Merge pull request #1792 from xusheng6/binja_symtab

binja: add support for symtab names. Fix #1504

CHANGELOG.md

@@ -5,6 +5,7 @@
 ### New Features
 - ghidra: add Ghidra feature extractor and supporting code #1770 @colton-gabertan
 - ghidra: add entry script helping users run capa against a loaded Ghidra database #1767 @mike-hunhoff
+- binja: add support for symtab names #1504 @xusheng6
 
 ### Breaking Changes
 

capa/features/extractors/binja/file.py

@@ -125,15 +125,17 @@ def extract_file_function_names(bv: BinaryView) -> Iterator[Tuple[Feature, Addre
     """
     for sym_name in bv.symbols:
         for sym in bv.symbols[sym_name]:
-            if sym.type == SymbolType.LibraryFunctionSymbol:
-                name = sym.short_name
-                yield FunctionName(name), sym.address
-                if name.startswith("_"):
-                    # some linkers may prefix linked routines with a `_` to avoid name collisions.
-                    # extract features for both the mangled and un-mangled representations.
-                    # e.g. `_fwrite` -> `fwrite`
-                    # see: https://stackoverflow.com/a/2628384/87207
-                    yield FunctionName(name[1:]), sym.address
+            if sym.type not in [SymbolType.LibraryFunctionSymbol, SymbolType.FunctionSymbol]:
+                continue
+
+            name = sym.short_name
+            yield FunctionName(name), sym.address
+            if name.startswith("_"):
+                # some linkers may prefix linked routines with a `_` to avoid name collisions.
+                # extract features for both the mangled and un-mangled representations.
+                # e.g. `_fwrite` -> `fwrite`
+                # see: https://stackoverflow.com/a/2628384/87207
+                yield FunctionName(name[1:]), sym.address
 
 
 def extract_file_format(bv: BinaryView) -> Iterator[Tuple[Feature, Address]]:

capa/features/extractors/binja/function.py

@@ -7,8 +7,9 @@
 # See the License for the specific language governing permissions and limitations under the License.
 from typing import Tuple, Iterator
 
-from binaryninja import Function, BinaryView, RegisterValueType, LowLevelILOperation
+from binaryninja import Function, BinaryView, SymbolType, RegisterValueType, LowLevelILOperation
 
+from capa.features.file import FunctionName
 from capa.features.common import Feature, Characteristic
 from capa.features.address import Address, AbsoluteVirtualAddress
 from capa.features.extractors import loops
@@ -73,10 +74,31 @@ def extract_recursive_call(fh: FunctionHandle):
             yield Characteristic("recursive call"), fh.address
 
 
+def extract_function_name(fh: FunctionHandle):
+    """extract function names (e.g., symtab names)"""
+    func: Function = fh.inner
+    bv: BinaryView = func.view
+    if bv is None:
+        return
+
+    for sym in bv.get_symbols(func.start):
+        if sym.type not in [SymbolType.LibraryFunctionSymbol, SymbolType.FunctionSymbol]:
+            continue
+
+        name = sym.short_name
+        yield FunctionName(name), sym.address
+        if name.startswith("_"):
+            # some linkers may prefix linked routines with a `_` to avoid name collisions.
+            # extract features for both the mangled and un-mangled representations.
+            # e.g. `_fwrite` -> `fwrite`
+            # see: https://stackoverflow.com/a/2628384/87207
+            yield FunctionName(name[1:]), sym.address
+
+
 def extract_features(fh: FunctionHandle) -> Iterator[Tuple[Feature, Address]]:
     for func_handler in FUNCTION_HANDLERS:
         for feature, addr in func_handler(fh):
             yield feature, addr
 
 
-FUNCTION_HANDLERS = (extract_function_calls_to, extract_function_loop, extract_recursive_call)
+FUNCTION_HANDLERS = (extract_function_calls_to, extract_function_loop, extract_recursive_call, extract_function_name)

capa/features/extractors/binja/insn.py

@@ -94,28 +94,32 @@ def extract_insn_api_features(fh: FunctionHandle, bbh: BBHandle, ih: InsnHandle)
                 candidate_addrs.append(stub_addr)
 
             for address in candidate_addrs:
-                sym = func.view.get_symbol_at(address)
-                if sym is None or sym.type not in [SymbolType.ImportAddressSymbol, SymbolType.ImportedFunctionSymbol]:
-                    continue
+                for sym in func.view.get_symbols(address):
+                    if sym is None or sym.type not in [
+                        SymbolType.ImportAddressSymbol,
+                        SymbolType.ImportedFunctionSymbol,
+                        SymbolType.FunctionSymbol,
+                    ]:
+                        continue
 
-                sym_name = sym.short_name
+                    sym_name = sym.short_name
 
-                lib_name = ""
-                import_lib = bv.lookup_imported_object_library(sym.address)
-                if import_lib is not None:
-                    lib_name = import_lib[0].name
-                    if lib_name.endswith(".dll"):
-                        lib_name = lib_name[:-4]
-                    elif lib_name.endswith(".so"):
-                        lib_name = lib_name[:-3]
+                    lib_name = ""
+                    import_lib = bv.lookup_imported_object_library(sym.address)
+                    if import_lib is not None:
+                        lib_name = import_lib[0].name
+                        if lib_name.endswith(".dll"):
+                            lib_name = lib_name[:-4]
+                        elif lib_name.endswith(".so"):
+                            lib_name = lib_name[:-3]
 
-                for name in capa.features.extractors.helpers.generate_symbols(lib_name, sym_name):
-                    yield API(name), ih.address
-
-                if sym_name.startswith("_"):
-                    for name in capa.features.extractors.helpers.generate_symbols(lib_name, sym_name[1:]):
+                    for name in capa.features.extractors.helpers.generate_symbols(lib_name, sym_name):
                         yield API(name), ih.address
 
+                    if sym_name.startswith("_"):
+                        for name in capa.features.extractors.helpers.generate_symbols(lib_name, sym_name[1:]):
+                            yield API(name), ih.address
+
 
 def extract_insn_number_features(
     fh: FunctionHandle, bbh: BBHandle, ih: InsnHandle

tests/test_binja_features.py

@@ -36,7 +36,7 @@ except ImportError:
 @pytest.mark.skipif(binja_present is False, reason="Skip binja tests if the binaryninja Python API is not installed")
 @fixtures.parametrize(
     "sample,scope,feature,expected",
-    fixtures.FEATURE_PRESENCE_TESTS,
+    fixtures.FEATURE_PRESENCE_TESTS + fixtures.FEATURE_SYMTAB_FUNC_TESTS,
     indirect=["sample", "scope"],
 )
 def test_binja_features(sample, scope, feature, expected):