Fix escaping of string literals (#428)

* Fix string literal escaping * Run Black obfuscation * Revert _format_string_literal back * Revert _format_string_literal back 2
fkie-cad · Aug 1, 2024 · ad2f8a3 · ad2f8a3
1 parent 3ed0094
commit ad2f8a3
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 4 deletions.
diff --git a/decompiler/backend/cexpressiongenerator.py b/decompiler/backend/cexpressiongenerator.py
@@ -163,6 +163,10 @@ class CExpressionGenerator(DataflowObjectVisitorInterface):
         # OperationType.adc: "adc",
     }
 
+    ESCAPE_TABLE = str.maketrans(
+        {"\\": r"\\", '"': r"\"", "'": r"\'", "\n": r"\n", "\r": r"\r", "\t": r"\t", "\v": r"\v", "\b": r"\b", "\f": r"\f", "\0": r"\0"}
+    )
+
     def visit_unknown_expression(self, expr: expressions.UnknownExpression) -> str:
         """Return the error message for this UnknownExpression."""
         return expr.msg
@@ -197,16 +201,16 @@ def visit_constant_composition(self, expr: expressions.ConstantComposition):
         """Visit a Constant Array."""
         match expr.type.type:
             case CustomType(text="wchar16") | CustomType(text="wchar32"):
-                val = "".join([x.value for x in expr.value])
+                val = "".join([x.value for x in expr.value]).translate(self.ESCAPE_TABLE)
                 return f'L"{val}"' if len(val) <= MAX_GLOBAL_INIT_LENGTH else f'L"{val[:MAX_GLOBAL_INIT_LENGTH]}..."'
             case Integer(size=8, signed=False):
                 val = "".join([f"\\x{x.value:02X}" for x in expr.value][:MAX_GLOBAL_INIT_LENGTH])
                 return f'"{val}"' if len(val) <= MAX_GLOBAL_INIT_LENGTH else f'"{val[:MAX_GLOBAL_INIT_LENGTH]}..."'
             case Integer(8):
-                val = "".join([x.value for x in expr.value][:MAX_GLOBAL_INIT_LENGTH])
+                val = "".join([x.value for x in expr.value][:MAX_GLOBAL_INIT_LENGTH]).translate(self.ESCAPE_TABLE)
                 return f'"{val}"' if len(val) <= MAX_GLOBAL_INIT_LENGTH else f'"{val[:MAX_GLOBAL_INIT_LENGTH]}..."'
             case _:
-                return f'{", ".join([self.visit(x) for x in expr.value])}'  # Todo: Should we print every member? Could get pretty big
+                return f'{", ".join([self.visit(x) for x in expr.value]).translate(self.ESCAPE_TABLE)}'  # Todo: Should we print every member? Could get pretty big
 
     def visit_variable(self, expr: expressions.Variable) -> str:
         """Return a string representation of the variable."""

diff --git a/decompiler/backend/codevisitor.py b/decompiler/backend/codevisitor.py
@@ -251,7 +251,11 @@ def _condition_string(self, condition: ConditionVar) -> str:
     def _format_integer_literal(self, type_info: Integer, value: int) -> str:
         """Format the integer based on the codegenerators settings."""
 
-        byte_format_handler = {"char": lambda x: f"'{chr(x)}'", "hex": lambda x: f"{hex(x)}", "dec": lambda x: f"{x}"}
+        byte_format_handler = {
+            "char": lambda x: f"'{chr(x).translate(self.ESCAPE_TABLE)}'",
+            "hex": lambda x: f"{hex(x)}",
+            "dec": lambda x: f"{x}",
+        }
         if self._possibly_char_in_ascii_range(type_info, value):
             if value_handler := byte_format_handler.get(self._byte_format, None):
                 if hint_handler := byte_format_handler.get(self._byte_format_hint, None):