Add a new Decompilation artifact with mappings

.github/workflows/dec-tests.yml

@@ -15,7 +15,7 @@ on:
 jobs:
   build:
     runs-on: ubuntu-latest
-    timeout-minutes: 10
+    timeout-minutes: 15
     steps:
     - uses: actions/checkout@v2
     - name: Set up Python 3.10
@@ -48,4 +48,4 @@ jobs:
       run: |
         # these two test must be run in separate python environments, due to the way ghidra bridge works
         # you also must run these tests in the exact order shown here
-        pytest ./tests/test_remote_ghidra.py && pytest ./tests/test_decompilers.py
+        pytest ./tests/test_remote_ghidra.py && pytest ./tests/test_decompilers.py -s

libbs/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "1.12.1"
+__version__ = "1.13.0"
 
 
 import logging

libbs/api/artifact_lifter.py

@@ -60,7 +60,7 @@ def lower_stack_offset(self, offset: int, func_addr: int) -> int:
     #
 
     def _lift_or_lower_artifact(self, artifact, mode):
-        target_attrs = ("type", "offset", "addr", "func_addr")
+        target_attrs = ("type", "offset", "addr", "func_addr", "line_map")
         if mode not in ("lower", "lift"):
             return None
 
@@ -81,6 +81,14 @@ def _lift_or_lower_artifact(self, artifact, mode):
                         continue
                     lifting_func = getattr(self, f"{mode}_stack_offset")
                     setattr(lifted_art, attr, lifting_func(curr_val, lifted_art.addr))
+                # special handling for decompilation
+                elif attr == "line_map":
+                    lifted_line_map = {}
+                    lift_or_lower_func = self.lift_addr if mode == "lift" else self.lower_addr
+                    for k, v in curr_val.items():
+                        lifted_line_map[k] = {lift_or_lower_func(_v) for _v in v}
+
+                    setattr(lifted_art, attr, lifted_line_map)
                 else:
                     attr_func_name = attr if attr != "func_addr" else "addr"
                     lifting_func = getattr(self, f"{mode}_{attr_func_name}")

libbs/api/decompiler_interface.py

@@ -17,7 +17,7 @@
     Artifact,
     Function, FunctionHeader, StackVariable,
     Comment, GlobalVariable, Patch,
-    Enum, Struct, FunctionArgument
+    Enum, Struct, FunctionArgument, Decompilation
 )
 from libbs.decompilers import SUPPORTED_DECOMPILERS, ANGR_DECOMPILER, \
     BINJA_DECOMPILER, IDA_DECOMPILER, GHIDRA_DECOMPILER
@@ -110,6 +110,7 @@ def __init__(
         else:
             self._init_headless_components()
 
+        self.info(f"Using configuration file: {self.config.save_location}")
         self.config.save()
 
     def _init_headless_components(self, *args, check_dec_path=True, **kwargs):
@@ -270,30 +271,37 @@ def decompiler_available(self) -> bool:
         """
         return True
 
-    def decompile(self, addr: int) -> Optional[str]:
+    def decompile(self, addr: int, map_lines=False, **kwargs) -> Optional[Decompilation]:
         lowered_addr = self.art_lifter.lower_addr(addr)
         if not self.decompiler_available:
             _l.error("Decompiler is not available.")
             return None
 
-        # TODO: make this a function call after transitioning decompiler artifacts to LiveState
         sorted_funcs = sorted(self._functions().items(), key=lambda x: x[0])
-        for func_addr, func in sorted_funcs:
-            if func.addr <= lowered_addr < (func.addr + func.size):
-                break
+        func_by_addr = {_addr: func for _addr, func in sorted_funcs}
+        func = None
+        if lowered_addr in func_by_addr:
+            func = func_by_addr[lowered_addr]
         else:
-            func = None
+            _l.debug(f"Address is not a function start, searching for function...")
+            for func_addr, _func in sorted_funcs:
+                if _func.addr <= lowered_addr < (_func.addr + _func.size):
+                    func = _func
+                    break
 
         if func is None:
             self.warning(f"Failed to find function for address {hex(lowered_addr)}")
             return None
 
         try:
-            decompilation = self._decompile(func)
+            decompilation = self._decompile(func, map_lines=map_lines, **kwargs)
         except Exception as e:
             self.warning(f"Failed to decompile function at {hex(lowered_addr)}: {e}")
             decompilation = None
 
+        if decompilation is not None:
+            decompilation = self.art_lifter.lift(decompilation)
+
         return decompilation
 
     def xrefs_to(self, artifact: Artifact) -> List[Artifact]:
@@ -307,7 +315,7 @@ def xrefs_to(self, artifact: Artifact) -> List[Artifact]:
     def get_func_containing(self, addr: int) -> Optional[Function]:
         raise NotImplementedError
 
-    def _decompile(self, function: Function) -> Optional[str]:
+    def _decompile(self, function: Function, map_lines=False, **kwargs) -> Optional[Decompilation]:
         raise NotImplementedError
 
     def get_decompilation_object(self, function: Function, **kwargs) -> Optional[object]:

libbs/artifacts/decompilation.py

@@ -6,24 +6,27 @@
 class Decompilation(Artifact):
     __slots__ = Artifact.__slots__ + (
         "addr",
-        "decompilation",
+        "text",
+        "line_map",
         "decompiler",
     )
 
     def __init__(
         self,
         addr: int = None,
-        decompilation: str = None,
+        text: str = None,
+        line_map: dict = None,
         decompiler: str = None,
         **kwargs
     ):
         super().__init__(**kwargs)
         self.addr = addr
-        self.decompilation = decompilation
+        self.text = text
+        self.line_map = line_map or {}
         self.decompiler = decompiler
 
     def __str__(self):
-        return f"//ADDR: {hex(self.addr)}\n// SOURCE: {self.decompiler}\n{self.decompilation}"
+        return f"//ADDR: {hex(self.addr)}\n// SOURCE: {self.decompiler}\n{self.text}"
 
     def __repr__(self):
-        return f"<Decompilation: {self.decompiler}@{hex(self.addr)} len={len(self.decompilation)}>"
+        return f"<Decompilation: {self.decompiler}@{hex(self.addr)} len={len(self.text)}>"

libbs/configuration.py

@@ -46,7 +46,7 @@ def save(self):
             with open(self.save_location, "w") as fp:
                 toml.dump(dump_dict, fp)
 
-        _l.info(f"Saved config to {self.save_location}")
+        _l.debug(f"Saved config to %s", self.save_location)
         return True
 
     def load(self):

libbs/decompilers/angr/compat.py

@@ -1,5 +1,6 @@
 # pylint: disable=wrong-import-position,wrong-import-order
 import logging
+from collections import defaultdict
 from typing import Optional
 
 from angrmanagement.plugins import BasePlugin
@@ -94,23 +95,23 @@ def handle_stack_var_renamed(self, func, offset, old_name, new_name):
         if func is None:
             return False
 
-        decompilation = self.interface.decompile_function(func)
+        decompilation = self.interface.decompile_function(func).codegen
         stack_var = self.interface.find_stack_var_in_codegen(decompilation, offset)
         print("handle_stack_var_renamed signal sent out to everyone")
         self.interface.stack_variable_changed(StackVariable(offset, new_name, None, stack_var.size, func.addr))
         return True
 
     # pylint: disable=unused-argument
     def handle_stack_var_retyped(self, func, offset, old_type, new_type):
-        decompilation = self.interface.decompile_function(func)
+        decompilation = self.interface.decompile_function(func).codegen
         stack_var = self.interface.find_stack_var_in_codegen(decompilation, offset)
         var_type = AngrInterface.stack_var_type_str(decompilation, stack_var)
         self.interface.stack_variable_changed(StackVariable(offset, stack_var.name, var_type, stack_var.size, func.addr))
         return True
 
     # pylint: disable=unused-argument
     def handle_func_arg_renamed(self, func, offset, old_name, new_name):
-        decompilation = self.interface.decompile_function(func)
+        decompilation = self.interface.decompile_function(func).codegen
         func_args = AngrInterface.func_args_as_libbs_args(decompilation)
         self.interface.function_header_changed(
             FunctionHeader(
@@ -127,7 +128,7 @@ def handle_func_arg_renamed(self, func, offset, old_name, new_name):
 
     # pylint: disable=unused-argument
     def handle_func_arg_retyped(self, func, offset, old_type, new_type):
-        decompilation = self.interface.decompile_function(func)
+        decompilation = self.interface.decompile_function(func).codegen
         func_args = AngrInterface.func_args_as_libbs_args(decompilation)
         self.interface.function_header_changed(
             FunctionHeader(