Merge branch 'master' into dynamic-feature-extraction

.github/pyinstaller/pyinstaller.spec

@@ -17,6 +17,7 @@ a = Analysis(
         # when invoking pyinstaller from the project root,
         # this gets invoked from the directory of the spec file,
         # i.e. ./.github/pyinstaller
+        ("../../assets", "assets"),
         ("../../rules", "rules"),
         ("../../sigs", "sigs"),
         ("../../cache", "cache"),

CHANGELOG.md

@@ -11,6 +11,7 @@
 - ghidra: add entry script helping users run capa against a loaded Ghidra database #1767 @mike-hunhoff
 - binja: add support for forwarded exports #1646 @xusheng6
 - binja: add support for symtab names #1504 @xusheng6
+- add com class/interface features #322 @Aayush-goel-04
 
 ### Breaking Changes
 
@@ -1606,4 +1607,4 @@ Download a standalone binary below and checkout the readme [here on GitHub](http
 ### Raw diffs
 
   - [capa v1.0.0...v1.1.0](https://github.com/mandiant/capa/compare/v1.0.0...v1.1.0)
-  - [capa-rules v1.0.0...v1.1.0](https://github.com/mandiant/capa-rules/compare/v1.0.0...v1.1.0)
+  - [capa-rules v1.0.0...v1.1.0](https://github.com/mandiant/capa-rules/compare/v1.0.0...v1.1.0)

capa/rules/__init__.py

@@ -8,6 +8,8 @@
 
 import io
 import re
+import gzip
+import json
 import uuid
 import codecs
 import logging
@@ -326,6 +328,68 @@ def ensure_feature_valid_for_scopes(scopes: Scopes, feature: Union[Feature, Stat
             raise InvalidRule(f"feature {feature} not supported for scopes {scopes}")
 
 
+class ComType(Enum):
+    CLASS = "class"
+    INTERFACE = "interface"
+
+
+# COM data source https://github.com/stevemk14ebr/COM-Code-Helper/tree/master
+VALID_COM_TYPES = {
+    ComType.CLASS: {"db_path": "assets/classes.json.gz", "prefix": "CLSID_"},
+    ComType.INTERFACE: {"db_path": "assets/interfaces.json.gz", "prefix": "IID_"},
+}
+
+
+@lru_cache(maxsize=None)
+def load_com_database(com_type: ComType) -> Dict[str, List[str]]:
+    com_db_path: Path = capa.main.get_default_root() / VALID_COM_TYPES[com_type]["db_path"]
+
+    if not com_db_path.exists():
+        raise IOError(f"COM database path '{com_db_path}' does not exist or cannot be accessed")
+
+    try:
+        with gzip.open(com_db_path, "rb") as gzfile:
+            return json.loads(gzfile.read().decode("utf-8"))
+    except Exception as e:
+        raise IOError(f"Error loading COM database from '{com_db_path}'") from e
+
+
+def translate_com_feature(com_name: str, com_type: ComType) -> ceng.Or:
+    com_db = load_com_database(com_type)
+    guid_strings: Optional[List[str]] = com_db.get(com_name)
+    if guid_strings is None or len(guid_strings) == 0:
+        logger.error(" %s doesn't exist in COM %s database", com_name, com_type)
+        raise InvalidRule(f"'{com_name}' doesn't exist in COM {com_type} database")
+
+    com_features: List = []
+    for guid_string in guid_strings:
+        hex_chars = guid_string.replace("-", "")
+        h = [hex_chars[i : i + 2] for i in range(0, len(hex_chars), 2)]
+        reordered_hex_pairs = [
+            h[3],
+            h[2],
+            h[1],
+            h[0],
+            h[5],
+            h[4],
+            h[7],
+            h[6],
+            h[8],
+            h[9],
+            h[10],
+            h[11],
+            h[12],
+            h[13],
+            h[14],
+            h[15],
+        ]
+        guid_bytes = bytes.fromhex("".join(reordered_hex_pairs))
+        prefix = VALID_COM_TYPES[com_type]["prefix"]
+        com_features.append(capa.features.common.StringFactory(guid_string, f"{prefix+com_name} as GUID string"))
+        com_features.append(capa.features.common.Bytes(guid_bytes, f"{prefix+com_name} as bytes"))
+    return ceng.Or(com_features)
+
+
 def parse_int(s: str) -> int:
     if s.startswith("0x"):
         return int(s, 0x10)
@@ -742,6 +806,13 @@ def build_statements(d, scopes: Scopes):
         ensure_feature_valid_for_scopes(scopes, feature)
         return feature
 
+    elif key.startswith("com/"):
+        com_type = str(key[len("com/") :]).upper()
+        if com_type not in [item.name for item in ComType]:
+            raise InvalidRule(f"unexpected COM type: {com_type}")
+        value, description = parse_description(d[key], key, d.get("description"))
+        return translate_com_feature(value, ComType[com_type])
+
     else:
         Feature = parse_feature(key)
         value, description = parse_description(d[key], key, d.get("description"))

tests/fixtures.py

@@ -318,6 +318,8 @@ def get_data_path_by_name(name) -> Path:
         return CD / "data" / "499c2a85f6e8142c3f48d4251c9c7cd6.raw32"
     elif name.startswith("9324d"):
         return CD / "data" / "9324d1a8ae37a36ae560c37448c9705a.exe_"
+    elif name.startswith("395eb"):
+        return CD / "data" / "395eb0ddd99d2c9e37b6d0b73485ee9c.exe_"
     elif name.startswith("a1982"):
         return CD / "data" / "a198216798ca38f280dc413f8c57f2c2.exe_"
     elif name.startswith("a933a"):
@@ -1346,6 +1348,11 @@ def z9324d_extractor():
     return get_extractor(get_data_path_by_name("9324d..."))
 
 
+@pytest.fixture
+def z395eb_extractor():
+    return get_extractor(get_data_path_by_name("395eb..."))
+
+
 @pytest.fixture
 def pma12_04_extractor():
     return get_extractor(get_data_path_by_name("pma12-04"))

tests/test_main.py

@@ -401,6 +401,30 @@ def test_byte_matching(z9324d_extractor):
     assert "byte match test" in capabilities
 
 
+def test_com_feature_matching(z395eb_extractor):
+    rules = capa.rules.RuleSet(
+        [
+            capa.rules.Rule.from_yaml(
+                textwrap.dedent(
+                    """
+                    rule:
+                        meta:
+                            name: initialize IWebBrowser2
+                            scope: basic block
+                        features:
+                            - and:
+                                - api: ole32.CoCreateInstance
+                                - com/class: InternetExplorer #bytes: 01 DF 02 00 00 00 00 00 C0 00 00 00 00 00 00 46 = CLSID_InternetExplorer
+                                - com/interface: IWebBrowser2 #bytes: 61 16 0C D3 AF CD D0 11 8A 3E 00 C0 4F C9 E2 6E = IID_IWebBrowser2
+                    """
+                )
+            )
+        ]
+    )
+    capabilities, meta = capa.main.find_capabilities(rules, z395eb_extractor)
+    assert "initialize IWebBrowser2" in capabilities
+
+
 def test_count_bb(z9324d_extractor):
     rules = capa.rules.RuleSet(
         [

tests/test_rules.py

@@ -1531,3 +1531,72 @@ def test_property_access_symbol():
         )
         is True
     )
+
+
+def test_translate_com_features():
+    r = capa.rules.Rule.from_yaml(
+        textwrap.dedent(
+            """
+            rule:
+                meta:
+                    name: test rule
+                features:
+                    - com/class: WICPngDecoder
+                    # 389ea17b-5078-4cde-b6ef-25c15175c751 WICPngDecoder
+                    # e018945b-aa86-4008-9bd4-6777a1e40c11 WICPngDecoder
+            """
+        )
+    )
+    com_name = "WICPngDecoder"
+    com_features = [
+        capa.features.common.Bytes(b"{\xa1\x9e8xP\xdeL\xb6\xef%\xc1Qu\xc7Q", f"CLSID_{com_name} as bytes"),
+        capa.features.common.StringFactory("389ea17b-5078-4cde-b6ef-25c15175c751", f"CLSID_{com_name} as GUID string"),
+        capa.features.common.Bytes(b"[\x94\x18\xe0\x86\xaa\x08@\x9b\xd4gw\xa1\xe4\x0c\x11", f"IID_{com_name} as bytes"),
+        capa.features.common.StringFactory("e018945b-aa86-4008-9bd4-6777a1e40c11", f"IID_{com_name} as GUID string"),
+    ]
+    assert set(com_features) == set(r.statement.get_children())
+
+
+def test_invalid_com_features():
+    # test for unknown COM class
+    with pytest.raises(capa.rules.InvalidRule):
+        _ = capa.rules.Rule.from_yaml(
+            textwrap.dedent(
+                """
+                rule:
+                    meta:
+                        name: test rule
+                    features:
+                        - com/class: invalid_com
+                """
+            )
+        )
+
+    # test for unknown COM interface
+    with pytest.raises(capa.rules.InvalidRule):
+        _ = capa.rules.Rule.from_yaml(
+            textwrap.dedent(
+                """
+                rule:
+                    meta:
+                        name: test rule
+                    features:
+                        - com/interface: invalid_com
+                """
+            )
+        )
+
+    # test for invalid COM type
+    # valid_com_types = "class", "interface"
+    with pytest.raises(capa.rules.InvalidRule):
+        _ = capa.rules.Rule.from_yaml(
+            textwrap.dedent(
+                """
+                rule:
+                    meta:
+                        name: test rule
+                    features:
+                        - com/invalid_COM_type: WICPngDecoder
+                """
+            )
+        )