feat: Added in support for number-based date representations

Author: art1f1c3R

src/macaron/malware_analyzer/pypi_heuristics/metadata/anomalistic_version.py

@@ -37,6 +37,7 @@ class AnomalisticVersionAnalyzer(BaseHeuristicAnalyzer):
     """
 
     DETAIL_INFO_KEY: str = "versioning"
+    DIGIT_DATE_FORMATS: list[str] = ["%Y%m%d", "%Y%d%m", "%d%m%Y", "%m%d%Y", "%y%m%d", "%y%d%m", "%d%m%y", "%m%d%y"]
 
     def __init__(self) -> None:
         super().__init__(
@@ -146,10 +147,11 @@ def analyze(self, pypi_package_json: PyPIPackageJsonAsset) -> tuple[HeuristicRes
                 calendar = True
             else:
                 calendar_semantic = True
-        # check for calendar versioning: M[M].D[D].YY[YY]... or D[D].M[M].YY[YY]...
+        # check for calendar versioning: M[M].D[D].YY[YY]... or D[D].M[M].YY[YY]... or the whole digit rerpesenting a datetime
         elif (
-            (version.major in months and version.minor in days) or (version.major in days and version.minor in months)
-        ) and version.micro in years:
+            ((version.major in months and version.minor in days) or (version.major in days and version.minor in months))
+            and version.micro in years
+        ) or self.__integer_date(version.major, years, months, days):
             # must include day and year for this to be calendar
             calendar = True
 
@@ -180,6 +182,16 @@ def analyze(self, pypi_package_json: PyPIPackageJsonAsset) -> tuple[HeuristicRes
 
         return HeuristicResult.PASS, detail_info
 
+    def __integer_date(self, value: int, years: list[int], months: list[int], days: list[int]) -> bool:
+        for date_format in self.DIGIT_DATE_FORMATS:
+            if (date := parse_datetime(str(value), date_format)) is None:
+                continue
+
+            if date.year in years and date.month in months and date.day in days:
+                return True
+
+        return False
+
 
 class Versioning(Enum):
     """Enum used to assign different versioning methods."""

tests/malware_analyzer/pypi/test_anomalistic_version.py

@@ -67,6 +67,58 @@ def test_analyze_no_information(pypi_package_json: MagicMock) -> None:
         pytest.param(
             "16.10.16", "2016-10-13", HeuristicResult.PASS, Versioning.CALENDAR.value, id="test_calendar_MM.DD.YY_pass"
         ),
+        pytest.param(
+            "20161011.0",
+            "2016-10-13",
+            HeuristicResult.PASS,
+            Versioning.CALENDAR.value,
+            id="test_calendar_YYYYMMDD_pass",
+        ),
+        pytest.param(
+            "20161210.6.1",
+            "2016-10-13",
+            HeuristicResult.PASS,
+            Versioning.CALENDAR.value,
+            id="test_calendar_YYYYDDMM_pass",
+        ),
+        pytest.param(
+            "161013.9.0.5",
+            "2016-10-13",
+            HeuristicResult.PASS,
+            Versioning.CALENDAR.value,
+            id="test_calendar_YYDDMM_pass",
+        ),
+        pytest.param(
+            "161410.2.5.7",
+            "2016-10-13",
+            HeuristicResult.PASS,
+            Versioning.CALENDAR.value,
+            id="test_calendar_YYMMDD_pass",
+        ),
+        pytest.param(
+            "10102016.0",
+            "2016-10-13",
+            HeuristicResult.PASS,
+            Versioning.CALENDAR.value,
+            id="test_calendar_MMDDYYYY_pass",
+        ),
+        pytest.param(
+            "09102016",
+            "2016-10-13",
+            HeuristicResult.PASS,
+            Versioning.CALENDAR.value,
+            id="test_calendar_DDMMYYYY_pass",
+        ),
+        pytest.param(
+            "101516.5.7", "2016-10-13", HeuristicResult.PASS, Versioning.CALENDAR.value, id="test_calendar_DDMMYY_pass"
+        ),
+        pytest.param(
+            "161016.0.0.0.0",
+            "2016-10-13",
+            HeuristicResult.PASS,
+            Versioning.CALENDAR.value,
+            id="test_calendar_MMDDYY_pass",
+        ),
         pytest.param(
             "2!16.10.17.2.5.3",
             "2016-10-13",