Explicit hash/eq for dataflow objects (#417)

* Explicit hash/eqs

* black

---------

Co-authored-by: Niklas Bergmann <[email protected]>

decompiler/structures/pseudo/expressions.py

@@ -35,7 +35,7 @@
 
 from ...util.insertion_ordered_set import InsertionOrderedSet
 from .complextypes import Enum
-from .typing import ArrayType, CustomType, Type, UnknownType
+from .typing import CustomType, Type, UnknownType
 
 T = TypeVar("T")
 DecompiledType = TypeVar("DecompiledType", bound=Type)
@@ -58,18 +58,6 @@ class DataflowObject(ABC):
     def __init__(self, tags: Optional[Tuple[Tag, ...]] = None):
         self.tags = tags
 
-    def __eq__(self, other) -> bool:
-        """Check for equality."""
-        return type(other) == type(self) and hash(self) == hash(other)
-
-    def __hash__(self) -> int:
-        """Return a hash value for the expression."""
-        return hash(repr(self))
-
-    def __repr__(self):
-        """Return a debug representation."""
-        return str(self)
-
     @abstractmethod
     def __iter__(self) -> Iterator[DataflowObject]:
         """Iterate all nested DataflowObjects."""
@@ -149,6 +137,12 @@ def __init__(self, msg: str, tags: Optional[Tuple[Tag, ...]] = None):
         self.msg = msg
         super().__init__(tags)
 
+    def __eq__(self, __value):
+        return isinstance(__value, UnknownExpression) and self.msg == __value.msg
+
+    def __hash__(self):
+        return hash(self.msg)
+
     def __str__(self) -> str:
         """Return the error message as string representation."""
         return self.msg
@@ -183,6 +177,17 @@ def __init__(
         self._pointee = pointee
         super().__init__(tags)
 
+    def __eq__(self, __value):
+        return (
+            isinstance(__value, Constant)
+            and self.value == __value.value
+            and self._type == __value._type
+            and self._pointee == __value.pointee
+        )
+
+    def __hash__(self):
+        return hash((tuple(self.value) if isinstance(self.value, list) else self.value, self._type, self._pointee))
+
     def __repr__(self) -> str:
         value = str(self) if isinstance(self.value, str) else self.value
         if self.pointee:
@@ -235,6 +240,12 @@ class NotUseableConstant(Constant):
     def __init__(self, value: str, tags: Optional[Tuple[Tag, ...]] = None):
         super().__init__(value, CustomType("double", 0), tags=tags)
 
+    def __eq__(self, __value):
+        return isinstance(__value, NotUseableConstant) and self.value == __value.value
+
+    def __hash__(self):
+        return hash(self.value)
+
     def __str__(self) -> str:
         """Return a string because NotUseableConstant are string only"""
         return self.value
@@ -255,6 +266,12 @@ def __init__(self, name: str, value: Union[int, float], vartype: Type = UnknownT
         super().__init__(value, vartype, tags=tags)
         self._name = name
 
+    def __eq__(self, __value):
+        return isinstance(__value, Symbol) and self._name == __value._name and self.value == __value.value
+
+    def __hash__(self):
+        return hash((self._name, self.value))
+
     @property
     def name(self) -> str:
         return self._name
@@ -278,13 +295,25 @@ def copy(self) -> Symbol:
 class FunctionSymbol(Symbol):
     """Represents a function name"""
 
+    def __eq__(self, __value):
+        return isinstance(__value, FunctionSymbol) and super().__eq__(__value)
+
+    def __hash__(self):
+        return super().__hash__()
+
     def copy(self) -> FunctionSymbol:
         return FunctionSymbol(self.name, self.value, self._type.copy(), self.tags)
 
 
 class ImportedFunctionSymbol(FunctionSymbol):
     """Represents an imported function name"""
 
+    def __eq__(self, __value):
+        return isinstance(__value, ImportedFunctionSymbol) and super().__eq__(__value)
+
+    def __hash__(self):
+        return super().__hash__()
+
     def copy(self) -> ImportedFunctionSymbol:
         return ImportedFunctionSymbol(self._name, self.value, self._type.copy(), self.tags)
 
@@ -297,6 +326,12 @@ class IntrinsicSymbol(FunctionSymbol):
     def __init__(self, name: str):
         super().__init__(name, self.INTRINSIC_ADDRESS)
 
+    def __eq__(self, __value):
+        return isinstance(__value, IntrinsicSymbol) and self.name == __value.name
+
+    def __hash__(self):
+        return hash(self.name)
+
     def __repr__(self):
         return f"intrinsic '{self.name}'"
 
@@ -324,6 +359,18 @@ def __init__(
         self.ssa_name = ssa_name
         super().__init__(tags)
 
+    def __eq__(self, __value):
+        return (
+            isinstance(__value, Variable)
+            and self._name == __value._name
+            and self.ssa_label == __value.ssa_label
+            and self._type == __value._type
+            and self.is_aliased == __value.is_aliased
+        )
+
+    def __hash__(self):
+        return hash((self._name, self.ssa_label, self._type, self.is_aliased))
+
     def __repr__(self) -> str:
         """Return a debug representation of the variable, which includes all the attributes"""
         return f"{self.name}#{self.ssa_label} (type: {self.type} aliased: {self.is_aliased})"
@@ -399,6 +446,12 @@ def __init__(
         self.initial_value = initial_value
         self.is_constant = is_constant
 
+    def __eq__(self, __value):
+        return isinstance(__value, GlobalVariable) and super().__eq__(__value)
+
+    def __hash__(self):
+        return super().__hash__()
+
     def copy(
         self,
         name: str = None,
@@ -445,6 +498,14 @@ def __init__(self, high: Variable, low: Variable, vartype: Type = UnknownType(),
         self._low = low
         self._type = vartype
 
+    def __eq__(self, __value):
+        return (
+            isinstance(__value, RegisterPair) and self._high == __value._high and self._low == __value._low and self._type == __value._type
+        )
+
+    def __hash__(self):
+        return hash((self._high, self._low, self._type))
+
     def __repr__(self) -> str:
         """Return debug representation of register pair"""
         return f"{repr(self._high)}:{repr(self._low)} type: {self.type}"
@@ -507,6 +568,12 @@ def __init__(self, value: list[Constant], vartype: DecompiledType = UnknownType(
             tags,
         )
 
+    def __eq__(self, __value):
+        return isinstance(__value, ConstantComposition) and super().__eq__(__value)
+
+    def __hash__(self):
+        return super().__hash__()
+
     def __str__(self) -> str:
         """Return a string representation of the ConstantComposition"""
         return "{" + ",".join([str(x) for x in self.value]) + "}"

decompiler/structures/pseudo/instructions.py

@@ -57,6 +57,12 @@ def __init__(self, comment: str, comment_style: str = "C", tags: Optional[Tuple[
         self._comment_style = comment_style
         self._open_comment, self._close_comment = self.STYLES.get(comment_style, self.STYLES[self.DEFAULT_STYLE])
 
+    def __eq__(self, __value):
+        return isinstance(__value, Comment) and self._comment == __value._comment and self._comment_style == __value._comment_style
+
+    def __hash__(self):
+        return hash((self._comment, self._comment_style))
+
     def __repr__(self) -> str:
         """Return representation of comment."""
         return f"{self._open_comment} {self._comment} {self._close_comment}"
@@ -161,6 +167,12 @@ def __init__(self, destination: Expression, value: Expression, tags: Optional[Tu
         """Init a new Assignment."""
         super(Assignment, self).__init__(destination, value, tags=tags)
 
+    def __eq__(self, __value):
+        return isinstance(__value, Assignment) and self._destination == __value._destination and self._value == __value._value
+
+    def __hash__(self):
+        return hash((self._destination, self._value))
+
     def __str__(self) -> str:
         """Return a string representation starting with the lhs."""
         if isinstance(self._destination, ListOperation) and not self._destination.operands:
@@ -211,6 +223,12 @@ def __init__(self, destination: Variable, value: Variable, tags: Optional[Tuple[
         """Init a new Relation."""
         super(Relation, self).__init__(destination, value, tags=tags)
 
+    def __eq__(self, __value):
+        return isinstance(__value, Relation) and self._destination == __value._destination and self._value == __value._value
+
+    def __hash__(self):
+        return hash((self._destination, self._value))
+
     def __str__(self) -> str:
         """Return a string representation starting with the lhs."""
         return f"{self.destination} -> {self.value}"
@@ -314,6 +332,12 @@ def __init__(self, condition: Condition, tags: Optional[Tuple[Tag, ...]] = None)
         """Init a new branch instruction."""
         super(Branch, self).__init__(condition, tags=tags)
 
+    def __eq__(self, __value):
+        return isinstance(__value, Branch) and self._condition == __value._condition
+
+    def __hash__(self):
+        return hash(self._condition)
+
     def __repr__(self) -> str:
         """Return a debug representation of a branch"""
         return f"if {repr(self.condition)}"
@@ -333,6 +357,12 @@ def __init__(self, condition: Expression, tags: Optional[Tuple[Tag, ...]] = None
         """Init a new branch instruction."""
         super(IndirectBranch, self).__init__(condition, tags=tags)
 
+    def __eq__(self, __value):
+        return isinstance(__value, IndirectBranch) and self._condition
+
+    def __hash__(self):
+        return hash(self._condition)
+
     def __repr__(self) -> str:
         """Return a debug representation of a branch"""
         return f"jmp {repr(self.condition)}"
@@ -355,6 +385,12 @@ def __init__(self, values, tags: Optional[Tuple[Tag, ...]] = None):
         super().__init__(tags)
         self._values = ListOperation(values)
 
+    def __eq__(self, __value):
+        return isinstance(__value, Return) and self._values == __value._values
+
+    def __hash__(self):
+        return hash(self._values)
+
     def __repr__(self) -> str:
         return f"return {repr(self._values)}"
 
@@ -395,6 +431,12 @@ def accept(self, visitor: DataflowObjectVisitorInterface[T]) -> T:
 
 
 class Break(Instruction):
+    def __eq__(self, __value):
+        return isinstance(__value, Break)
+
+    def __hash__(self):
+        return hash(Break)
+
     def __iter__(self) -> Iterator[Expression]:
         yield from ()
 
@@ -417,6 +459,12 @@ def accept(self, visitor: DataflowObjectVisitorInterface[T]) -> T:
 
 
 class Continue(Instruction):
+    def __eq__(self, __value):
+        return isinstance(__value, Continue)
+
+    def __hash__(self):
+        return hash(Continue)
+
     def __iter__(self) -> Iterator[Expression]:
         yield from ()
 
@@ -457,6 +505,12 @@ def __init__(
         self._origin_block = origin_block if origin_block else {}
         super().__init__(destination, ListOperation(value), tags=tags)
 
+    def __eq__(self, __value):
+        return isinstance(__value, Phi) and self._destination == __value._destination and self._value == __value._value
+
+    def __hash__(self):
+        return hash((self._destination, self._value))
+
     def __repr__(self):
         return f"{repr(self.destination)} = ϕ({repr(self.value)})"
 
@@ -516,6 +570,12 @@ class MemPhi(Phi):
     def __init__(self, destination_var: Variable, source_vars: Sequence[Variable], tags: Optional[Tuple[Tag, ...]] = None):
         super().__init__(destination_var, source_vars, tags=tags)
 
+    def __eq__(self, __value):
+        return isinstance(__value, MemPhi) and super().__eq__(__value)
+
+    def __hash__(self):
+        return super().__hash__()
+
     def __str__(self) -> str:
         return f"{self.destination} = ϕ({self.value})"