Add RecognizeConstantData to StringRecognizer API

Adds a new recognize_constant_data method to the StringRecognizer API,
enabling plugins to decode constant data expressions (HLIL_CONST_DATA)
recovered by the outline resolver. The recognizer receives the full
instruction, giving access to the data buffer and builtin type via the
constant_data accessor.

Also fixes a pre-existing ctypes mismatch in get_constant_data_and_builtin
where BNBuiltinType (uint8_t) was incorrectly passed as c_int.

2 files changed, 41 insertions, 1 deletions

diff --git a/python/function.py b/python/function.py
index 93eb7d15..bc8def47 100644
--- a/python/function.py
+++ b/python/function.py
@@ -1938,7 +1938,7 @@ class Function:
 	def get_constant_data_and_builtin(
 			self, state: RegisterValueType, value: int, size: int = 0
 	) -> Tuple[databuffer.DataBuffer, BuiltinType]:
-		builtin = ctypes.c_int()
+		builtin = ctypes.c_ubyte()
 		db = databuffer.DataBuffer(
 			handle=core.BNGetConstantData(self.handle, state, value, size, ctypes.byref(builtin)))
 		return db, BuiltinType(builtin.value)
diff --git a/python/stringrecognizer.py b/python/stringrecognizer.py
index 055e236d..bdf9423a 100644
--- a/python/stringrecognizer.py
+++ b/python/stringrecognizer.py
@@ -195,6 +195,9 @@ class StringRecognizer(metaclass=_StringRecognizerMetaClass):
 			self._cb.recognizeExternPointer = self._cb.recognizeExternPointer.__class__(self._recognize_extern_pointer)
 		if self.recognize_import.__func__ != StringRecognizer.recognize_import:
 			self._cb.recognizeImport = self._cb.recognizeImport.__class__(self._recognize_import)
+		if self.recognize_constant_data.__func__ != StringRecognizer.recognize_constant_data:
+			self._cb.recognizeConstantData = self._cb.recognizeConstantData.__class__(
+				self._recognize_constant_data)
 		self.handle = core.BNRegisterStringRecognizer(self.__class__.recognizer_name, self._cb)
 		self.__class__._registered_recognizers.append(self)
 
@@ -263,6 +266,19 @@ class StringRecognizer(metaclass=_StringRecognizerMetaClass):
 			log_error_for_exception("Unhandled Python exception in StringRecognizer._recognize_import")
 			return False
 
+	def _recognize_constant_data(self, ctxt, hlil, expr, result):
+		try:
+			hlil = highlevelil.HighLevelILFunction(handle=core.BNNewHighLevelILFunctionReference(hlil))
+			instr = hlil.get_expr(highlevelil.ExpressionIndex(expr))
+			ref = self.recognize_constant_data(instr)
+			if ref is None:
+				return False
+			result[0] = ref._to_core_struct(True)
+			return True
+		except Exception:
+			log_error_for_exception("Unhandled Python exception in StringRecognizer._recognize_constant_data")
+			return False
+
 	@property
 	def name(self) -> str:
 		if hasattr(self, 'handle'):
@@ -347,6 +363,22 @@ class StringRecognizer(metaclass=_StringRecognizerMetaClass):
 		"""
 		return None
 
+	def recognize_constant_data(
+		self, instr: 'highlevelil.HighLevelILInstruction'
+	) -> Optional['binaryview.DerivedString']:
+		"""
+		Can be overridden to recognize strings for constant data expressions (HLIL_CONST_DATA).
+		These expressions are generated by the outline resolver when it recovers constant data
+		streams from scattered stores. The instruction provides access to the data buffer and
+		builtin type via its ``constant_data`` accessor.
+
+		If a string is found, return a :py:class:`~binaryninja.binaryview.DerivedString` with the string information.
+
+		:param instr: High level expression containing the constant data
+		:return: Optional :py:class:`~binaryninja.binaryview.DerivedString` for any string that is found.
+		"""
+		return None
+
 
 _recognizer_cache = {}
 
@@ -402,3 +434,11 @@ class CoreStringRecognizer(StringRecognizer):
 		if not core.BNStringRecognizerRecognizeImport(self.handle, instr.function.handle, instr.expr_index, type.handle, val, string):
 			return None
 		return binaryview.DerivedString._from_core_struct(string, True)
+
+	def recognize_constant_data(
+		self, instr: 'highlevelil.HighLevelILInstruction'
+	) -> Optional['binaryview.DerivedString']:
+		string = core.BNDerivedString()
+		if not core.BNStringRecognizerRecognizeConstantData(self.handle, instr.function.handle, instr.expr_index, string):
+			return None
+		return binaryview.DerivedString._from_core_struct(string, True)