From 699ec160821785271db1cbf41961f9d81d522281 Mon Sep 17 00:00:00 2001
From: Antonin Delpeuch <antonin@delpeuch.eu>
Date: Fri, 12 Jan 2018 19:32:45 +0000
Subject: [PATCH] Report duplicate items in DistinctValuesScrutinizer

---
 .../wikidata/module/langs/translation-en.json      |  2 +-
 .../qa/scrutinizers/DistinctValuesScrutinizer.java | 14 ++++++++------
 2 files changed, 9 insertions(+), 7 deletions(-)

diff --git a/extensions/wikidata/module/langs/translation-en.json b/extensions/wikidata/module/langs/translation-en.json
index e51280508..c54bcc76b 100644
--- a/extensions/wikidata/module/langs/translation-en.json
+++ b/extensions/wikidata/module/langs/translation-en.json
@@ -95,7 +95,7 @@
                 },
                 "identical-values-for-distinct-valued-property": {
                         "title": "Identical values for {property_entity}",
-                        "body": "This property should have distinct values."
+                        "body": "This property should have distinct values, but the same value was found on {item1_entity} and {item2_entity} for instance."
                 },
                 "no-edit-generated": {
                         "title": "No edit was generated.",
diff --git a/extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/DistinctValuesScrutinizer.java b/extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/DistinctValuesScrutinizer.java
index 27a26080d..336ccacb7 100644
--- a/extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/DistinctValuesScrutinizer.java
+++ b/extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/DistinctValuesScrutinizer.java
@@ -20,7 +20,7 @@ import org.wikidata.wdtk.datamodel.interfaces.Value;
  */
 public class DistinctValuesScrutinizer extends StatementScrutinizer {
     
-    private Map<PropertyIdValue, Set<Value>> _seenValues;
+    private Map<PropertyIdValue, Map<Value, EntityIdValue>> _seenValues;
     
     public DistinctValuesScrutinizer() {
         _seenValues = new HashMap<>();
@@ -31,22 +31,24 @@ public class DistinctValuesScrutinizer extends StatementScrutinizer {
         PropertyIdValue pid = statement.getClaim().getMainSnak().getPropertyId();
         if (_fetcher.hasDistinctValues(pid)) {
             Value mainSnakValue = statement.getClaim().getMainSnak().getValue();
-            Set<Value> seen = _seenValues.get(pid);
+            Map<Value, EntityIdValue> seen = _seenValues.get(pid);
             if (seen == null) {
-                seen = new HashSet<Value>();
+                seen = new HashMap<Value, EntityIdValue>();
                 _seenValues.put(pid, seen);
             }
-            if (seen.contains(mainSnakValue)) {
+            if (seen.containsKey(mainSnakValue)) {
+                EntityIdValue otherId = seen.get(mainSnakValue);
                 QAWarning issue = new QAWarning(
                         "identical-values-for-distinct-valued-property",
                         pid.getId(),
                         QAWarning.Severity.IMPORTANT,
                         1);
                 issue.setProperty("property_entity", pid);
-                // TODO also report the items on which the property is duplicated
+                issue.setProperty("item1_entity", entityId);
+                issue.setProperty("item2_entity", otherId);
                 addIssue(issue);
             } else {
-                seen.add(mainSnakValue);
+                seen.put(mainSnakValue, entityId);
             }
         }
     }