new: [Cluster] load cluster from existing cluster

pymispgalaxies/__init__.py

@@ -1,4 +1,4 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 
-from .api import Galaxies, Clusters, EncodeGalaxies, EncodeClusters, UnableToRevertMachinetag
+from .api import Galaxies, Galaxy, Clusters, Cluster, EncodeGalaxies, EncodeClusters, UnableToRevertMachinetag

pymispgalaxies/api.py

@@ -357,7 +357,7 @@ class Cluster(Mapping):  # type: ignore
         duplicates (List[Tuple[str, str]]): A list of tuples representing duplicate values in the cluster, where each tuple contains the name of the cluster and the duplicate value.
 
     Methods:
-        __init__(self, cluster: Dict[str, Any], skip_duplicates: bool = False): Initializes a Cluster object.
+        __init__(self, cluster: Dict[str, Any] | str, skip_duplicates: bool = False): Initializes a Cluster object from a dict or existing cluster file
         search(self, query: str, return_tags: bool = False) -> Union[List[ClusterValue], List[str]]: Searches for values in the cluster that match the given query.
         machinetags(self) -> List[str]: Returns a list of machine tags for the cluster.
         get_by_external_id(self, external_id: str) -> ClusterValue: Returns the cluster value with the specified external ID.
@@ -368,14 +368,21 @@ class Cluster(Mapping):  # type: ignore
         to_json(self) -> str: Converts the Cluster object to a JSON string.
         to_dict(self) -> Dict[str, Any]: Converts the Cluster object to a dictionary.
     """
-    def __init__(self, cluster: Dict[str, Any], skip_duplicates: bool = False):
+    def __init__(self, cluster: Dict[str, Any] | str, skip_duplicates: bool = False):
         """
         Initializes a Cluster object.
 
         Args:
-            cluster (Dict[str, Any]): A dictionary containing the cluster data.
+            cluster (Dict[str, Any] | str): A dictionary containing the cluster data, or the name of the existing cluster to load from the data folder.
             skip_duplicates (bool, optional): Flag indicating whether to skip duplicate values. Defaults to False.
         """
+        if isinstance(cluster, str):
+            root_dir_clusters = os.path.join(os.path.abspath(os.path.dirname(sys.modules['pymispgalaxies'].__file__)), 'data', 'misp-galaxy', 'clusters')
+            cluster_file = os.path.join(root_dir_clusters, f"{cluster}.json")
+            with open(cluster_file, 'r') as f:
+                self.__init__(json.load(f), skip_duplicates=skip_duplicates)
+
+        else:
             self.cluster = cluster
             self.name = self.cluster['name']
             self.type = self.cluster['type']

tests/tests.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 import unittest
-from pymispgalaxies import Galaxies, Clusters, UnableToRevertMachinetag
+from pymispgalaxies import Galaxies, Clusters, Cluster, UnableToRevertMachinetag
 from glob import glob
 import os
 import json
@@ -129,7 +129,7 @@ class TestPyMISPGalaxies(unittest.TestCase):
         self.assertFalse(errors)
 
     def test_get_by_external_id(self):
-        cluster = self.clusters.get('mitre-attack-pattern')
+        cluster = Cluster(cluster='mitre-attack-pattern')
         self.assertIsNotNone(cluster)
         cluster_by_external_id = cluster.get_by_external_id('T1525')
         cluster_by_value = cluster.get('Implant Internal Image - T1525')