int StringCell::compare(const StringCell *other, UnicodeChar (*converter)(UnicodeChar)) const
{
const std::uint8_t *ourScanPtr = const_cast<StringCell*>(this)->utf8Data();
const std::uint8_t *ourEndPtr = &constUtf8Data()[byteLength()];
const std::uint8_t *theirScanPtr = const_cast<StringCell*>(other)->utf8Data();
const std::uint8_t *theirEndPtr = &other->constUtf8Data()[other->byteLength()];
while(true)
{
int ourEnd = ourScanPtr == ourEndPtr;
int theirEnd = theirScanPtr == theirEndPtr;
if (ourEnd || theirEnd)
{
// One of the strings ended
return theirEnd - ourEnd;
}
const UnicodeChar ourChar = utf8::decodeChar(&ourScanPtr);
const UnicodeChar theirChar = utf8::decodeChar(&theirScanPtr);
int charCompare = ourChar.compare(theirChar, converter);
if (charCompare != 0)
{
return charCompare;
}
}
return 0;
}
SymbolCell* SymbolCell::copy(alloc::Heap &heap)
{
void *cellPlacement = heap.allocate();
if (dataIsInline())
{
auto inlineThis = static_cast<InlineSymbolCell*>(this);
auto inlineCopy = new (cellPlacement) InlineSymbolCell(
inlineThis->inlineByteLength(),
inlineThis->inlineCharLength()
);
memcpy(inlineCopy->m_inlineData, constUtf8Data(), inlineThis->inlineByteLength());
return inlineCopy;
}
else
{
auto heapThis = static_cast<HeapSymbolCell*>(this);
return new (cellPlacement) HeapSymbolCell(
heapThis->heapByteArray()->ref(),
heapThis->heapByteLength(),
heapThis->heapCharLength()
);
}
}
int StringCell::compare(const StringCell *other) const
{
ByteLengthType compareBytes = std::min(byteLength(), other->byteLength());
// Bytewise comparisons in UTF-8 sort Unicode code points correctly
int result = memcmp(constUtf8Data(), other->constUtf8Data(), compareBytes);
if (result != 0)
{
return result;
}
else
{
return byteLength() - other->byteLength();
}
}
StringCell *StringCell::toConvertedString(World &world, UnicodeChar (*converter)(UnicodeChar))
{
const std::uint8_t *scanPtr = utf8Data();
const std::uint8_t *endPtr = &constUtf8Data()[byteLength()];
StringCellBuilder builder(charLength());
while(scanPtr != endPtr)
{
const UnicodeChar originalChar = utf8::decodeChar(&scanPtr);
const UnicodeChar convertedChar = (*converter)(originalChar);
builder << convertedChar;
}
return builder.result(world);
}