Fixed incorrect string length being stored

CMakeLists.txt

@@ -10,9 +10,6 @@ include_directories(${CMAKE_BINARY_DIR})
 set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
 set(CXX_STANDARD 23)
 
-set(BUILD_TESTS OFF)
-set(BUILD_BENCHMARK OFF)
-
 set(CAVM_RUNTIME_CHECK_SUPPORT ON)
 
 add_subdirectory(src)
@@ -26,4 +23,6 @@ message("src " ${CONSTANT_POOL_TAG_SRC})
 add_executable(parser
     ${CONSTANT_POOL_TAG_SRC}
 )
+if (UNIX)
     set(CMAKE_CXX_FLAGS "-std=c++${CXX_STANDARD} -O0 -g -Wpedantic")
+endif()

include/BigEndian.h

@@ -4,6 +4,17 @@
 #include <Types.h>
 #include <cstddef>
 
+#ifdef _MSC_VER
+
+#include <intrin.h>
+
+#define __builtin_bswap16 _byteswap_ushort
+#define __builtin_bswap32 _byteswap_ulong
+#define __builtin_bswap64 _byteswap_uint64
+
+#endif
+
+
 template <typename T>
 class BigEndian
 {
@@ -33,8 +44,8 @@ class BigEndian
 				return __builtin_bswap32(val);
 			case 8:
 				return __builtin_bswap64(val);
-			// Clang doesn't support __builtin_bswap128, so we fall back to a generic byte-order swap implementation
+			// Only gcc has a 128-bit byte-order swap builtin, so we fall back to a generic byte-order swap implementation for clang and for msvc
-			#ifndef __clang__
+			#if not defined(__clang__) && not defined(_MSC_VER)
 			case 16:
 				return __builtin_bswap128(val);
 			#endif

include/ClassFormat.h

@@ -58,6 +58,12 @@ enum class ACCESS_FLAGS
 };
 } // namespace Method
 
+struct BaseAttribute
+{
+	u2 attribute_name_index;
+	u4 attribute_length;
+};
+
 struct attribute_info
 {
 	u2 attribute_name_index;
@@ -65,12 +71,6 @@ struct attribute_info
 	std::vector<u1> info;
 };
 
-struct BaseAttribute
-{
-	u2 attribute_name_index;
-	u4 attribute_length;
-};
-
 struct method_info
 {
 	u2 access_flags;

include/ConstantPoolTags/Tags.h

@@ -4,7 +4,7 @@
 #include <SpecTypes.h>
 #include <cstdint>
 #include <cstdlib>
-
+#include <string>
 
 struct JavaClassFormat;
 
@@ -47,7 +47,7 @@ struct ConstantUtf8Info : ConstantPoolInfo
 {
 	ConstantPoolTags::Tags tag;
 	u2 length;
-	unsigned char *bytes;
+	std::string bytes;
 	// std::vector<unsigned char> bytes;
 };
 

src/AttributeParser.cpp

@@ -1,4 +1,4 @@
-#include "ConstantPoolTags/Tags.h"
+#include <ConstantPoolTags/Tags.h>
 #include <Globals.h>
 #include <Utils/Utils.h>
 #include <Attributes.h>
@@ -8,6 +8,7 @@
 #include <cassert>
 #include <string>
 #include <Utils/CRC32_CompileTime.hpp>
+#include <algorithm>
 
 attribute_info Attributes::AttributesParser(JavaClassFormat& jc)
 {
@@ -28,7 +29,7 @@ attribute_info Attributes::AttributesParser(JavaClassFormat& jc)
 	int idx = ret.info.data()->ToHostFormat();
 
 	ASSERT(VEC_ELEM_TYPE(jc.constant_pool[ret.attribute_name_index.ToHostFormat() - 1]) == TYPE(ConstantPoolTags::ConstantUtf8Info));
-	std::string str = reinterpret_cast<const char*>(std::any_cast<ConstantPoolTags::ConstantUtf8Info>(jc.constant_pool[ret.attribute_name_index.ToHostFormat() - 1]).bytes);
+	std::string str = std::any_cast<ConstantPoolTags::ConstantUtf8Info>(jc.constant_pool[ret.attribute_name_index.ToHostFormat() - 1]).bytes;
 
 	switch ((ClassAttributes)crc32_rec(CRC32, str.c_str()))
 	{
@@ -39,6 +40,7 @@ attribute_info Attributes::AttributesParser(JavaClassFormat& jc)
 					  << std::any_cast<ConstantPoolTags::ConstantUtf8Info>(jc.constant_pool[idx - 1]).bytes
 					  << std::endl;
 			SOURCE_FILE_Attribute attr;
+			attr.sourcefile_index = idx;
 			// attr.
 			break;
         }

src/ConstantPoolTags/PoolParser.cpp

@@ -14,31 +14,34 @@ void ConstantPoolTags::ConstantPoolParser(ConstantPoolTags::Tags t, JavaClassFor
 		case Tags::Utf8: {
 			// std::cout << "FIXME: Implement Utf8 Constant Pool Tag" << std::endl;
 			ConstantUtf8Info ref;
-			std::vector<uint8_t> length = Utils::ReadFromStart(Globals::buffer(), 2);
+			ref.length = Utils::ReadFromStartIntoVal<uint16_t>(Globals::buffer(), 2);
+			// std::vector<uint8_t> length = Utils::ReadFromStart(Globals::buffer(), 2);
 			ref.tag = t;
-			ref.length = Utils::vecToVal<uint16_t>(length);
+			// ref.length = Utils::vecToVal<uint16_t>(length);
 
 			unsigned char x = Utils::ReadFromStartIntoVal<unsigned char>(Globals::buffer(), 1);
 			bool valid = true;
 
-			ref.bytes = new unsigned char[ref.length.ToHostFormat()];
+			unsigned char* bytes = new unsigned char[ref.length.ToHostFormat()];
+
+			int len = 0;
 
 			for (int i = 0; i < ref.length.ToHostFormat(); i++)
 			{
 				if ((x & 0x80) == 0)
 				{
-					ref.bytes[i] = x;
+					bytes[i] = x;
 				}
 				else if ((x & 0x800) == 0)
 				{
 					unsigned char y = Utils::ReadFromStartIntoVal<unsigned char>(Globals::buffer(), 1);
-					ref.bytes[i] = ((x & 0x1f) << 6) + (y & 0x3f);
+					bytes[i] = ((x & 0x1f) << 6) + (y & 0x3f);
 				}
 				else if ((x & 0x10000) == 0)
 				{
 					unsigned char y = Utils::ReadFromStartIntoVal<unsigned char>(Globals::buffer(), 1);
 					unsigned char z = Utils::ReadFromStartIntoVal<unsigned char>(Globals::buffer(), 1);
-					ref.bytes[i] = ((x & 0xf) << 12) + ((y & 0x3f) << 6) + (z & 0x3f);
+					bytes[i] = ((x & 0xf) << 12) + ((y & 0x3f) << 6) + (z & 0x3f);
 				}
 				else
 				{
@@ -46,11 +49,20 @@ void ConstantPoolTags::ConstantPoolParser(ConstantPoolTags::Tags t, JavaClassFor
 					exit(1);
 				}
 
-				if (i != ref.length.ToHostFormat() - 1)
+				if (i < ref.length.ToHostFormat() - 1)
 					x = Utils::ReadFromStartIntoVal<unsigned char>(Globals::buffer(), 1);
+
+				len += 1;
 			}
+
+			std::string bytes1;
+			
+			bytes1.append(reinterpret_cast<const char*>(bytes), 0, len);
+
+			ref.bytes = bytes1;
 			jc.constant_pool.push_back(ref);
 
+			ref.length = len;
 			break;
 		}
 		case Tags::Integer:

src/Main.cpp

@@ -163,7 +163,8 @@ int main(int argc, char **argv)
 
 		if (x == TYPE(ConstantPoolTags::ConstantUtf8Info))
 		{
-			std::cout << "utf8:\t" << std::any_cast<ConstantPoolTags::ConstantUtf8Info>(jc.constant_pool[i]).bytes << std::endl;
+			ConstantPoolTags::ConstantUtf8Info ref = std::any_cast<ConstantPoolTags::ConstantUtf8Info>(jc.constant_pool[i]);
+			std::cout << "utf8:\t" << "(length): " << ref.length.ToHostFormat() << ",\t(str): " << ref.bytes << std::endl;
 		}
 		else if (x == TYPE(ConstantPoolTags::ConstantClassInfo))
 		{

src/Method/MethodParser.cpp

@@ -32,7 +32,7 @@ void Methods::MethodParser(method_info &p, JavaClassFormat &jc)
 		ASSERT(VEC_ELEM_TYPE(v) == TYPE(ConstantPoolTags::ConstantUtf8Info));
 		ConstantPoolTags::ConstantUtf8Info str = std::any_cast<ConstantPoolTags::ConstantUtf8Info>(v);
 
-		ASSERT(strcmp(reinterpret_cast<const char *>(str.bytes), "Code") == 0);
+		ASSERT(strcmp(str.bytes.c_str(), "Code") == 0);
 
 		u4 attribute_length = Utils::ReadFromStartIntoVal<uint32_t>(Globals::buffer());