feat: improve sql schema

cvelib/cve.py

@@ -298,6 +298,13 @@ def setPriority(self, s: str) -> None:
 
     def setDiscoveredBy(self, s: str) -> None:
         """Set Discovered-by"""
+        seen: set = set()
+        for d in s.split(","):
+            d = d.strip()
+            if d != "":
+                if d in seen:
+                    raise CveException("duplicate discoverer '%s'" % d)
+                seen.add(d)
         self.discoveredBy = s
         self.data["Discovered-by"] = self.discoveredBy
 

cvelib/github.py

@@ -340,21 +340,28 @@ def parse(s: str) -> List[Union[GHDependabot, GHSecret, GHCode]]:
         raise CveException("invalid yaml:\n'%s'" % s)
 
     ghas: List[Union[GHDependabot, GHSecret, GHCode]] = []
+    seen_urls: set = set()
     for item in yml:
         if "type" not in item:
             raise CveException("invalid GHAS document: 'type' missing for item")
 
+        obj: Union[GHDependabot, GHSecret, GHCode]
         if item["type"] == "dependabot":
-            ghas.append(GHDependabot(item))
+            obj = GHDependabot(item)
         elif item["type"] == "secret-scanning":
-            ghas.append(GHSecret(item))
+            obj = GHSecret(item)
         elif item["type"] == "code-scanning":
-            ghas.append(GHCode(item))
+            obj = GHCode(item)
         else:
             raise CveException(
                 "invalid GHAS document: unknown GHAS type '%s'" % item["type"]
             )
 
+        if obj.url != "unavailable" and obj.url in seen_urls:
+            raise CveException("duplicate GHAS url '%s'" % obj.url)
+        seen_urls.add(obj.url)
+        ghas.append(obj)
+
     return ghas
 
 

cvelib/pkg.py

@@ -166,6 +166,8 @@ def setTags(self, tagList: List[Tuple[str, str]]) -> None:
                 t = t.strip()
                 if not rePatterns["pkg-tags"].search(t):
                     raise CveException("invalid tag '%s'" % t)
+                if t in self.tags[tagKey]:
+                    raise CveException("duplicate tag '%s'" % t)
                 self.tags[tagKey].append(t)
 
     def setPriorities(self, priorityList: List[Tuple[str, str]]) -> None:

cvelib/scan.py

@@ -267,17 +267,27 @@ def parse(s: str) -> List[ScanOCI]:
         raise CveException("invalid yaml:\n'%s'" % s)
 
     mans: List[ScanOCI] = []
+    seen_keys: set = set()
     for item in yml:
         if "type" not in item:
             raise CveException("invalid Scan-Reports document: 'type' missing for item")
 
         if item["type"] == "oci":
-            mans.append(ScanOCI(item))
+            obj = ScanOCI(item)
         else:
             raise CveException(
                 "invalid Scan-Reports document: unknown type '%s'" % item["type"]
             )
 
+        key = (obj.component, obj.detectedIn, obj.advisory, obj.url)
+        if key in seen_keys:
+            raise CveException(
+                "duplicate scan entry '%s' '%s' '%s' '%s'"
+                % (obj.component, obj.detectedIn, obj.advisory, obj.url)
+            )
+        seen_keys.add(key)
+        mans.append(obj)
+
     return mans