) {
+ const LChar* c1 = s1.characters8();
+ const LChar* c2 = s2.characters8();
+ unsigned length = 0;
+ while (length < lmin && *c1 == *c2) {
+ c1++;
+ c2++;
+ length++;
+ }
+ if (length < lmin)
+ return (c1[0] < c2[0]);
+
+ return (l1 < l2);
+ }
+ const UChar* c1 = s1.characters();
+ const UChar* c2 = s2.characters();
+ unsigned length = 0;
+ while (length < lmin && *c1 == *c2) {
+ c1++;
+ c2++;
+ length++;
}
+ if (length < lmin)
+ return (c1[0] < c2[0]);
- if (ok)
- *ok = b;
-
- return static_cast<uint32_t>(d);
+ return (l1 < l2);
}
-uint32_t UString::toUInt32(bool* ok, bool tolerateEmptyString) const
+bool operator>(const UString& s1, const UString& s2)
{
- double d = toDouble(false, tolerateEmptyString);
- bool b = true;
-
- if (d != static_cast<uint32_t>(d)) {
- b = false;
- d = 0;
+ const unsigned l1 = s1.length();
+ const unsigned l2 = s2.length();
+ const unsigned lmin = l1 < l2 ? l1 : l2;
+ const UChar* c1 = s1.characters();
+ const UChar* c2 = s2.characters();
+ unsigned l = 0;
+ while (l < lmin && *c1 == *c2) {
+ c1++;
+ c2++;
+ l++;
}
+ if (l < lmin)
+ return (c1[0] > c2[0]);
- if (ok)
- *ok = b;
-
- return static_cast<uint32_t>(d);
+ return (l1 > l2);
}
-uint32_t UString::toStrictUInt32(bool* ok) const
+CString UString::ascii() const
{
- if (ok)
- *ok = false;
-
- // Empty string is not OK.
- int len = m_rep->len;
- if (len == 0)
- return 0;
- const UChar* p = m_rep->data();
- unsigned short c = p[0];
-
- // If the first digit is 0, only 0 itself is OK.
- if (c == '0') {
- if (len == 1 && ok)
- *ok = true;
- return 0;
- }
-
- // Convert to UInt32, checking for overflow.
- uint32_t i = 0;
- while (1) {
- // Process character, turning it into a digit.
- if (c < '0' || c > '9')
- return 0;
- const unsigned d = c - '0';
+ // Basic Latin1 (ISO) encoding - Unicode characters 0..255 are
+ // preserved, characters outside of this range are converted to '?'.
- // Multiply by 10, checking for overflow out of 32 bits.
- if (i > 0xFFFFFFFFU / 10)
- return 0;
- i *= 10;
+ unsigned length = this->length();
- // Add in the digit, checking for overflow out of 32 bits.
- const unsigned max = 0xFFFFFFFFU - d;
- if (i > max)
- return 0;
- i += d;
-
- // Handle end of string.
- if (--len == 0) {
- if (ok)
- *ok = true;
- return i;
+ if (this->is8Bit()) {
+ const LChar* characters = this->characters8();
+
+ char* characterBuffer;
+ CString result = CString::newUninitialized(length, characterBuffer);
+
+ for (unsigned i = 0; i < length; ++i) {
+ LChar ch = characters[i];
+ characterBuffer[i] = ch && (ch < 0x20 || ch > 0x7f) ? '?' : ch;
}
-
- // Get next character.
- c = *(++p);
+
+ return result;
}
-}
-int UString::find(const UString& f, int pos) const
-{
- int fsz = f.size();
+ const UChar* characters = this->characters16();
- if (pos < 0)
- pos = 0;
+ char* characterBuffer;
+ CString result = CString::newUninitialized(length, characterBuffer);
- if (fsz == 1) {
- UChar ch = f[0];
- const UChar* end = data() + size();
- for (const UChar* c = data() + pos; c < end; c++) {
- if (*c == ch)
- return static_cast<int>(c - data());
- }
- return -1;
- }
-
- int sz = size();
- if (sz < fsz)
- return -1;
- if (fsz == 0)
- return pos;
- const UChar* end = data() + sz - fsz;
- int fsizeminusone = (fsz - 1) * sizeof(UChar);
- const UChar* fdata = f.data();
- unsigned short fchar = fdata[0];
- ++fdata;
- for (const UChar* c = data() + pos; c <= end; c++) {
- if (c[0] == fchar && !memcmp(c + 1, fdata, fsizeminusone))
- return static_cast<int>(c - data());
+ for (unsigned i = 0; i < length; ++i) {
+ UChar ch = characters[i];
+ characterBuffer[i] = ch && (ch < 0x20 || ch >= 0x7f) ? '?' : ch;
}
- return -1;
+ return result;
}
-int UString::find(UChar ch, int pos) const
+CString UString::latin1() const
{
- if (pos < 0)
- pos = 0;
- const UChar* end = data() + size();
- for (const UChar* c = data() + pos; c < end; c++) {
- if (*c == ch)
- return static_cast<int>(c - data());
- }
-
- return -1;
-}
+ // Basic Latin1 (ISO) encoding - Unicode characters 0..255 are
+ // preserved, characters outside of this range are converted to '?'.
-int UString::rfind(const UString& f, int pos) const
-{
- int sz = size();
- int fsz = f.size();
- if (sz < fsz)
- return -1;
- if (pos < 0)
- pos = 0;
- if (pos > sz - fsz)
- pos = sz - fsz;
- if (fsz == 0)
- return pos;
- int fsizeminusone = (fsz - 1) * sizeof(UChar);
- const UChar* fdata = f.data();
- for (const UChar* c = data() + pos; c >= data(); c--) {
- if (*c == *fdata && !memcmp(c + 1, fdata + 1, fsizeminusone))
- return static_cast<int>(c - data());
- }
+ unsigned length = this->length();
+ const UChar* characters = this->characters();
- return -1;
-}
+ char* characterBuffer;
+ CString result = CString::newUninitialized(length, characterBuffer);
-int UString::rfind(UChar ch, int pos) const
-{
- if (isEmpty())
- return -1;
- if (pos + 1 >= size())
- pos = size() - 1;
- for (const UChar* c = data() + pos; c >= data(); c--) {
- if (*c == ch)
- return static_cast<int>(c - data());
+ for (unsigned i = 0; i < length; ++i) {
+ UChar ch = characters[i];
+ characterBuffer[i] = ch > 0xff ? '?' : ch;
}
- return -1;
-}
-
-UString UString::substr(int pos, int len) const
-{
- int s = size();
-
- if (pos < 0)
- pos = 0;
- else if (pos >= s)
- pos = s;
- if (len < 0)
- len = s;
- if (pos + len >= s)
- len = s - pos;
-
- if (pos == 0 && len == s)
- return *this;
-
- return UString(Rep::create(m_rep, pos, len));
-}
-
-bool operator==(const UString& s1, const UString& s2)
-{
- int size = s1.size();
- switch (size) {
- case 0:
- return !s2.size();
- case 1:
- return s2.size() == 1 && s1.data()[0] == s2.data()[0];
- default:
- return s2.size() == size && memcmp(s1.data(), s2.data(), size * sizeof(UChar)) == 0;
- }
+ return result;
}
-bool operator==(const UString& s1, const char *s2)
+// Helper to write a three-byte UTF-8 code point to the buffer, caller must check room is available.
+static inline void putUTF8Triple(char*& buffer, UChar ch)
{
- if (s2 == 0)
- return s1.isEmpty();
-
- const UChar* u = s1.data();
- const UChar* uend = u + s1.size();
- while (u != uend && *s2) {
- if (u[0] != (unsigned char)*s2)
- return false;
- s2++;
- u++;
- }
-
- return u == uend && *s2 == 0;
+ ASSERT(ch >= 0x0800);
+ *buffer++ = static_cast<char>(((ch >> 12) & 0x0F) | 0xE0);
+ *buffer++ = static_cast<char>(((ch >> 6) & 0x3F) | 0x80);
+ *buffer++ = static_cast<char>((ch & 0x3F) | 0x80);
}
-bool operator<(const UString& s1, const UString& s2)
+CString UString::utf8(bool strict) const
{
- const int l1 = s1.size();
- const int l2 = s2.size();
- const int lmin = l1 < l2 ? l1 : l2;
- const UChar* c1 = s1.data();
- const UChar* c2 = s2.data();
- int l = 0;
- while (l < lmin && *c1 == *c2) {
- c1++;
- c2++;
- l++;
- }
- if (l < lmin)
- return (c1[0] < c2[0]);
+ unsigned length = this->length();
- return (l1 < l2);
-}
+ if (!length)
+ return CString("", 0);
-bool operator>(const UString& s1, const UString& s2)
-{
- const int l1 = s1.size();
- const int l2 = s2.size();
- const int lmin = l1 < l2 ? l1 : l2;
- const UChar* c1 = s1.data();
- const UChar* c2 = s2.data();
- int l = 0;
- while (l < lmin && *c1 == *c2) {
- c1++;
- c2++;
- l++;
- }
- if (l < lmin)
- return (c1[0] > c2[0]);
+ // Allocate a buffer big enough to hold all the characters
+ // (an individual UTF-16 UChar can only expand to 3 UTF-8 bytes).
+ // Optimization ideas, if we find this function is hot:
+ // * We could speculatively create a CStringBuffer to contain 'length'
+ // characters, and resize if necessary (i.e. if the buffer contains
+ // non-ascii characters). (Alternatively, scan the buffer first for
+ // ascii characters, so we know this will be sufficient).
+ // * We could allocate a CStringBuffer with an appropriate size to
+ // have a good chance of being able to write the string into the
+ // buffer without reallocing (say, 1.5 x length).
+ if (length > numeric_limits<unsigned>::max() / 3)
+ return CString();
- return (l1 > l2);
-}
+ Vector<char, 1024> bufferVector(length * 3);
+ char* buffer = bufferVector.data();
-int compare(const UString& s1, const UString& s2)
-{
- const int l1 = s1.size();
- const int l2 = s2.size();
- const int lmin = l1 < l2 ? l1 : l2;
- const UChar* c1 = s1.data();
- const UChar* c2 = s2.data();
- int l = 0;
- while (l < lmin && *c1 == *c2) {
- c1++;
- c2++;
- l++;
- }
+ if (is8Bit()) {
+ const LChar* characters = this->characters8();
- if (l < lmin)
- return (c1[0] > c2[0]) ? 1 : -1;
+ ConversionResult result = convertLatin1ToUTF8(&characters, characters + length, &buffer, buffer + bufferVector.size());
+ ASSERT_UNUSED(result, result != targetExhausted); // (length * 3) should be sufficient for any conversion
+ } else {
+ const UChar* characters = this->characters16();
- if (l1 == l2)
- return 0;
+ ConversionResult result = convertUTF16ToUTF8(&characters, characters + length, &buffer, buffer + bufferVector.size(), strict);
+ ASSERT(result != targetExhausted); // (length * 3) should be sufficient for any conversion
- return (l1 > l2) ? 1 : -1;
-}
+ // Only produced from strict conversion.
+ if (result == sourceIllegal)
+ return CString();
-bool equal(const UString::Rep* r, const UString::Rep* b)
-{
- int length = r->len;
- if (length != b->len)
- return false;
- const UChar* d = r->data();
- const UChar* s = b->data();
- for (int i = 0; i != length; ++i) {
- if (d[i] != s[i])
- return false;
+ // Check for an unconverted high surrogate.
+ if (result == sourceExhausted) {
+ if (strict)
+ return CString();
+ // This should be one unpaired high surrogate. Treat it the same
+ // was as an unpaired high surrogate would have been handled in
+ // the middle of a string with non-strict conversion - which is
+ // to say, simply encode it to UTF-8.
+ ASSERT((characters + 1) == (this->characters() + length));
+ ASSERT((*characters >= 0xD800) && (*characters <= 0xDBFF));
+ // There should be room left, since one UChar hasn't been converted.
+ ASSERT((buffer + 3) <= (buffer + bufferVector.size()));
+ putUTF8Triple(buffer, *characters);
+ }
}
- return true;
-}
-
-CString UString::UTF8String(bool strict) const
-{
- // Allocate a buffer big enough to hold all the characters.
- const int length = size();
- Vector<char, 1024> buffer(length * 3);
-
- // Convert to runs of 8-bit characters.
- char* p = buffer.data();
- const UChar* d = reinterpret_cast<const UChar*>(&data()[0]);
- ConversionResult result = convertUTF16ToUTF8(&d, d + length, &p, p + buffer.size(), strict);
- if (result != conversionOK)
- return CString();
- return CString(buffer.data(), p - buffer.data());
-}
-
-// For use in error handling code paths -- having this not be inlined helps avoid PIC branches to fetch the global on Mac OS X.
-NEVER_INLINE void UString::makeNull()
-{
- m_rep = &Rep::null();
-}
-
-// For use in error handling code paths -- having this not be inlined helps avoid PIC branches to fetch the global on Mac OS X.
-NEVER_INLINE UString::Rep* UString::nullRep()
-{
- return &Rep::null();
+ return CString(bufferVector.data(), buffer - bufferVector.data());
}
} // namespace JSC
projects / apple / javascriptcore.git / blobdiff