Add outlining support for wmemcpy.

6 files changed, 21 insertions, 13 deletions

diff --git a/binaryninjaapi.h b/binaryninjaapi.h
index bab743d3..0ca84526 100644
--- a/binaryninjaapi.h
+++ b/binaryninjaapi.h
@@ -7127,8 +7127,7 @@ namespace BinaryNinja {
 		void ClearUserGlobalPointerValue();
 		void SetUserGlobalPointerValue(const Confidence<RegisterValue>& value);
 
-		std::optional<std::pair<std::string, BNStringType>> StringifyUnicodeData(
-			Architecture* arch, const DataBuffer& buffer, bool allowShortStrings = false);
+		std::optional<std::pair<std::string, BNStringType>> StringifyUnicodeData(Architecture* arch, const DataBuffer& buffer, bool nullTerminates = true, bool allowShortStrings = false);
 	};
 
 	class MemoryMap
diff --git a/binaryninjacore.h b/binaryninjacore.h
index deca98e5..b2f02949 100644
--- a/binaryninjacore.h
+++ b/binaryninjacore.h
@@ -37,14 +37,14 @@
 // Current ABI version for linking to the core. This is incremented any time
 // there are changes to the API that affect linking, including new functions,
 // new types, or modifications to existing functions or types.
-#define BN_CURRENT_CORE_ABI_VERSION 115
+#define BN_CURRENT_CORE_ABI_VERSION 116
 
 // Minimum ABI version that is supported for loading of plugins. Plugins that
 // are linked to an ABI version less than this will not be able to load and
 // will require rebuilding. The minimum version is increased when there are
 // incompatible changes that break binary compatibility, such as changes to
 // existing types or functions.
-#define BN_MINIMUM_CORE_ABI_VERSION 115
+#define BN_MINIMUM_CORE_ABI_VERSION 116
 
 #ifdef __GNUC__
 	#ifdef BINARYNINJACORE_LIBRARY
@@ -3582,7 +3582,8 @@ extern "C"
 		BuiltinMemset,
 		BuiltinStrncpy,
 		BuiltinStrcpy,
-		BuiltinWcscpy
+		BuiltinWcscpy,
+		BuiltinWmemcpy
 	} BNBuiltinType;
 
 	typedef struct BNSegmentInfo {
@@ -4394,8 +4395,7 @@ extern "C"
 	BINARYNINJACOREAPI void BNClearUserGlobalPointerValue(BNBinaryView* view);
 	BINARYNINJACOREAPI void BNSetUserGlobalPointerValue(BNBinaryView* view, BNRegisterValueWithConfidence value);
 
-	BINARYNINJACOREAPI bool BNStringifyUnicodeData(BNBinaryView* data, BNArchitecture* arch, const BNDataBuffer* buffer,
-		bool allowShortStrings, char** string, BNStringType* type);
+	BINARYNINJACOREAPI bool BNStringifyUnicodeData(BNBinaryView* data, BNArchitecture* arch, const BNDataBuffer* buffer, bool nullTerminates, bool allowShortStrings, char** string, BNStringType* type);
 
 	// Raw binary data view
 	BINARYNINJACOREAPI BNBinaryView* BNCreateBinaryDataView(BNFileMetadata* file);
diff --git a/binaryview.cpp b/binaryview.cpp
index f2b33dc3..4f6c9354 100644
--- a/binaryview.cpp
+++ b/binaryview.cpp
@@ -5530,11 +5530,11 @@ void BinaryView::SetUserGlobalPointerValue(const Confidence<RegisterValue>& valu
 }
 
 
-optional<pair<string, BNStringType>> BinaryView::StringifyUnicodeData(Architecture* arch, const DataBuffer& buffer, bool allowShortStrings)
+optional<pair<string, BNStringType>> BinaryView::StringifyUnicodeData(Architecture* arch, const DataBuffer& buffer, bool nullTerminates, bool allowShortStrings)
 {
 	char* str = nullptr;
 	BNStringType type = AsciiString;
-	if (!BNStringifyUnicodeData(m_object, arch ? arch->GetObject() : nullptr, buffer.GetBufferObject(), allowShortStrings, &str, &type))
+	if (!BNStringifyUnicodeData(m_object, arch ? arch->GetObject() : nullptr, buffer.GetBufferObject(), nullTerminates, allowShortStrings, &str, &type))
 		return nullopt;
 
 	string result(str);
diff --git a/lang/c/pseudoc.cpp b/lang/c/pseudoc.cpp
index 6f3c58ab..4d585966 100644
--- a/lang/c/pseudoc.cpp
+++ b/lang/c/pseudoc.cpp
@@ -1260,6 +1260,7 @@ void PseudoCFunction::GetExprTextInternal(const HighLevelILInstruction& instr, H
 			const ConstantData& data = instr.GetConstantData<HLIL_CONST_DATA>();
 			if (auto [db, builtin] = data.ToDataBuffer(); db.GetLength())
 			{
+				bool nullTerminates = true;
 				switch (builtin)
 				{
 					case BuiltinStrcpy:
@@ -1284,9 +1285,12 @@ void PseudoCFunction::GetExprTextInternal(const HighLevelILInstruction& instr, H
 						tokens.Append(BraceToken, "}");
 						break;
 					}
+					case BuiltinWmemcpy:
+						nullTerminates = false;
+						// FALL_THROUGH
 					default:
 					{
-						if (auto unicode = GetFunction()->GetView()->StringifyUnicodeData(instr.function->GetArchitecture(), db); unicode.has_value())
+						if (auto unicode = GetFunction()->GetView()->StringifyUnicodeData(instr.function->GetArchitecture(), db, nullTerminates); unicode.has_value())
 						{
 							auto wideStringPrefix = (builtin == BuiltinWcscpy) ? "L" : "";
 							auto tokenContext = (builtin == BuiltinWcscpy) ? ConstStringDataTokenContext : ConstDataTokenContext;
diff --git a/lang/rust/pseudorust.cpp b/lang/rust/pseudorust.cpp
index 2f9b751c..429def51 100644
--- a/lang/rust/pseudorust.cpp
+++ b/lang/rust/pseudorust.cpp
@@ -1278,6 +1278,7 @@ void PseudoRustFunction::GetExprText(const HighLevelILInstruction& instr, HighLe
 			const ConstantData& data = instr.GetConstantData<HLIL_CONST_DATA>();
 			if (auto [db, builtin] = data.ToDataBuffer(); db.GetLength())
 			{
+				bool nullTerminates = true;
 				switch (builtin)
 				{
 					case BuiltinStrcpy:
@@ -1302,9 +1303,12 @@ void PseudoRustFunction::GetExprText(const HighLevelILInstruction& instr, HighLe
 						tokens.Append(BraceToken, "}");
 						break;
 					}
+					case BuiltinWmemcpy:
+						nullTerminates = false;
+						// FALL_THROUGH
 					default:
 					{
-						if (auto unicode = GetFunction()->GetView()->StringifyUnicodeData(instr.function->GetArchitecture(), db); unicode.has_value())
+						if (auto unicode = GetFunction()->GetView()->StringifyUnicodeData(instr.function->GetArchitecture(), db, nullTerminates); unicode.has_value())
 						{
 							auto wideStringPrefix = (builtin == BuiltinWcscpy) ? "L" : "";
 							auto tokenContext = (builtin == BuiltinWcscpy) ? ConstStringDataTokenContext : ConstDataTokenContext;
diff --git a/python/binaryview.py b/python/binaryview.py
index 7b14b57b..16c2dbfb 100644
--- a/python/binaryview.py
+++ b/python/binaryview.py
@@ -10141,11 +10141,12 @@ to a the type "tagRECT" found in the typelibrary "winX64common"
 		"""
 		return MemoryMap(handle=self.handle)
 
-	def stringify_unicode_data(self, arch: Optional['architecture.Architecture'], buffer: 'databuffer.DataBuffer', allow_short_strings: bool = False) -> Tuple[Optional[str], Optional[StringType]]:
+	def stringify_unicode_data(self, arch: Optional['architecture.Architecture'], buffer: 'databuffer.DataBuffer', null_terminates: bool = True, allow_short_strings: bool = False) -> Tuple[Optional[str], Optional[StringType]]:
 		"""
 		``stringify_unicode_data`` converts a buffer of unicode data into a string representation.
 		:param arch: The architecture to use for stringification, or None to use the current architecture of the BinaryView
 		:param buffer: The DataBuffer containing the unicode data to stringify
+		:param null_terminates: If True, stops stringification at the first null character, otherwise continues until the end of the buffer
 		:param allow_short_strings: If True, allows short strings to be returned, otherwise only long strings are returned
 		:return: A tuple containing the string representation and its type, or (None, None) if the stringification fails
 		:rtype: Tuple[Optional[str], Optional[StringType]]
@@ -10156,7 +10157,7 @@ to a the type "tagRECT" found in the typelibrary "winX64common"
 		string_type = ctypes.c_int()
 		if arch is not None:
 			arch = arch.handle
-		if not core.BNStringifyUnicodeData(self.handle, arch, buffer.handle, allow_short_strings, ctypes.byref(string), ctypes.byref(string_type)):
+		if not core.BNStringifyUnicodeData(self.handle, arch, buffer.handle, null_terminates, allow_short_strings, ctypes.byref(string), ctypes.byref(string_type)):
 			return None, None
 		result = string.value.decode('utf-8')
 		core.free_string(string)