Merge pull request #187 from kit-data-manager/fix-186-handle-loops-in-data-graph

Fix handling of possible loops in data graphs

Author: Pfeil

src/main/java/edu/kit/datamanager/ro_crate/reader/RoCrateReader.java

@@ -177,22 +177,22 @@ protected Set<String> getDataEntityIds(RootDataEntity root, JsonNode graph) {
     if (root == null) { return Set.of(); }
     Map<String, Set<String>> network = makeEntityGraph(graph);
     Set<String> directDataEntities = new HashSet<>(root.hasPart);
-    return Stream.concat(
-            directDataEntities.stream(),
-            directDataEntities.stream().flatMap(entity -> getDataEntityIdsRecursive(entity, network))
-    ).collect(Collectors.toSet());
-  }
 
-  protected Stream<String> getDataEntityIdsRecursive(
-          String parent,
-          Map<String, Set<String>> network
-  ) {
-    return Stream.concat(
-            Stream.of(parent),
-            network.getOrDefault(parent, new HashSet<>()).stream()
-                    .flatMap(s -> getDataEntityIdsRecursive(s, network))
-                    .filter(Objects::nonNull)
-    );
+    Stack<String> processingQueue = new Stack<>();
+    processingQueue.addAll(directDataEntities);
+    Set<String> result = new HashSet<>();
+
+    while (!processingQueue.empty()) {
+      String currentId = processingQueue.pop();
+      result.add(currentId);
+      network.getOrDefault(currentId, new HashSet<>()).stream()
+              .filter(subId -> !result.contains(subId)) // avoid loops!
+              .forEach(subId -> {
+                result.add(subId);
+                processingQueue.add(subId);
+              });
+    }
+    return result;
   }
 
   protected String unpackId(JsonNode node) {

src/test/java/edu/kit/datamanager/ro_crate/crate/ReadAndWriteTest.java

@@ -53,6 +53,6 @@ void testReadCrateWithHasPartHierarchy() {
     RoCrateReader reader = new RoCrateReader(new FolderReader());
     RoCrate crate = reader.readCrate(ReadAndWriteTest.class.getResource("/crates/hasPartHierarchy").getPath());
     assertEquals(1, crate.getAllContextualEntities().size());
-    assertEquals(4, crate.getAllDataEntities().size());
+    assertEquals(6, crate.getAllDataEntities().size());
   }
 }

src/test/resources/crates/hasPartHierarchy/README.md

@@ -0,0 +1,12 @@
+# hasPart/isPartOf Structure in this crate
+
+This documents the hierarchy in this json file for the hasPart property.
+"isPartOf" occurences are handled the inverse way (as if it was a hasPart property the other way around).
+
+- ./
+  - sub-a.file
+  - sub-b.file
+    - sub-b-sub-a.file
+      - looper.file --> sub-b.file  
+    - sub-b-sub-b.file (isPartOf)
+    - looper-isPartOf.file --> sub-b.file

src/test/resources/crates/hasPartHierarchy/ro-crate-metadata.json

@@ -42,12 +42,24 @@
         },
         {
             "@id": "sub-b-sub-a.file",
-            "@type": "File"
+            "@type": "File",
+            "hasPart": {"@id": "looper.file"}
         },
         {
             "@id": "sub-b-sub-b.file",
             "@type": "File",
             "isPartOf": { "@id": "sub-b.file" }
+        },
+        {
+            "@id": "looper.file",
+            "@type": "File",
+            "hasPart": {"@id": "sub-b.file"}
+        },
+        {
+            "@id": "looper-isPartOf.file",
+            "@type": "File",
+            "hasPart": {"@id": "sub-b.file"},
+            "isPartOf": {"@id": "sub-b.file"}
         }
     ]
 }