refactor: Refactoring from notes of code review

3 files changed, 14 insertions, 6 deletions

diff --git a/src/sec_certs/dataset/cve.py b/src/sec_certs/dataset/cve.py
index 6960c6fc..566bc2e4 100644
--- a/src/sec_certs/dataset/cve.py
+++ b/src/sec_certs/dataset/cve.py
@@ -35,6 +35,7 @@ class CVEDataset(JSONPathDataset, ComplexSerializableType):
         self.cves = cves
         self.json_path = Path(json_path)
         self.cpe_to_cve_ids_lookup: dict[str, set[str]] = dict()
+        self.cves_with_vulnerable_configurations: list[CVE] = []
 
     @property
     def serialized_attributes(self) -> list[str]:
diff --git a/src/sec_certs/sample/cpe.py b/src/sec_certs/sample/cpe.py
index 29e078b6..0c8cd149 100644
--- a/src/sec_certs/sample/cpe.py
+++ b/src/sec_certs/sample/cpe.py
@@ -19,6 +19,12 @@ class CPEConfiguration(ComplexSerializableType):
         self.platform: str = platform
         self.cpes: set[str] = cpes
 
+    def __hash__(self) -> int:
+        return hash(self.platform) + sum([hash(cpe) for cpe in self.cpes])
+
+    def __lt__(self, other: CPEConfiguration) -> bool:
+        return self.platform < other.platform
+
     def __eq__(self, other: Any) -> bool:
         return isinstance(other, self.__class__) and self.platform == other.platform and self.cpes == other.cpes
 
diff --git a/src/sec_certs/sample/cve.py b/src/sec_certs/sample/cve.py
index d3ea1590..caf2b4d7 100644
--- a/src/sec_certs/sample/cve.py
+++ b/src/sec_certs/sample/cve.py
@@ -47,8 +47,8 @@ class CVE(PandasSerializableType, ComplexSerializableType):
             raise ValueError("NIST Dict for CVE Impact badly formatted.")
 
     cve_id: str
-    vulnerable_cpes: list[CPE]
-    vulnerable_cpe_configurations: list[CPEConfiguration]
+    vulnerable_cpes: set[CPE]
+    vulnerable_cpe_configurations: set[CPEConfiguration]
     impact: Impact
     published_date: datetime.datetime | None
     cwe_ids: set[str] | None
@@ -69,8 +69,8 @@ class CVE(PandasSerializableType, ComplexSerializableType):
     def __init__(
         self,
         cve_id: str,
-        vulnerable_cpes: list[CPE],
-        vulnerable_cpe_configurations: list[CPEConfiguration],
+        vulnerable_cpes: set[CPE],
+        vulnerable_cpe_configurations: set[CPEConfiguration],
         impact: Impact,
         published_date: str,
         cwe_ids: set[str] | None,
@@ -220,8 +220,9 @@ class CVE(PandasSerializableType, ComplexSerializableType):
         cve_id = dct["cve"]["CVE_data_meta"]["ID"]
         impact = cls.Impact.from_nist_dict(dct)
         cpe_and_cpe_configurations = get_vulnerable_cpes_from_nist_dict(dct)
-        vulnerable_cpes = list(itertools.chain.from_iterable(cpe_and_cpe_configurations[0]))
-        vulnerable_cpe_configurations = list(itertools.chain.from_iterable(cpe_and_cpe_configurations[1]))
+        # There exist CVEs such as (CVE-2022-0177) which are rejected and do not contain any assinged CPEs
+        vulnerable_cpes = set(itertools.chain.from_iterable(cpe_and_cpe_configurations[0])) if cpe_and_cpe_configurations else set()
+        vulnerable_cpe_configurations = set(itertools.chain.from_iterable(cpe_and_cpe_configurations[1])) if cpe_and_cpe_configurations else set()
         published_date = dct["publishedDate"]
         cwe_ids = cls.parse_cwe_data(dct)