Add self-referential scrutinizer

extensions/wikidata/module/langs/translation-en.json

@@ -49,6 +49,10 @@
                 "missing-inverse-statements": {
                         "title": "Inverse statements will not be added.",
                         "body": "Some of the properties that you are using require inverse statements. You should add them in your schema."
+                },
+                "self-referential-statements": {
+                        "title": "Some statements are self-referential.",
+                        "body": "While not forbidden, self-referential statements are generally suspicious. You could have reconciliation issues."
                 }
        }
 }

extensions/wikidata/src/org/openrefine/wikidata/qa/EditInspector.java

@@ -8,6 +8,7 @@ import org.openrefine.wikidata.qa.scrutinizers.EditScrutinizer;
 import org.openrefine.wikidata.qa.scrutinizers.FormatConstraintScrutinizer;
 import org.openrefine.wikidata.qa.scrutinizers.InverseConstraintScrutinizer;
 import org.openrefine.wikidata.qa.scrutinizers.NewItemScrutinizer;
+import org.openrefine.wikidata.qa.scrutinizers.SelfReferentialScrutinizer;
 import org.openrefine.wikidata.schema.ItemUpdate;
 
 /**
@@ -27,6 +28,7 @@ public class EditInspector {
         register(new NewItemScrutinizer());
         register(new FormatConstraintScrutinizer());
         register(new InverseConstraintScrutinizer());
+        register(new SelfReferentialScrutinizer());
     }
     
     /**

extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/FormatConstraintScrutinizer.java

@@ -5,10 +5,17 @@ import java.util.Map;
 import java.util.regex.Pattern;
 
 import org.openrefine.wikidata.qa.ConstraintFetcher;
+import org.wikidata.wdtk.datamodel.interfaces.EntityIdValue;
 import org.wikidata.wdtk.datamodel.interfaces.Snak;
 import org.wikidata.wdtk.datamodel.interfaces.StringValue;
 
-
+/**
+ * A scrutinizer that detects incorrect formats in text values
+ * (mostly identifiers).
+ * 
+ * @author antonin
+ *
+ */
 public class FormatConstraintScrutinizer extends SnakScrutinizer {
     
     private Map<String, Pattern> _patterns;
@@ -41,7 +48,7 @@ public class FormatConstraintScrutinizer extends SnakScrutinizer {
     }
 
     @Override
-    public void scrutinize(Snak snak, boolean added) {
+    public void scrutinize(Snak snak, EntityIdValue entityId, boolean added) {
         if(StringValue.class.isInstance(snak.getValue())) {
             String value = ((StringValue) snak.getValue()).getString();
             String pid = snak.getPropertyId().getId();

extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/InverseConstraintScrutinizer.java

@@ -12,7 +12,13 @@ import org.wikidata.wdtk.datamodel.interfaces.ItemIdValue;
 import org.wikidata.wdtk.datamodel.interfaces.Statement;
 import org.wikidata.wdtk.datamodel.interfaces.Value;
 
-
+/**
+ * A scrutinizer that checks for missing inverse statements in
+ * edit batches.
+ * 
+ * @author antonin
+ *
+ */
 public class InverseConstraintScrutinizer extends StatementScrutinizer {
 
     private ConstraintFetcher _fetcher;

extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/SelfReferentialScrutinizer.java

@@ -0,0 +1,22 @@
+package org.openrefine.wikidata.qa.scrutinizers;
+
+import org.wikidata.wdtk.datamodel.interfaces.EntityIdValue;
+import org.wikidata.wdtk.datamodel.interfaces.Snak;
+
+/**
+ * A scrutinizer that checks for self-referential statements.
+ * These statements are flagged by Wikibase as suspicious.
+ * 
+ * @author antonin
+ *
+ */
+public class SelfReferentialScrutinizer extends SnakScrutinizer {
+
+    @Override
+    public void scrutinize(Snak snak, EntityIdValue entityId, boolean added) {
+        if (entityId.equals(snak.getValue())) {
+            warning("self-referential-statements");
+        }
+    }
+
+}

extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/SnakScrutinizer.java

@@ -7,28 +7,41 @@ import org.wikidata.wdtk.datamodel.interfaces.Reference;
 import org.wikidata.wdtk.datamodel.interfaces.Snak;
 import org.wikidata.wdtk.datamodel.interfaces.Statement;
 
+/**
+ * A scrutinizer that inspects snaks individually, no matter whether they
+ * appear as main snaks, qualifiers or references.
+ * 
+ * @author antonin
+ *
+ */
 public abstract class SnakScrutinizer extends StatementScrutinizer {
     
-    public abstract void scrutinize(Snak snak, boolean added);
+    /**
+     * This is the method that subclasses should override to implement their checks.
+     * @param snak: the snak to inspect
+     * @param entityId: the item on which it is going to (dis)appear
+     * @param added: whether this snak is going to be added or deleted
+     */
+    public abstract void scrutinize(Snak snak, EntityIdValue entityId, boolean added);
     
     @Override
     public void scrutinize(Statement statement, EntityIdValue entityId, boolean added) {
         // Main snak
-        scrutinize(statement.getClaim().getMainSnak(), added);
+        scrutinize(statement.getClaim().getMainSnak(), entityId, added);
         
         // Qualifiers
-        scrutinizeSnakSet(statement.getClaim().getAllQualifiers(), added);
+        scrutinizeSnakSet(statement.getClaim().getAllQualifiers(), entityId, added);
         
         // References
         for(Reference ref : statement.getReferences()) {
-            scrutinizeSnakSet(ref.getAllSnaks(), added);
+            scrutinizeSnakSet(ref.getAllSnaks(), entityId, added);
         }
     }
     
-    private void scrutinizeSnakSet(Iterator<Snak> snaks, boolean added) {
+    private void scrutinizeSnakSet(Iterator<Snak> snaks, EntityIdValue entityId, boolean added) {
         while(snaks.hasNext()) {
             Snak snak = snaks.next();
-            scrutinize(snak, added);
+            scrutinize(snak, entityId, added);
         }
     }
 }