Method ProcessCFStrings

objectivec/objc.cpp:1465–1580

void ObjCProcessor::ProcessCFStrings()
{
	m_symbolQueue = new SymbolQueue();
	uint64_t ptrSize = m_data->GetAddressSize();
	// https://github.com/apple/llvm-project/blob/next/clang/lib/CodeGen/CodeGenModule.cpp#L6129
	// See also ASTContext.cpp ctrl+f __NSConstantString_tag

	// The place these flags are used is unclear, along with any clear flag definitions, but they are useful for
	// introspection
	EnumerationBuilder __cfStringFlagBuilder;
	__cfStringFlagBuilder.AddMemberWithValue("SwiftABI", 0b1);
	__cfStringFlagBuilder.AddMemberWithValue("Swift4_1", 0b100);
	// LLVM also sets 0x7c0 (0b11111000000) on both UTF8 and UTF16 strings however it is unclear what this denotes.
	__cfStringFlagBuilder.AddMemberWithValue("UTF8", 0b1000);
	__cfStringFlagBuilder.AddMemberWithValue("UTF16", 0b10000);
	auto type = finalizeEnumerationBuilder(m_data, __cfStringFlagBuilder, ptrSize, {"CFStringFlag"});
	m_typeNames.cfStringFlag = type.first;

	StructureBuilder __cfStringStructBuilder;
	__cfStringStructBuilder.AddMember(Type::PointerType(ptrSize, Type::VoidType()), "isa");
	__cfStringStructBuilder.AddMember(Type::NamedType(m_data, m_typeNames.cfStringFlag), "flags");
	__cfStringStructBuilder.AddMember(Type::PointerType(ptrSize, Type::IntegerType(1, true)), "data");
	__cfStringStructBuilder.AddMember(Type::IntegerType(ptrSize, false), "length");
	type = finalizeStructureBuilder(m_data, __cfStringStructBuilder, "__NSConstantString");
	m_typeNames.cfString = type.first;

	StructureBuilder __cfStringUTF16StructBuilder;
	__cfStringUTF16StructBuilder.AddMember(Type::PointerType(ptrSize, Type::VoidType()), "isa");
	__cfStringUTF16StructBuilder.AddMember(Type::NamedType(m_data, m_typeNames.cfStringFlag), "flags");
	__cfStringUTF16StructBuilder.AddMember(Type::PointerType(ptrSize, Type::IntegerType(2, true)), "data");
	__cfStringUTF16StructBuilder.AddMember(Type::IntegerType(ptrSize, false), "length");
	type = finalizeStructureBuilder(m_data, __cfStringUTF16StructBuilder, "__NSConstantString_UTF16");
	m_typeNames.cfStringUTF16 = type.first;

	auto reader = GetReader();
	if (auto cfstrings = GetSectionWithName("__cfstring"))
	{
		auto start = cfstrings->GetStart();
		auto end = cfstrings->GetEnd();
		auto typeWidth = Type::NamedType(m_data, m_typeNames.cfString)->GetWidth();
		m_data->BeginBulkModifySymbols();
		for (view_ptr_t i = start; i < end; i += typeWidth)
		{
			reader->Seek(i + ptrSize);
			uint64_t flags = reader->ReadPointer();
			auto strLoc = ReadPointerAccountingForRelocations(reader.get());
			auto size = reader->ReadPointer();
			std::string str;
			if (flags & 0b10000)  // UTF16
			{
				auto data = m_data->ReadBuffer(strLoc, size * 2);

				str = "";
				for (uint64_t bufferOff = 0; bufferOff < size * 2; bufferOff += 2)
				{
					uint8_t* rawData = static_cast<uint8_t*>(data.GetData());
					uint8_t* offsetAddress = rawData + bufferOff;
					uint16_t c = *reinterpret_cast<uint16_t*>(offsetAddress);