From fffb1b4af77dc9c1863d01731167a16f6006ea56 Mon Sep 17 00:00:00 2001
From: Manuel Blatt <manuel.blatt@fkie.fraunhofer.de>
Date: Tue, 21 May 2024 17:55:29 +0200
Subject: [PATCH] add type names to struct defs and detect complex strings

---
 decompiler/backend/cexpressiongenerator.py | 42 ++++++++++++++++++++++
 decompiler/backend/variabledeclarations.py | 12 +++++--
 2 files changed, 52 insertions(+), 2 deletions(-)

diff --git a/decompiler/backend/cexpressiongenerator.py b/decompiler/backend/cexpressiongenerator.py
index 77cfbe77e..5496f69d4 100644
--- a/decompiler/backend/cexpressiongenerator.py
+++ b/decompiler/backend/cexpressiongenerator.py
@@ -15,6 +15,7 @@
 )
 from decompiler.structures.pseudo import instructions as instructions
 from decompiler.structures.pseudo import operations as operations
+from decompiler.structures.pseudo.complextypes import Struct
 from decompiler.structures.pseudo.operations import MemberAccess
 from decompiler.structures.visitors.interfaces import DataflowObjectVisitorInterface
 from decompiler.util.integer_util import normalize_int
@@ -22,6 +23,42 @@
 MAX_GLOBAL_INIT_LENGTH = 128
 
 
+def get_complex_string_struct_address_offset(vartype) -> str | None:
+    if not isinstance(vartype, Struct):
+        return None
+    if len(vartype.members) != 2:
+        return None
+    address_offset = None
+    length_offset = None
+    for offset, member in vartype.members.items():
+        match member.type:
+            case Pointer(type=Integer(size=8)):
+                address_offset = offset
+            case Integer():
+                length_offset = offset
+            case _:
+                return None
+    if address_offset is None or length_offset is None:
+        return None
+    return address_offset
+
+
+INLINE_COMPLEX_STRINGS = False
+DETECT_COMPLEX_STRINGS = True
+
+
+def is_complex_string_struct(vartype) -> bool:
+    if not DETECT_COMPLEX_STRINGS:
+        return False
+    return get_complex_string_struct_address_offset(vartype) is not None
+
+
+def get_data_of_complex_string_struct(variable) -> str:
+    address_offset = get_complex_string_struct_address_offset(variable.type)
+    address = variable.initial_value.value[address_offset]
+    return address
+
+
 def inline_global_variable(var) -> bool:
     if not var.is_constant:
         return False
@@ -29,6 +66,9 @@ def inline_global_variable(var) -> bool:
         case ArrayType():
             if var.type.type in [Integer.char(), CustomType.wchar16(), CustomType.wchar32()]:
                 return True
+        case Struct():
+            if INLINE_COMPLEX_STRINGS and is_complex_string_struct(var.type):
+                return True
         case _:
             return False
     return False
@@ -212,6 +252,8 @@ def visit_variable(self, expr: expressions.Variable) -> str:
     def visit_global_variable(self, expr: expressions.GlobalVariable):
         """Inline a global variable if its initial value is constant and not of void type"""
         if inline_global_variable(expr):
+            if is_complex_string_struct(expr.type):
+                return self.visit(get_data_of_complex_string_struct(expr))
             return self.visit(expr.initial_value)
         return expr.name
 
diff --git a/decompiler/backend/variabledeclarations.py b/decompiler/backend/variabledeclarations.py
index 7b2d8926b..efd08a37d 100644
--- a/decompiler/backend/variabledeclarations.py
+++ b/decompiler/backend/variabledeclarations.py
@@ -3,7 +3,12 @@
 from collections import defaultdict
 from typing import Iterable, Iterator, List
 
-from decompiler.backend.cexpressiongenerator import CExpressionGenerator, inline_global_variable
+from decompiler.backend.cexpressiongenerator import (
+    CExpressionGenerator,
+    get_data_of_complex_string_struct,
+    inline_global_variable,
+    is_complex_string_struct,
+)
 from decompiler.structures.ast.syntaxtree import AbstractSyntaxTree
 from decompiler.structures.pseudo import GlobalVariable, Integer, Variable
 from decompiler.structures.pseudo.complextypes import Struct
@@ -69,7 +74,10 @@ def _generate_definitions(global_variables: set[GlobalVariable]) -> Iterator[str
                         br, bl = "{", "}"
                     yield f"{base}{variable.type.type} {variable.name}[{hex(variable.type.elements)}] = {br}{CExpressionGenerator().visit(variable.initial_value)}{bl};"
                 case Struct():
-                    string = f"struct {variable.name}" + "{\n"
+                    if is_complex_string_struct(variable.type):
+                        yield base + f"struct {variable.type.name} {variable.name} = {CExpressionGenerator().visit(get_data_of_complex_string_struct(variable))};"
+                        continue
+                    string = f"struct {variable.type.name} {variable.name}" + "{\n"
                     for m_type, m_value in zip(variable.type.members.values(), variable.initial_value.value.values()):
                         value = CExpressionGenerator().visit(m_value)
                         string += f"\t.{m_type.name} = {value};\n"