Merge variables API/UI, variable liveness API for determining soundness of variable merges

3 files changed, 87 insertions, 2 deletions

diff --git a/python/function.py b/python/function.py
index 9a0f099a..d2855234 100644
--- a/python/function.py
+++ b/python/function.py
@@ -1377,6 +1377,26 @@ class Function:
 			return None
 		return flowgraph.CoreFlowGraph(graph)
 
+	@property
+	def merged_vars(self) -> Dict['variable.Variable', List['variable.Variable']]:
+		"""
+		Map of merged variables, organized by target variable (read-only). Use ``merge_vars`` and
+		``unmerge_vars`` to update merged variables.
+		"""
+		count = ctypes.c_ulonglong()
+		data = core.BNGetMergedVariables(self.handle, count)
+
+		result = {}
+		for i in range(count.value):
+			target = Variable.from_BNVariable(self, data[i].target)
+			sources = []
+			for j in range(data[i].sourceCount):
+				sources.append(Variable.from_BNVariable(self, data[i].sources[j]))
+			result[target] = sources
+
+		core.BNFreeMergedVariableList(data, count.value)
+		return result
+
 	def mark_recent_use(self) -> None:
 		core.BNMarkFunctionAsRecentlyUsed(self.handle)
 
@@ -3316,6 +3336,44 @@ class Function:
 			return start.value
 		return None
 
+	def merge_vars(
+	    self, target: 'variable.Variable', sources: Union[List['variable.Variable'], 'variable.Variable']
+	) -> None:
+		"""
+		``merge_vars`` merges one or more varibles in ``sources`` into the ``target`` variable. All
+		variable accesses to the variables in ``sources`` will be rewritten to use ``target``.
+
+		:param Variable target: target variable
+		:param list(Variable) sources: list of source variables
+		"""
+		if isinstance(sources, variable.Variable):
+			sources = [sources]
+		source_list = (core.BNVariable * len(sources))()
+		for i in range(0, len(sources)):
+			source_list[i].type = sources[i].source_type
+			source_list[i].index = sources[i].index
+			source_list[i].storage = sources[i].storage
+		core.BNMergeVariables(self.handle, target.to_BNVariable(), source_list, len(sources))
+
+	def unmerge_vars(
+	    self, target: 'variable.Variable', sources: Union[List['variable.Variable'], 'variable.Variable']
+	) -> None:
+		"""
+		``unmerge_vars`` undoes variable merging performed with ``merge_vars``. The variables in
+		``sources`` will no longer be merged into the ``target`` variable.
+
+		:param Variable target: target variable
+		:param list(Variable) sources: list of source variables
+		"""
+		if isinstance(sources, variable.Variable):
+			sources = [sources]
+		source_list = (core.BNVariable * len(sources))()
+		for i in range(0, len(sources)):
+			source_list[i].type = sources[i].source_type
+			source_list[i].index = sources[i].index
+			source_list[i].storage = sources[i].storage
+		core.BNUnmergeVariables(self.handle, target.to_BNVariable(), source_list, len(sources))
+
 
 class AdvancedFunctionAnalysisDataRequestor:
 	def __init__(self, func: 'Function' = None):
diff --git a/python/mediumlevelil.py b/python/mediumlevelil.py
index 07ca7a22..ceb33f50 100644
--- a/python/mediumlevelil.py
+++ b/python/mediumlevelil.py
@@ -3024,6 +3024,28 @@ class MediumLevelILFunction:
 		finally:
 			core.BNFreeILInstructionList(instrs)
 
+	def get_live_instructions_for_var(self, var: 'variable.Variable', include_last_use: bool = True) -> List[MediumLevelILInstruction]:
+		"""
+		``get_live_instructions_for_var`` computes the list of instructions for which ``var`` is live.
+		If ``include_last_use`` is False, the last use of the variable will not be included in the
+		list (this allows for easier computation of overlaps in liveness between two variables).
+		If the variable is never used, this function will return an empty list.
+
+		:param SSAVariable var: the variable to query
+		:param bool include_last_use: whether to include the last use of the variable in the list of instructions
+		:return: list of instructions for which ``var`` is live
+		:rtype: list(MediumLevelILInstruction)
+		"""
+		count = ctypes.c_ulonglong()
+		var_data = var.to_BNVariable()
+		instrs = core.BNGetMediumLevelILLiveInstructionsForVariable(self.handle, var_data, include_last_use, count)
+		assert instrs is not None, "core.BNGetMediumLevelILLiveInstructionsForVariable returned None"
+		result = []
+		for i in range(0, count.value):
+			result.append(self[instrs[i]])
+		core.BNFreeILInstructionList(instrs)
+		return result
+
 	def get_ssa_var_value(self, ssa_var: SSAVariable) -> 'variable.RegisterValue':
 		var_data = ssa_var.var.to_BNVariable()
 		value = core.BNGetMediumLevelILSSAVarValue(self.handle, var_data, ssa_var.version)
diff --git a/python/variable.py b/python/variable.py
index 07b8863c..839e87ce 100644
--- a/python/variable.py
+++ b/python/variable.py
@@ -663,7 +663,7 @@ class Variable(CoreVariable):
 		if self.type is not None:
 			return f"<var {self.type.get_string_before_name()} {self.name}{self.type.get_string_after_name()}>"
 		else:
-			return repr(super())
+			return f"<var {self.name}>"
 
 	def __str__(self):
 		return self.name
@@ -712,7 +712,7 @@ class Variable(CoreVariable):
 	@property
 	def name(self) -> str:
 		"""Name of the variable, Settings thisslow because it ensures that analysis has been updated. """
-		return core.BNGetRealVariableName(self._function.handle, self._function.arch.handle, self.to_BNVariable())
+		return core.BNGetVariableNameOrDefault(self._function.handle, self.to_BNVariable())
 
 	@name.setter
 	def name(self, name: Optional[str]) -> None:
@@ -720,6 +720,11 @@ class Variable(CoreVariable):
 		self._function.view.update_analysis_and_wait()
 
 	@property
+	def last_seen_name(self) -> str:
+		"""Name of the variable, or the name most recently assigned if the variable has since been removed (read-only). """
+		return core.BNGetLastSeenVariableNameOrDefault(self._function.handle, self.to_BNVariable())
+
+	@property
 	def type(self) -> Optional['binaryninja.types.Type']:
 		var_type_conf = core.BNGetVariableType(self._function.handle, self.to_BNVariable())
 		if var_type_conf.type: