feat(heuristics): add two analyzers to detect dependency confusion and distinguish from stub packages (#1117)

This PR adds three new heuristic analyzers designed to detect potential dependency confusion attacks and differentiate them from harmless stub or placeholder packages.

Signed-off-by: Amine <[email protected]>

Author: AmineRaouane

src/macaron/malware_analyzer/README.md

@@ -70,6 +70,17 @@ When a heuristic fails, with `HeuristicResult.FAIL`, then that is an indicator b
     - **Rule**: Return `HeuristicResult.FAIL` if the email is invalid; otherwise, return `HeuristicResult.PASS`.
     - **Dependency**: None.
 
+
+12. **Minimal Content**
+    - **Description**: Checks if the package has a small number of files.
+    - **Rule**: Return `HeuristicResult.FAIL` if the number of files is strictly less than FILES_THRESHOLD; otherwise, return `HeuristicResult.PASS`.
+    - **Dependency**: None.
+
+13. **Unsecure Description**
+    - **Description**: Checks if the package description is unsecure, such as not having a descriptive keywords that indicates its a stub package .
+    - **Rule**: Return `HeuristicResult.FAIL` if no descriptive word is found in the package description or summary ; otherwise, return `HeuristicResult.PASS`.
+    - **Dependency**: None.
+
 ### Source Code Analysis with Semgrep
 **PyPI Source Code Analyzer**
 - **Description**: Uses Semgrep, with default rules written in `src/macaron/resources/pypi_malware_rules` and custom rules available by supplying a path to `custom_semgrep_rules` in `defaults.ini`, to scan the package `.tar` source code.

src/macaron/malware_analyzer/pypi_heuristics/heuristics.py

@@ -49,6 +49,12 @@ class Heuristics(str, Enum):
     #: Indicates that the package has a similar structure to other packages maintained by the same user.
     SIMILAR_PROJECTS = "similar_projects"
 
+    #: Indicates that the package has minimal content.
+    MINIMAL_CONTENT = "minimal_content"
+
+    #: Indicates that the package's description is unsecure, such as not having a descriptive keywords.
+    UNSECURE_DESCRIPTION = "unsecure_description"
+
 
 class HeuristicResult(str, Enum):
     """Result type indicating the outcome of a heuristic."""

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

@@ -0,0 +1,54 @@
+# Copyright (c) 2024 - 2025, Oracle and/or its affiliates. All rights reserved.
+# Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/.
+
+"""This analyzer checks if a PyPI package has minimal content."""
+
+import logging
+import os
+
+from macaron.errors import SourceCodeError
+from macaron.json_tools import JsonType
+from macaron.malware_analyzer.pypi_heuristics.base_analyzer import BaseHeuristicAnalyzer
+from macaron.malware_analyzer.pypi_heuristics.heuristics import HeuristicResult, Heuristics
+from macaron.slsa_analyzer.package_registry.pypi_registry import PyPIPackageJsonAsset
+
+logger: logging.Logger = logging.getLogger(__name__)
+
+
+class MinimalContentAnalyzer(BaseHeuristicAnalyzer):
+    """Check whether the package has minimal content."""
+
+    FILES_THRESHOLD = 50
+
+    def __init__(self) -> None:
+        super().__init__(
+            name="minimal_content_analyzer",
+            heuristic=Heuristics.MINIMAL_CONTENT,
+            depends_on=None,
+        )
+
+    def analyze(self, pypi_package_json: PyPIPackageJsonAsset) -> tuple[HeuristicResult, dict[str, JsonType]]:
+        """Analyze the package.
+
+        Parameters
+        ----------
+        pypi_package_json: PyPIPackageJsonAsset
+            The PyPI package JSON asset object.
+
+        Returns
+        -------
+        tuple[HeuristicResult, dict[str, JsonType]]:
+            The result and related information collected during the analysis.
+        """
+        result = pypi_package_json.download_sourcecode()
+        if not result:
+            error_msg = "No source code files have been downloaded"
+            logger.debug(error_msg)
+            raise SourceCodeError(error_msg)
+
+        file_count = sum(len(files) for _, _, files in os.walk(pypi_package_json.package_sourcecode_path))
+
+        if file_count >= self.FILES_THRESHOLD:
+            return HeuristicResult.PASS, {"message": "Package has sufficient content"}
+
+        return HeuristicResult.FAIL, {"message": "Not enough files found"}

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

@@ -0,0 +1,56 @@
+# Copyright (c) 2024 - 2025, Oracle and/or its affiliates. All rights reserved.
+# Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/.
+
+"""This analyzer checks if a PyPI package has unsecure description."""
+
+import logging
+import re
+
+from macaron.errors import HeuristicAnalyzerValueError
+from macaron.json_tools import JsonType, json_extract
+from macaron.malware_analyzer.pypi_heuristics.base_analyzer import BaseHeuristicAnalyzer
+from macaron.malware_analyzer.pypi_heuristics.heuristics import HeuristicResult, Heuristics
+from macaron.slsa_analyzer.package_registry.pypi_registry import PyPIPackageJsonAsset
+
+logger: logging.Logger = logging.getLogger(__name__)
+
+
+class UnsecureDescriptionAnalyzer(BaseHeuristicAnalyzer):
+    """Check whether the package's description is unsecure."""
+
+    SECURE_DESCRIPTION_REGEX = re.compile(
+        r"\b(?:internal|private|stub|placeholder|dependency confusion|security|namespace protection|reserved|harmless|prevent)\b",
+        re.IGNORECASE,
+    )
+
+    def __init__(self) -> None:
+        super().__init__(
+            name="unsecure_description_analyzer", heuristic=Heuristics.UNSECURE_DESCRIPTION, depends_on=None
+        )
+
+    def analyze(self, pypi_package_json: PyPIPackageJsonAsset) -> tuple[HeuristicResult, dict[str, JsonType]]:
+        """Analyze the package.
+
+        Parameters
+        ----------
+        pypi_package_json: PyPIPackageJsonAsset
+            The PyPI package JSON asset object.
+
+        Returns
+        -------
+        tuple[HeuristicResult, dict[str, JsonType]]:
+            The result and related information collected during the analysis.
+        """
+        package_json = pypi_package_json.package_json
+        info = package_json.get("info", {})
+        if not info:
+            error_msg = "No package info found in metadata"
+            logger.debug(error_msg)
+            raise HeuristicAnalyzerValueError(error_msg)
+
+        description = json_extract(package_json, ["info", "description"], str)
+        summary = json_extract(package_json, ["info", "summary"], str)
+        data = f"{description} {summary}"
+        if self.SECURE_DESCRIPTION_REGEX.search(data):
+            return HeuristicResult.PASS, {"message": "Package description is secure"}
+        return HeuristicResult.FAIL, {"message": "Package description is unsecure"}

src/macaron/slsa_analyzer/checks/detect_malicious_metadata_check.py

@@ -22,11 +22,13 @@
 from macaron.malware_analyzer.pypi_heuristics.metadata.empty_project_link import EmptyProjectLinkAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.metadata.fake_email import FakeEmailAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.metadata.high_release_frequency import HighReleaseFrequencyAnalyzer
+from macaron.malware_analyzer.pypi_heuristics.metadata.minimal_content import MinimalContentAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.metadata.one_release import OneReleaseAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.metadata.similar_projects import SimilarProjectAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.metadata.source_code_repo import SourceCodeRepoAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.metadata.typosquatting_presence import TyposquattingPresenceAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.metadata.unchanged_release import UnchangedReleaseAnalyzer
+from macaron.malware_analyzer.pypi_heuristics.metadata.unsecure_description import UnsecureDescriptionAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.metadata.wheel_absence import WheelAbsenceAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.sourcecode.pypi_sourcecode_analyzer import PyPISourcecodeAnalyzer
 from macaron.malware_analyzer.pypi_heuristics.sourcecode.suspicious_setup import SuspiciousSetupAnalyzer
@@ -366,6 +368,8 @@ def run_check(self, ctx: AnalyzeContext) -> CheckResultData:
         TyposquattingPresenceAnalyzer,
         FakeEmailAnalyzer,
         SimilarProjectAnalyzer,
+        UnsecureDescriptionAnalyzer,
+        MinimalContentAnalyzer,
     ]
 
     # name used to query the result of all problog rules, so it can be accessed outside the model.
@@ -419,6 +423,13 @@ def run_check(self, ctx: AnalyzeContext) -> CheckResultData:
     {Confidence.HIGH.value}::trigger(malware_high_confidence_4) :-
         quickUndetailed, forceSetup, failed({Heuristics.TYPOSQUATTING_PRESENCE.value}).
 
+    % Package released with dependency confusion .
+    {Confidence.HIGH.value}::trigger(malware_high_confidence_5) :-
+        forceSetup,
+        passed({Heuristics.MINIMAL_CONTENT.value}),
+        failed({Heuristics.ANOMALOUS_VERSION.value}),
+        failed({Heuristics.UNSECURE_DESCRIPTION.value}).
+
     % Package released recently with little detail, with multiple releases as a trust marker, but frequent and with
     % the same code.
     {Confidence.MEDIUM.value}::trigger(malware_medium_confidence_1) :-
@@ -431,7 +442,8 @@ def run_check(self, ctx: AnalyzeContext) -> CheckResultData:
     {Confidence.MEDIUM.value}::trigger(malware_medium_confidence_2) :-
         quickUndetailed,
         failed({Heuristics.ONE_RELEASE.value}),
-        failed({Heuristics.ANOMALOUS_VERSION.value}).
+        failed({Heuristics.ANOMALOUS_VERSION.value}),
+        failed({Heuristics.UNSECURE_DESCRIPTION.value}).
 
     % Package has no links, one release or multiple quick releases, and a suspicious maintainer who recently
     % joined, has a fake email address, and other similarly-structured projects.
@@ -445,13 +457,15 @@ def run_check(self, ctx: AnalyzeContext) -> CheckResultData:
         failed({Heuristics.SIMILAR_PROJECTS.value}),
         failed({Heuristics.HIGH_RELEASE_FREQUENCY.value}),
         failed({Heuristics.FAKE_EMAIL.value}).
+
     % ----- Evaluation -----
 
     % Aggregate result
     {problog_result_access} :- trigger(malware_high_confidence_1).
     {problog_result_access} :- trigger(malware_high_confidence_2).
     {problog_result_access} :- trigger(malware_high_confidence_3).
     {problog_result_access} :- trigger(malware_high_confidence_4).
+    {problog_result_access} :- trigger(malware_high_confidence_5).
     {problog_result_access} :- trigger(malware_medium_confidence_1).
     {problog_result_access} :- trigger(malware_medium_confidence_2).
     {problog_result_access} :- trigger(malware_medium_confidence_3).

tests/malware_analyzer/pypi/test_minimal_content.py

@@ -0,0 +1,107 @@
+# Copyright (c) 2024 - 2025, Oracle and/or its affiliates. All rights reserved.
+# Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/.
+
+"""Tests for the MinimalContentAnalyzer heuristic."""
+
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+from macaron.errors import SourceCodeError
+from macaron.malware_analyzer.pypi_heuristics.heuristics import HeuristicResult
+from macaron.malware_analyzer.pypi_heuristics.metadata.minimal_content import MinimalContentAnalyzer
+
+
+@pytest.fixture(name="analyzer")
+def analyzer_() -> MinimalContentAnalyzer:
+    """Pytest fixture to create a MinimalContentAnalyzer instance."""
+    return MinimalContentAnalyzer()
+
+
+def test_analyze_sufficient_files_pass(analyzer: MinimalContentAnalyzer, pypi_package_json: MagicMock) -> None:
+    """Test the analyzer passes when the package has sufficient files."""
+    pypi_package_json.download_sourcecode.return_value = True
+    pypi_package_json.package_sourcecode_path = "/fake/path"
+    with patch("os.walk") as mock_walk:
+        mock_walk.return_value = [("root", [], [f"file{i}.py" for i in range(60)])]
+        result, info = analyzer.analyze(pypi_package_json)
+
+    assert result == HeuristicResult.PASS
+    assert info == {"message": "Package has sufficient content"}
+    pypi_package_json.download_sourcecode.assert_called_once()
+
+
+def test_analyze_exactly_threshold_files_pass(analyzer: MinimalContentAnalyzer, pypi_package_json: MagicMock) -> None:
+    """Test the analyzer passes when the package has exactly the threshold number of files."""
+    pypi_package_json.download_sourcecode.return_value = True
+    pypi_package_json.package_sourcecode_path = "/fake/path"
+    with patch("os.walk") as mock_walk:
+        mock_walk.return_value = [("root", [], [f"file{i}.py" for i in range(50)])]
+        result, info = analyzer.analyze(pypi_package_json)
+
+    assert result == HeuristicResult.PASS
+    assert info == {"message": "Package has sufficient content"}
+
+
+def test_analyze_insufficient_files_fail(analyzer: MinimalContentAnalyzer, pypi_package_json: MagicMock) -> None:
+    """Test the analyzer fails when the package has insufficient files."""
+    pypi_package_json.download_sourcecode.return_value = True
+    pypi_package_json.package_sourcecode_path = "/fake/path"
+    with patch("os.walk") as mock_walk:
+        mock_walk.return_value = [("root", [], ["file1.py"])]
+        result, info = analyzer.analyze(pypi_package_json)
+
+    assert result == HeuristicResult.FAIL
+    assert info == {"message": "Not enough files found"}
+
+
+def test_analyze_no_files_fail(analyzer: MinimalContentAnalyzer, pypi_package_json: MagicMock) -> None:
+    """Test the analyzer fails when the package has no files."""
+    pypi_package_json.download_sourcecode.return_value = True
+    pypi_package_json.package_sourcecode_path = "/fake/path"
+    with patch("os.walk") as mock_walk:
+        mock_walk.return_value = [("root", [], [])]
+        result, info = analyzer.analyze(pypi_package_json)
+
+    assert result == HeuristicResult.FAIL
+    assert info == {"message": "Not enough files found"}
+
+
+def test_analyze_download_failed_raises_error(analyzer: MinimalContentAnalyzer, pypi_package_json: MagicMock) -> None:
+    """Test the analyzer raises SourceCodeError when source code download fails."""
+    pypi_package_json.download_sourcecode.return_value = False
+
+    with pytest.raises(SourceCodeError) as exc_info:
+        analyzer.analyze(pypi_package_json)
+
+    assert "No source code files have been downloaded" in str(exc_info.value)
+    pypi_package_json.download_sourcecode.assert_called_once()
+
+
+@pytest.mark.parametrize(
+    ("file_count", "expected_result"),
+    [
+        (0, HeuristicResult.FAIL),
+        (1, HeuristicResult.FAIL),
+        (2, HeuristicResult.FAIL),
+        (55, HeuristicResult.PASS),
+        (70, HeuristicResult.PASS),
+    ],
+)
+def test_analyze_various_file_counts(
+    analyzer: MinimalContentAnalyzer,
+    pypi_package_json: MagicMock,
+    file_count: int,
+    expected_result: HeuristicResult,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Test the analyzer with various file counts."""
+    pypi_package_json.download_sourcecode.return_value = True
+    pypi_package_json.package_sourcecode_path = "/fake/path"
+    files = [f"file{i}.py" for i in range(file_count)]
+    mock_walk = MagicMock(return_value=[("root", [], files)])
+    monkeypatch.setattr("os.walk", mock_walk)
+
+    result, _ = analyzer.analyze(pypi_package_json)
+
+    assert result == expected_result

tests/malware_analyzer/pypi/test_unsecure_description.py

@@ -0,0 +1,67 @@
+# Copyright (c) 2024 - 2025, Oracle and/or its affiliates. All rights reserved.
+# Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/.
+
+"""Tests for the UnsecureDescriptionAnalyzer heuristic."""
+
+from unittest.mock import MagicMock
+
+import pytest
+
+from macaron.errors import HeuristicAnalyzerValueError
+from macaron.malware_analyzer.pypi_heuristics.heuristics import HeuristicResult
+from macaron.malware_analyzer.pypi_heuristics.metadata.unsecure_description import UnsecureDescriptionAnalyzer
+
+
+@pytest.fixture(name="analyzer")
+def analyzer_() -> UnsecureDescriptionAnalyzer:
+    """Pytest fixture to create an UnsecureDescriptionAnalyzer instance."""
+    return UnsecureDescriptionAnalyzer()
+
+
+def test_analyze_secure_description_pass(analyzer: UnsecureDescriptionAnalyzer, pypi_package_json: MagicMock) -> None:
+    """Test the analyzer passes when the package description is secure."""
+    pypi_package_json.package_json = {"info": {"description": "This is an internal package."}}
+    result, info = analyzer.analyze(pypi_package_json)
+    assert result == HeuristicResult.PASS
+    assert info["message"] == "Package description is secure"
+
+
+def test_analyze_unsecure_description_fail(analyzer: UnsecureDescriptionAnalyzer, pypi_package_json: MagicMock) -> None:
+    """Test the analyzer fails when the package description is unsecure."""
+    pypi_package_json.package_json = {"info": {"description": "A public utility library."}}
+    result, info = analyzer.analyze(pypi_package_json)
+    assert result == HeuristicResult.FAIL
+    assert info["message"] == "Package description is unsecure"
+
+
+def test_analyze_no_info_skip(analyzer: UnsecureDescriptionAnalyzer, pypi_package_json: MagicMock) -> None:
+    """Test the analyzer raises an error when no package info is found."""
+    pypi_package_json.package_json = {}
+    with pytest.raises(HeuristicAnalyzerValueError) as exc_info:
+        analyzer.analyze(pypi_package_json)
+    assert "No package info found in metadata" in str(exc_info.value)
+
+
+@pytest.mark.parametrize(
+    ("metadata", "expected_result"),
+    [
+        ({"description": "For internal use only"}, HeuristicResult.PASS),
+        ({"summary": "This is a private package"}, HeuristicResult.PASS),
+        ({"description": "A placeholder for a future project"}, HeuristicResult.PASS),
+        ({"summary": "Used for dependency confusion testing"}, HeuristicResult.PASS),
+        ({"description": "A package for security research"}, HeuristicResult.PASS),
+        ({"summary": "This name is reserved for namespace protection"}, HeuristicResult.PASS),
+        ({"description": "This is a stub package"}, HeuristicResult.PASS),
+        ({"description": "A regular package", "summary": "Does regular things"}, HeuristicResult.FAIL),
+    ],
+)
+def test_analyze_scenarios(
+    analyzer: UnsecureDescriptionAnalyzer,
+    pypi_package_json: MagicMock,
+    metadata: dict,
+    expected_result: HeuristicResult,
+) -> None:
+    """Test the analyzer with various metadata scenarios."""
+    pypi_package_json.package_json = {"info": metadata}
+    result, _ = analyzer.analyze(pypi_package_json)
+    assert result == expected_result