Merge branch 'dev' into dependency-analysis

2 files changed, 43 insertions, 37 deletions

diff --git a/sec_certs/model/cpe_matching.py b/sec_certs/model/cpe_matching.py
index 610b409a..db63f712 100644
--- a/sec_certs/model/cpe_matching.py
+++ b/sec_certs/model/cpe_matching.py
@@ -180,6 +180,29 @@ class CPEClassifier(BaseEstimator):
             string = string.lower().replace(CPEClassifier._replace_special_chars_with_space(x.lower()), "").strip()
         return string
 
+    def _process_manufacturer(self, manufacturer: str, result: Set) -> Optional[List[str]]:
+        tokenized = manufacturer.split()
+        if tokenized[0] in self.vendors_:
+            result.add(tokenized[0])
+        if len(tokenized) > 1 and tokenized[0] + tokenized[1] in self.vendors_:
+            result.add(tokenized[0] + tokenized[1])
+
+        # Below are completely manual fixes
+        if "hewlett" in tokenized or "hewlett-packard" in tokenized or manufacturer == "hewlett packard":
+            result.add("hp")
+        if "thales" in tokenized:
+            result.add("thalesesecurity")
+            result.add("thalesgroup")
+        if "stmicroelectronics" in tokenized:
+            result.add("st")
+        if "athena" in tokenized and "smartcard" in tokenized:
+            result.add("athena-scs")
+        if tokenized[0] == "the" and not result:
+            candidate_result = self.get_candidate_list_of_vendors(" ".join(tokenized[1:]))
+            return list(candidate_result) if candidate_result else None
+
+        return list(result) if result else None
+
     def get_candidate_list_of_vendors(self, manufacturer: str) -> Optional[List[str]]:
         """
         Given manufacturer name, this method will find list of plausible vendors from CPE dataset that are likely related.
@@ -203,27 +226,7 @@ class CPEClassifier(BaseEstimator):
         if manufacturer in self.vendors_:
             result.add(manufacturer)
 
-        tokenized = manufacturer.split()
-        if tokenized[0] in self.vendors_:
-            result.add(tokenized[0])
-        if len(tokenized) > 1 and tokenized[0] + tokenized[1] in self.vendors_:
-            result.add(tokenized[0] + tokenized[1])
-
-        # Below are completely manual fixes
-        if "hewlett" in tokenized or "hewlett-packard" in tokenized or manufacturer == "hewlett packard":
-            result.add("hp")
-        if "thales" in tokenized:
-            result.add("thalesesecurity")
-            result.add("thalesgroup")
-        if "stmicroelectronics" in tokenized:
-            result.add("st")
-        if "athena" in tokenized and "smartcard" in tokenized:
-            result.add("athena-scs")
-        if tokenized[0] == "the" and not result:
-            candidate_result = self.get_candidate_list_of_vendors(" ".join(tokenized[1:]))
-            return list(candidate_result) if candidate_result else None
-
-        return list(result) if result else None
+        return self._process_manufacturer(manufacturer, result)
 
     def get_candidate_vendor_version_pairs(
         self, cert_candidate_cpe_vendors: List[str], cert_candidate_versions: List[str]
diff --git a/sec_certs/model/dependency_finder.py b/sec_certs/model/dependency_finder.py
index 8b769d68..a86bc198 100644
--- a/sec_certs/model/dependency_finder.py
+++ b/sec_certs/model/dependency_finder.py
@@ -20,6 +20,24 @@ class DependencyFinder:
             referenced_by[cert_id].append(this_cert_id)
 
     @staticmethod
+    def _process_references(referenced_by: ReferencedByDirect, referenced_by_indirect: ReferencedByIndirect):
+        new_change_detected = True
+        while new_change_detected:
+            new_change_detected = False
+            certs_id_list = referenced_by.keys()
+
+            for cert_id in certs_id_list:
+                tmp_referenced_by_indirect_nums = referenced_by_indirect[cert_id].copy()
+                for referencing in tmp_referenced_by_indirect_nums:
+                    if referencing in referenced_by.keys():
+                        tmp_referencing = referenced_by_indirect[referencing].copy()
+                        newly_discovered_references = [
+                            x for x in tmp_referencing if x not in referenced_by_indirect[cert_id]
+                        ]
+                        referenced_by_indirect[cert_id].update(newly_discovered_references)
+                        new_change_detected = True if newly_discovered_references else False
+
+    @staticmethod
     def _build_cert_references(certificates: Certificates) -> Tuple[ReferencedByDirect, ReferencedByIndirect]:
         referenced_by: ReferencedByDirect = {}
 
@@ -43,22 +61,7 @@ class DependencyFinder:
             for item in referenced_by[cert_id]:
                 referenced_by_indirect[cert_id].add(item)
 
-        new_change_detected = True
-        while new_change_detected:
-            new_change_detected = False
-            certs_id_list = referenced_by.keys()
-
-            for cert_id in certs_id_list:
-                tmp_referenced_by_indirect_nums = referenced_by_indirect[cert_id].copy()
-                for referencing in tmp_referenced_by_indirect_nums:
-                    if referencing in referenced_by.keys():
-                        tmp_referencing = referenced_by_indirect[referencing].copy()
-                        newly_discovered_references = [
-                            x for x in tmp_referencing if x not in referenced_by_indirect[cert_id]
-                        ]
-                        referenced_by_indirect[cert_id].update(newly_discovered_references)
-                        new_change_detected = True if newly_discovered_references else False
-
+        DependencyFinder._process_references(referenced_by, referenced_by_indirect)
         return referenced_by, referenced_by_indirect
 
     @staticmethod