Merge pull request #64 from casework/use_urirefs_for_facets

Use URIRefs for Facets
casework · Sep 1, 2022 · d78453d · d78453d
2 parents faeafe2 + 33ebd89
commit d78453d
Show file tree

Hide file tree

Showing 7 changed files with 304 additions and 288 deletions.
diff --git a/case_utils/case_file/__init__.py b/case_utils/case_file/__init__.py
@@ -82,17 +82,19 @@ def create_file_node(
     :returns: The File Observable Object's node.
     :rtype: rdflib.URIRef
     """
+    node_namespace = rdflib.Namespace(node_prefix)
+
     if node_iri is None:
         node_slug = "file-" + case_utils.local_uuid.local_uuid()
-        node_iri = rdflib.Namespace(node_prefix)[node_slug]
+        node_iri = node_namespace[node_slug]
     n_file = rdflib.URIRef(node_iri)
     graph.add((n_file, NS_RDF.type, NS_UCO_OBSERVABLE.File))
 
     basename = os.path.basename(filepath)
     literal_basename = rdflib.Literal(basename)
 
     file_stat = os.stat(filepath)
-    n_file_facet = rdflib.BNode()
+    n_file_facet = node_namespace["file-facet-" + case_utils.local_uuid.local_uuid()]
     graph.add(
         (
             n_file_facet,
@@ -119,7 +121,9 @@ def create_file_node(
         graph.add((n_file_facet, NS_UCO_OBSERVABLE.modifiedTime, literal_mtime))
 
     if not disable_hashes:
-        n_contentdata_facet = rdflib.BNode()
+        n_contentdata_facet = node_namespace[
+            "content-data-facet-" + case_utils.local_uuid.local_uuid()
+        ]
         graph.add((n_file, NS_UCO_CORE.hasFacet, n_contentdata_facet))
         graph.add(
             (n_contentdata_facet, NS_RDF.type, NS_UCO_OBSERVABLE.ContentDataFacet)
@@ -191,7 +195,7 @@ def create_file_node(
         for key in successful_hashdict._fields:
             if key not in ("md5", "sha1", "sha256", "sha512"):
                 continue
-            n_hash = rdflib.BNode()
+            n_hash = node_namespace["hash-" + case_utils.local_uuid.local_uuid()]
             graph.add((n_contentdata_facet, NS_UCO_OBSERVABLE.hash, n_hash))
             graph.add((n_hash, NS_RDF.type, NS_UCO_TYPES.Hash))
             graph.add(

diff --git a/tests/case_utils/case_file/kb.json b/tests/case_utils/case_file/kb.json
@@ -8,92 +8,120 @@
         "xsd": "http://www.w3.org/2001/XMLSchema#"
     },
     "@graph": [
+        {
+            "@id": "kb:content-data-facet-1833f979-1f19-5543-9d82-6cefd144b169",
+            "@type": "uco-observable:ContentDataFacet",
+            "uco-observable:hash": [
+                {
+                    "@id": "kb:hash-24644904-83ea-5911-aea8-be687a9f3caf"
+                },
+                {
+                    "@id": "kb:hash-295bdeb5-7f23-5a3f-8b7f-4bb1191b7c21"
+                },
+                {
+                    "@id": "kb:hash-39127f5c-598b-51d4-a720-2e949f18f85f"
+                },
+                {
+                    "@id": "kb:hash-49e81fee-c6b3-5f5f-af8b-0746d32e4932"
+                }
+            ],
+            "uco-observable:sizeInBytes": {
+                "@type": "xsd:integer",
+                "@value": "4"
+            }
+        },
         {
             "@id": "kb:file-800784de-5c9e-5eb2-b843-0ac51a1bd4b9",
             "@type": "uco-observable:File",
             "uco-core:hasFacet": {
-                "@type": "uco-observable:FileFacet",
-                "uco-observable:fileName": "sample.txt",
-                "uco-observable:modifiedTime": {
-                    "@type": "xsd:dateTime",
-                    "@value": "2010-01-02T03:04:56+00:00"
-                },
-                "uco-observable:sizeInBytes": {
-                    "@type": "xsd:integer",
-                    "@value": "4"
-                }
+                "@id": "kb:file-facet-ffa3e6bb-dffc-549d-a7c4-ffc5e90ac55d"
             }
         },
         {
             "@id": "kb:file-ace6460a-92a9-58b9-83ea-a18ae87f6e04",
             "@type": "uco-observable:File",
             "uco-core:hasFacet": [
                 {
-                    "@type": "uco-observable:ContentDataFacet",
-                    "uco-observable:hash": [
-                        {
-                            "@type": "uco-types:Hash",
-                            "uco-types:hashMethod": {
-                                "@type": "uco-vocabulary:HashNameVocab",
-                                "@value": "MD5"
-                            },
-                            "uco-types:hashValue": {
-                                "@type": "xsd:hexBinary",
-                                "@value": "098f6bcd4621d373cade4e832627b4f6"
-                            }
-                        },
-                        {
-                            "@type": "uco-types:Hash",
-                            "uco-types:hashMethod": {
-                                "@type": "uco-vocabulary:HashNameVocab",
-                                "@value": "SHA1"
-                            },
-                            "uco-types:hashValue": {
-                                "@type": "xsd:hexBinary",
-                                "@value": "a94a8fe5ccb19ba61c4c0873d391e987982fbbd3"
-                            }
-                        },
-                        {
-                            "@type": "uco-types:Hash",
-                            "uco-types:hashMethod": {
-                                "@type": "uco-vocabulary:HashNameVocab",
-                                "@value": "SHA256"
-                            },
-                            "uco-types:hashValue": {
-                                "@type": "xsd:hexBinary",
-                                "@value": "9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a08"
-                            }
-                        },
-                        {
-                            "@type": "uco-types:Hash",
-                            "uco-types:hashMethod": {
-                                "@type": "uco-vocabulary:HashNameVocab",
-                                "@value": "SHA512"
-                            },
-                            "uco-types:hashValue": {
-                                "@type": "xsd:hexBinary",
-                                "@value": "ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a8ff"
-                            }
-                        }
-                    ],
-                    "uco-observable:sizeInBytes": {
-                        "@type": "xsd:integer",
-                        "@value": "4"
-                    }
+                    "@id": "kb:content-data-facet-1833f979-1f19-5543-9d82-6cefd144b169"
                 },
                 {
-                    "@type": "uco-observable:FileFacet",
-                    "uco-observable:fileName": "sample.txt",
-                    "uco-observable:modifiedTime": {
-                        "@type": "xsd:dateTime",
-                        "@value": "2010-01-02T03:04:56+00:00"
-                    },
-                    "uco-observable:sizeInBytes": {
-                        "@type": "xsd:integer",
-                        "@value": "4"
-                    }
+                    "@id": "kb:file-facet-1297a4bd-563b-54c3-ad8a-f67f01ba9b10"
                 }
             ]
+        },
+        {
+            "@id": "kb:file-facet-1297a4bd-563b-54c3-ad8a-f67f01ba9b10",
+            "@type": "uco-observable:FileFacet",
+            "uco-observable:fileName": "sample.txt",
+            "uco-observable:modifiedTime": {
+                "@type": "xsd:dateTime",
+                "@value": "2010-01-02T03:04:56+00:00"
+            },
+            "uco-observable:sizeInBytes": {
+                "@type": "xsd:integer",
+                "@value": "4"
+            }
+        },
+        {
+            "@id": "kb:file-facet-ffa3e6bb-dffc-549d-a7c4-ffc5e90ac55d",
+            "@type": "uco-observable:FileFacet",
+            "uco-observable:fileName": "sample.txt",
+            "uco-observable:modifiedTime": {
+                "@type": "xsd:dateTime",
+                "@value": "2010-01-02T03:04:56+00:00"
+            },
+            "uco-observable:sizeInBytes": {
+                "@type": "xsd:integer",
+                "@value": "4"
+            }
+        },
+        {
+            "@id": "kb:hash-24644904-83ea-5911-aea8-be687a9f3caf",
+            "@type": "uco-types:Hash",
+            "uco-types:hashMethod": {
+                "@type": "uco-vocabulary:HashNameVocab",
+                "@value": "MD5"
+            },
+            "uco-types:hashValue": {
+                "@type": "xsd:hexBinary",
+                "@value": "098f6bcd4621d373cade4e832627b4f6"
+            }
+        },
+        {
+            "@id": "kb:hash-295bdeb5-7f23-5a3f-8b7f-4bb1191b7c21",
+            "@type": "uco-types:Hash",
+            "uco-types:hashMethod": {
+                "@type": "uco-vocabulary:HashNameVocab",
+                "@value": "SHA1"
+            },
+            "uco-types:hashValue": {
+                "@type": "xsd:hexBinary",
+                "@value": "a94a8fe5ccb19ba61c4c0873d391e987982fbbd3"
+            }
+        },
+        {
+            "@id": "kb:hash-39127f5c-598b-51d4-a720-2e949f18f85f",
+            "@type": "uco-types:Hash",
+            "uco-types:hashMethod": {
+                "@type": "uco-vocabulary:HashNameVocab",
+                "@value": "SHA256"
+            },
+            "uco-types:hashValue": {
+                "@type": "xsd:hexBinary",
+                "@value": "9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a08"
+            }
+        },
+        {
+            "@id": "kb:hash-49e81fee-c6b3-5f5f-af8b-0746d32e4932",
+            "@type": "uco-types:Hash",
+            "uco-types:hashMethod": {
+                "@type": "uco-vocabulary:HashNameVocab",
+                "@value": "SHA512"
+            },
+            "uco-types:hashValue": {
+                "@type": "xsd:hexBinary",
+                "@value": "ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a8ff"
+            }
         }
     ]
 }
diff --git a/tests/case_utils/case_file/kb.ttl b/tests/case_utils/case_file/kb.ttl
@@ -8,51 +8,65 @@
 @prefix uco-vocabulary: <https://ontology.unifiedcyberontology.org/uco/vocabulary/> .
 @prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
 
+kb:content-data-facet-1833f979-1f19-5543-9d82-6cefd144b169
+	a uco-observable:ContentDataFacet ;
+	uco-observable:hash
+		kb:hash-24644904-83ea-5911-aea8-be687a9f3caf ,
+		kb:hash-295bdeb5-7f23-5a3f-8b7f-4bb1191b7c21 ,
+		kb:hash-39127f5c-598b-51d4-a720-2e949f18f85f ,
+		kb:hash-49e81fee-c6b3-5f5f-af8b-0746d32e4932
+		;
+	uco-observable:sizeInBytes "4"^^xsd:integer ;
+	.
+
 kb:file-800784de-5c9e-5eb2-b843-0ac51a1bd4b9
 	a uco-observable:File ;
-	uco-core:hasFacet [
-		a uco-observable:FileFacet ;
-		uco-observable:fileName "sample.txt" ;
-		uco-observable:modifiedTime "2010-01-02T03:04:56+00:00"^^xsd:dateTime ;
-		uco-observable:sizeInBytes "4"^^xsd:integer ;
-	] ;
+	uco-core:hasFacet kb:file-facet-ffa3e6bb-dffc-549d-a7c4-ffc5e90ac55d ;
 	.
 
 kb:file-ace6460a-92a9-58b9-83ea-a18ae87f6e04
 	a uco-observable:File ;
 	uco-core:hasFacet
-		[
-			a uco-observable:ContentDataFacet ;
-			uco-observable:hash
-				[
-					a uco-types:Hash ;
-					uco-types:hashMethod "MD5"^^uco-vocabulary:HashNameVocab ;
-					uco-types:hashValue "098f6bcd4621d373cade4e832627b4f6"^^xsd:hexBinary ;
-				] ,
-				[
-					a uco-types:Hash ;
-					uco-types:hashMethod "SHA1"^^uco-vocabulary:HashNameVocab ;
-					uco-types:hashValue "a94a8fe5ccb19ba61c4c0873d391e987982fbbd3"^^xsd:hexBinary ;
-				] ,
-				[
-					a uco-types:Hash ;
-					uco-types:hashMethod "SHA256"^^uco-vocabulary:HashNameVocab ;
-					uco-types:hashValue "9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a08"^^xsd:hexBinary ;
-				] ,
-				[
-					a uco-types:Hash ;
-					uco-types:hashMethod "SHA512"^^uco-vocabulary:HashNameVocab ;
-					uco-types:hashValue "ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a8ff"^^xsd:hexBinary ;
-				]
-				;
-			uco-observable:sizeInBytes "4"^^xsd:integer ;
-		] ,
-		[
-			a uco-observable:FileFacet ;
-			uco-observable:fileName "sample.txt" ;
-			uco-observable:modifiedTime "2010-01-02T03:04:56+00:00"^^xsd:dateTime ;
-			uco-observable:sizeInBytes "4"^^xsd:integer ;
-		]
+		kb:content-data-facet-1833f979-1f19-5543-9d82-6cefd144b169 ,
+		kb:file-facet-1297a4bd-563b-54c3-ad8a-f67f01ba9b10
 		;
 	.
 
+kb:file-facet-1297a4bd-563b-54c3-ad8a-f67f01ba9b10
+	a uco-observable:FileFacet ;
+	uco-observable:fileName "sample.txt" ;
+	uco-observable:modifiedTime "2010-01-02T03:04:56+00:00"^^xsd:dateTime ;
+	uco-observable:sizeInBytes "4"^^xsd:integer ;
+	.
+
+kb:file-facet-ffa3e6bb-dffc-549d-a7c4-ffc5e90ac55d
+	a uco-observable:FileFacet ;
+	uco-observable:fileName "sample.txt" ;
+	uco-observable:modifiedTime "2010-01-02T03:04:56+00:00"^^xsd:dateTime ;
+	uco-observable:sizeInBytes "4"^^xsd:integer ;
+	.
+
+kb:hash-24644904-83ea-5911-aea8-be687a9f3caf
+	a uco-types:Hash ;
+	uco-types:hashMethod "MD5"^^uco-vocabulary:HashNameVocab ;
+	uco-types:hashValue "098f6bcd4621d373cade4e832627b4f6"^^xsd:hexBinary ;
+	.
+
+kb:hash-295bdeb5-7f23-5a3f-8b7f-4bb1191b7c21
+	a uco-types:Hash ;
+	uco-types:hashMethod "SHA1"^^uco-vocabulary:HashNameVocab ;
+	uco-types:hashValue "a94a8fe5ccb19ba61c4c0873d391e987982fbbd3"^^xsd:hexBinary ;
+	.
+
+kb:hash-39127f5c-598b-51d4-a720-2e949f18f85f
+	a uco-types:Hash ;
+	uco-types:hashMethod "SHA256"^^uco-vocabulary:HashNameVocab ;
+	uco-types:hashValue "9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a08"^^xsd:hexBinary ;
+	.
+
+kb:hash-49e81fee-c6b3-5f5f-af8b-0746d32e4932
+	a uco-types:Hash ;
+	uco-types:hashMethod "SHA512"^^uco-vocabulary:HashNameVocab ;
+	uco-types:hashValue "ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a8ff"^^xsd:hexBinary ;
+	.
+
diff --git a/tests/case_utils/case_file/sample.txt-disable_hashes.ttl b/tests/case_utils/case_file/sample.txt-disable_hashes.ttl
@@ -8,11 +8,13 @@
 
 kb:file-800784de-5c9e-5eb2-b843-0ac51a1bd4b9
 	a uco-observable:File ;
-	uco-core:hasFacet [
-		a uco-observable:FileFacet ;
-		uco-observable:fileName "sample.txt" ;
-		uco-observable:modifiedTime "2010-01-02T03:04:56+00:00"^^xsd:dateTime ;
-		uco-observable:sizeInBytes "4"^^xsd:integer ;
-	] ;
+	uco-core:hasFacet kb:file-facet-ffa3e6bb-dffc-549d-a7c4-ffc5e90ac55d ;
+	.
+
+kb:file-facet-ffa3e6bb-dffc-549d-a7c4-ffc5e90ac55d
+	a uco-observable:FileFacet ;
+	uco-observable:fileName "sample.txt" ;
+	uco-observable:modifiedTime "2010-01-02T03:04:56+00:00"^^xsd:dateTime ;
+	uco-observable:sizeInBytes "4"^^xsd:integer ;
 	.