Merge branch 'master' into pre-commit

floss/main.py

@@ -537,7 +537,11 @@ def main(argv=None) -> int:
     # can throw away result later if not desired in output
     time0 = time()
     interim = time0
+
     static_strings = get_static_strings(sample, args.min_length)
+    if static_strings == []:
+        return 0
+
     static_runtime = get_runtime_diff(interim)
 
     lang_id = identify_language(sample, static_strings)

floss/results.py

@@ -139,8 +139,6 @@ def from_utf8(cls, buf, addr, min_length):
         except UnicodeDecodeError:
             raise ValueError("not utf-8")
 
-        if not decoded_string.isprintable():
-            raise ValueError("not printable")
         if len(decoded_string) < min_length:
             raise ValueError("too short")
         return cls(string=decoded_string, offset=addr, encoding=StringEncoding.UTF8)

floss/utils.py

@@ -541,6 +541,11 @@ def get_static_strings(sample: Path, min_length: int) -> list:
     """
     Returns list of static strings from the file which are above the minimum length
     """
+
+    if sample.stat().st_size == 0:
+        logger.warning("File is empty")
+        return []
+
     with sample.open("r") as f:
         if hasattr(mmap, "MAP_PRIVATE"):
             # unix

setup.py

@@ -57,7 +57,7 @@
     extras_require={
         "dev": [
             "pre-commit==2.21.0",
-            "pyyaml==6.0",
+            "pyyaml==6.0.1",
             "pytest==7.4.0",
             "pytest-sugar==0.9.4",
             "pytest-instafail==0.5.0",
@@ -68,7 +68,7 @@
             "mypy==1.4.1",
             # type stubs for mypy
             "types-PyYAML==6.0.10",
-            "types-tabulate==0.9.0.2",
+            "types-tabulate==0.9.0.3",
         ],
         "build": [
             "pyinstaller==5.13.0",

tests/test_language_extract_go.py

@@ -9,14 +9,14 @@
 @pytest.fixture(scope="module")
 def go_strings32():
     n = 6
-    path = pathlib.Path(__file__).parent / "data" / "src" / "go-hello" / "bin" / "go-hello.exe"
+    path = pathlib.Path(__file__).parent / "data" / "language" / "go" / "go-hello" / "bin" / "go-hello.exe"
     return extract_go_strings(path, n)
 
 
 @pytest.fixture(scope="module")
 def go_strings64():
     n = 6
-    path = pathlib.Path(__file__).parent / "data" / "src" / "go-hello" / "bin" / "go-hello64.exe"
+    path = pathlib.Path(__file__).parent / "data" / "language" / "go" / "go-hello" / "bin" / "go-hello64.exe"
     return extract_go_strings(path, n)
 
 
@@ -124,6 +124,21 @@ def test_mov_lea_mov(request, string, offset, encoding, go_strings):
     assert StaticString(string=string, offset=offset, encoding=encoding) in request.getfixturevalue(go_strings)
 
 
+@pytest.mark.parametrize(
+    "string,offset,encoding,go_strings",
+    [
+        # 00000000004AB080  20 6E 6F 74 20 66 6F 75  6E 64 20 6D 61 72 6B 72   not found markr
+        # 00000000004AB090  6F 6F 74 20 6A 6F 62 73  20 64 6F 6E 65 0A 20 74  oot jobs done. t
+        pytest.param(" markroot jobs done\n", 0xAA68A, StringEncoding.UTF8, "go_strings64"),
+        # 004A3DE0  66 6F 75 6E 64 20 6D 61  72 6B 72 6F 6F 74 20 6A  found markroot j
+        # 004A3DF0  6F 62 73 20 64 6F 6E 65  0A 20 74 6F 20 75 6E 61  obs done. to una
+        pytest.param(" markroot jobs done\n", 0xA23E5, StringEncoding.UTF8, "go_strings32"),
+    ],
+)
+def test_strings_with_newline_char_0A(request, string, offset, encoding, go_strings):
+    assert StaticString(string=string, offset=offset, encoding=encoding) in request.getfixturevalue(go_strings)
+
+
 @pytest.mark.skip(reason="not extracted via go_strings")
 @pytest.mark.parametrize(
     "string,offset,encoding,go_strings",

tests/test_language_id.py

@@ -10,10 +10,10 @@
 @pytest.mark.parametrize(
     "binary_file, expected_result",
     [
-        ("data/src/go-hello/bin/go-hello.exe", Language.GO),
-        ("data/src/rust-hello/bin/rust-hello.exe", Language.RUST),
+        ("data/language/go/go-hello/bin/go-hello.exe", Language.GO),
+        ("data/language/rust/rust-hello/bin/rust-hello.exe", Language.RUST),
         ("data/test-decode-to-stack.exe", Language.UNKNOWN),
-        ("data/src/dotnet-hello/bin/dotnet-hello.exe", Language.DOTNET),
+        ("data/language/dotnet/dotnet-hello/bin/dotnet-hello.exe", Language.DOTNET),
         ("data/src/shellcode-stackstrings/bin/shellcode-stackstrings.bin", Language.UNKNOWN),
     ],
 )