Fix mutable default state.

7 files changed, 57 insertions, 47 deletions

diff --git a/src/sec_certs/dataset/cc.py b/src/sec_certs/dataset/cc.py
index fb90c142..b86a03f3 100644
--- a/src/sec_certs/dataset/cc.py
+++ b/src/sec_certs/dataset/cc.py
@@ -91,33 +91,34 @@ class CCDataset(Dataset[CCCertificate], ComplexSerializableType):
 
     def __init__(
         self,
-        certs: dict[str, CCCertificate] = {},
+        certs: dict[str, CCCertificate] | None = None,
         root_dir: str | Path = constants.DUMMY_NONEXISTING_PATH,
         name: str | None = None,
         description: str = "",
         state: Dataset.DatasetInternalState | None = None,
-        aux_handlers: dict[type[AuxiliaryDatasetHandler], AuxiliaryDatasetHandler] = {},
+        aux_handlers: dict[type[AuxiliaryDatasetHandler], AuxiliaryDatasetHandler] | None = None,
     ):
-        self.certs = certs
+        self.certs = certs if certs is not None else {}
         self.timestamp = datetime.now()
         self.sha256_digest = "not implemented"
         self.name = name if name else type(self).__name__ + " dataset"
         self.description = description if description else datetime.now().strftime("%d/%m/%Y %H:%M:%S")
         self.state = state if state else self.DatasetInternalState()
-        self.aux_handlers = aux_handlers
         self.root_dir = Path(root_dir)
-
-        if not self.aux_handlers:
-            self.aux_handlers[CPEDatasetHandler] = CPEDatasetHandler(self.auxiliary_datasets_dir)
-            self.aux_handlers[CVEDatasetHandler] = CVEDatasetHandler(self.auxiliary_datasets_dir)
-            self.aux_handlers[CPEMatchDictHandler] = CPEMatchDictHandler(self.auxiliary_datasets_dir)
-            self.aux_handlers[CCSchemeDatasetHandler] = CCSchemeDatasetHandler(self.auxiliary_datasets_dir)
-            self.aux_handlers[ProtectionProfileDatasetHandler] = ProtectionProfileDatasetHandler(
-                self.auxiliary_datasets_dir
-            )
-            self.aux_handlers[CCMaintenanceUpdateDatasetHandler] = CCMaintenanceUpdateDatasetHandler(
-                self.auxiliary_datasets_dir
-            )
+        self.aux_handlers = (
+            aux_handlers
+            if aux_handlers is not None
+            else {
+                CPEDatasetHandler: CPEDatasetHandler(self.auxiliary_datasets_dir),
+                CVEDatasetHandler: CVEDatasetHandler(self.auxiliary_datasets_dir),
+                CPEMatchDictHandler: CPEMatchDictHandler(self.auxiliary_datasets_dir),
+                CCSchemeDatasetHandler: CCSchemeDatasetHandler(self.auxiliary_datasets_dir),
+                ProtectionProfileDatasetHandler: ProtectionProfileDatasetHandler(self.auxiliary_datasets_dir),
+                CCMaintenanceUpdateDatasetHandler: CCMaintenanceUpdateDatasetHandler(self.auxiliary_datasets_dir),
+            }
+        )
+        # Make sure that the auxiliary handlers (if supplied by the user) have the correct root_dir
+        self._set_local_paths()
 
     def to_pandas(self) -> pd.DataFrame:
         """
@@ -820,14 +821,13 @@ class CCDatasetMaintenanceUpdates(CCDataset, ComplexSerializableType):
     # Quite difficult to achieve correct behaviour with MyPy here, opting for ignore
     def __init__(
         self,
-        certs: dict[str, CCMaintenanceUpdate] = {},  # type: ignore
+        certs: dict[str, CCMaintenanceUpdate] | None = None,  # type: ignore
         root_dir: Path = constants.DUMMY_NONEXISTING_PATH,
         name: str = "dataset name",
         description: str = "dataset_description",
         state: CCDataset.DatasetInternalState | None = None,
     ):
-        super().__init__(certs, root_dir, name, description, state)  # type: ignore
-        self.aux_handlers = {}
+        super().__init__(certs, root_dir, name, description, state, aux_handlers={})  # type: ignore
         self.state.meta_sources_parsed = True
 
     @property
diff --git a/src/sec_certs/dataset/cpe.py b/src/sec_certs/dataset/cpe.py
index 3257d2a4..43db8a79 100644
--- a/src/sec_certs/dataset/cpe.py
+++ b/src/sec_certs/dataset/cpe.py
@@ -31,11 +31,11 @@ class CPEDataset(JSONPathDataset, ComplexSerializableType):
 
     def __init__(
         self,
-        cpes: dict[str, CPE] = {},
+        cpes: dict[str, CPE] | None = None,
         json_path: str | Path = constants.DUMMY_NONEXISTING_PATH,
         last_update_timestamp: datetime = datetime.fromtimestamp(0),
     ):
-        self.cpes = cpes
+        self.cpes = cpes if cpes is not None else {}
         self.json_path = Path(json_path)
         self.last_update_timestamp = last_update_timestamp
 
diff --git a/src/sec_certs/dataset/cve.py b/src/sec_certs/dataset/cve.py
index f76211ee..1dbf8b37 100644
--- a/src/sec_certs/dataset/cve.py
+++ b/src/sec_certs/dataset/cve.py
@@ -29,11 +29,11 @@ class CVEDataset(JSONPathDataset, ComplexSerializableType):
 
     def __init__(
         self,
-        cves: dict[str, CVE] = {},
+        cves: dict[str, CVE] | None = None,
         json_path: str | Path = constants.DUMMY_NONEXISTING_PATH,
         last_update_timestamp: datetime = datetime.fromtimestamp(0),
     ):
-        self.cves = cves
+        self.cves = cves if cves is not None else {}
         self.json_path = Path(json_path)
         self._cpe_uri_to_cve_ids_lookup: dict[str, set[str]] = {}
         self._cves_with_vulnerable_configurations: list[CVE] = []
@@ -120,7 +120,7 @@ class CVEDataset(JSONPathDataset, ComplexSerializableType):
     def build_lookup_dict(
         self,
         cpe_match_feed: dict,
-        limit_to_cpes: set[CPE] = set(),
+        limit_to_cpes: set[CPE] | None = None,
     ):
         self._cpe_uri_to_cve_ids_lookup = {}
         cpe_uris_of_interest = {x.uri for x in limit_to_cpes} if limit_to_cpes else None
diff --git a/src/sec_certs/dataset/dataset.py b/src/sec_certs/dataset/dataset.py
index c17109f3..af7a56ae 100644
--- a/src/sec_certs/dataset/dataset.py
+++ b/src/sec_certs/dataset/dataset.py
@@ -51,22 +51,25 @@ class Dataset(Generic[CertSubType], ComplexSerializableType, ABC):
 
     def __init__(
         self,
-        certs: dict[str, CertSubType] = {},
+        certs: dict[str, CertSubType] | None = None,
         root_dir: str | Path = constants.DUMMY_NONEXISTING_PATH,
         name: str | None = None,
         description: str = "",
         state: DatasetInternalState | None = None,
-        aux_handlers: dict[type[AuxiliaryDatasetHandler], AuxiliaryDatasetHandler] = {},
+        aux_handlers: dict[type[AuxiliaryDatasetHandler], AuxiliaryDatasetHandler] | None = None,
     ):
-        self.certs = certs
+        # TODO: This is actually unused, none of the datasets call super().__init__.
+        self.certs = certs if certs is not None else {}
 
         self.timestamp = datetime.now()
         self.sha256_digest = "not implemented"
         self.name = name if name else type(self).__name__.lower() + "_dataset"
         self.description = description if description else "No description provided"
         self.state = state if state else self.DatasetInternalState()
-        self.aux_handlers = aux_handlers
         self.root_dir = Path(root_dir)
+        self.aux_handlers = aux_handlers if aux_handlers is not None else {}
+        # Make sure that the auxiliary handlers (if supplied by the user) have the correct root_dir
+        self._set_local_paths()
 
     @property
     def root_dir(self) -> Path:
@@ -250,8 +253,9 @@ class Dataset(Generic[CertSubType], ComplexSerializableType, ABC):
         return dset
 
     def _set_local_paths(self) -> None:
-        for handler in self.aux_handlers.values():
-            handler.set_local_paths(self.auxiliary_datasets_dir)
+        if hasattr(self, "aux_handlers"):
+            for handler in self.aux_handlers.values():
+                handler.set_local_paths(self.auxiliary_datasets_dir)
 
     def move_dataset(self, new_root_dir: str | Path) -> None:
         """
diff --git a/src/sec_certs/dataset/fips.py b/src/sec_certs/dataset/fips.py
index 34b1147d..90d6b5df 100644
--- a/src/sec_certs/dataset/fips.py
+++ b/src/sec_certs/dataset/fips.py
@@ -59,27 +59,32 @@ class FIPSDataset(Dataset[FIPSCertificate], ComplexSerializableType):
 
     def __init__(
         self,
-        certs: dict[str, FIPSCertificate] = {},
+        certs: dict[str, FIPSCertificate] | None = None,
         root_dir: str | Path = constants.DUMMY_NONEXISTING_PATH,
         name: str | None = None,
         description: str = "",
         state: Dataset.DatasetInternalState | None = None,
-        aux_handlers: dict[type[AuxiliaryDatasetHandler], AuxiliaryDatasetHandler] = {},
+        aux_handlers: dict[type[AuxiliaryDatasetHandler], AuxiliaryDatasetHandler] | None = None,
     ):
-        self.certs = certs
+        self.certs = certs if certs is not None else {}
         self.timestamp = datetime.datetime.now()
         self.sha256_digest = "not implemented"
         self.name = name if name else type(self).__name__ + " dataset"
         self.description = description if description else datetime.datetime.now().strftime("%d/%m/%Y %H:%M:%S")
         self.state = state if state else self.DatasetInternalState()
-        self.aux_handlers = aux_handlers
         self.root_dir = Path(root_dir)
-
-        if not self.aux_handlers:
-            self.aux_handlers[CPEDatasetHandler] = CPEDatasetHandler(self.auxiliary_datasets_dir)
-            self.aux_handlers[CVEDatasetHandler] = CVEDatasetHandler(self.auxiliary_datasets_dir)
-            self.aux_handlers[FIPSAlgorithmDatasetHandler] = FIPSAlgorithmDatasetHandler(self.auxiliary_datasets_dir)
-            self.aux_handlers[CPEMatchDictHandler] = CPEMatchDictHandler(self.auxiliary_datasets_dir)
+        self.aux_handlers = (
+            aux_handlers
+            if aux_handlers is not None
+            else {
+                CPEDatasetHandler: CPEDatasetHandler(self.auxiliary_datasets_dir),
+                CVEDatasetHandler: CVEDatasetHandler(self.auxiliary_datasets_dir),
+                FIPSAlgorithmDatasetHandler: FIPSAlgorithmDatasetHandler(self.auxiliary_datasets_dir),
+                CPEMatchDictHandler: CPEMatchDictHandler(self.auxiliary_datasets_dir),
+            }
+        )
+        # Make sure that the auxiliary handlers (if supplied by the user) have the correct root_dir
+        self._set_local_paths()
 
     LIST_OF_CERTS_HTML: Final[dict[str, str]] = {
         "fips_modules_active.html": constants.FIPS_ACTIVE_MODULES_URL,
@@ -112,7 +117,6 @@ class FIPSDataset(Dataset[FIPSCertificate], ComplexSerializableType):
     def _extract_data_from_html_modules(self) -> None:
         """
         Extracts data from html module file
-        :param bool fresh: if all certs should be processed, or only the failed ones. Defaults to True
         """
         logger.info("Extracting data from html modules.")
         certs_to_process = [x for x in self if x.state.module_is_ok_to_analyze()]
diff --git a/src/sec_certs/dataset/fips_algorithm.py b/src/sec_certs/dataset/fips_algorithm.py
index ee7d06a1..c9a8b3c5 100644
--- a/src/sec_certs/dataset/fips_algorithm.py
+++ b/src/sec_certs/dataset/fips_algorithm.py
@@ -21,8 +21,10 @@ logger = logging.getLogger(__name__)
 
 
 class FIPSAlgorithmDataset(JSONPathDataset, ComplexSerializableType):
-    def __init__(self, algs: dict[str, FIPSAlgorithm] = {}, json_path: str | Path = constants.DUMMY_NONEXISTING_PATH):
-        self.algs = algs
+    def __init__(
+        self, algs: dict[str, FIPSAlgorithm] | None = None, json_path: str | Path = constants.DUMMY_NONEXISTING_PATH
+    ):
+        self.algs = algs if algs is not None else {}
         self.json_path = Path(json_path)
         self.alg_number_to_algs: dict[str, set[FIPSAlgorithm]] = {}
 
diff --git a/src/sec_certs/dataset/protection_profile.py b/src/sec_certs/dataset/protection_profile.py
index 9b22ca75..d5457e7f 100644
--- a/src/sec_certs/dataset/protection_profile.py
+++ b/src/sec_certs/dataset/protection_profile.py
@@ -39,20 +39,20 @@ class ProtectionProfileDataset(Dataset[ProtectionProfile], ComplexSerializableTy
 
     def __init__(
         self,
-        certs: dict[str, ProtectionProfile] = {},
+        certs: dict[str, ProtectionProfile] | None = None,
         root_dir: str | Path = constants.DUMMY_NONEXISTING_PATH,
         name: str | None = None,
         description: str = "",
         state: Dataset.DatasetInternalState | None = None,
-        aux_handlers: dict[type[AuxiliaryDatasetHandler], AuxiliaryDatasetHandler] = {},
+        aux_handlers: dict[type[AuxiliaryDatasetHandler], AuxiliaryDatasetHandler] | None = None,
     ):
-        self.certs = certs
+        self.certs = certs if certs is not None else {}
         self.timestamp = datetime.now()
         self.sha256_digest = "not implemented"
         self.name = name if name else type(self).__name__ + " dataset"
         self.description = description if description else datetime.now().strftime("%d/%m/%Y %H:%M:%S")
         self.state = state if state else self.DatasetInternalState()
-        self.aux_handlers = aux_handlers
+        self.aux_handlers = aux_handlers if aux_handlers is not None else {}
         self.root_dir = Path(root_dir)
 
     @property