Add warning for early Gregorian dates in Wikidata schema (#2136)

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

@@ -154,5 +154,7 @@
     "warnings-messages/no-unit-provided/title": "Unit missing for {property_entity}",
     "warnings-messages/no-unit-provided/body": "Values such as <span class=\"wb-issue-preformat\">{example_value}</span> on {example_item_entity} are expected to have units.",
     "warnings-messages/invalid-entity-type/title": "{property_entity} used on items",
-    "warnings-messages/invalid-entity-type/body": "Uses of {property_entity} on items such as {example_entity} are invalid."
+    "warnings-messages/invalid-entity-type/body": "Uses of {property_entity} on items such as {example_entity} are invalid.",
+    "warnings-messages/early-gregorian-date/title": "Early dates in the Gregorian calendar",
+    "warnings-messages/early-gregorian-date/body": "Dates earlier than October 1582 (such as in year {example_year}) are unlikely to be expressed using the Gregorian calendar. See the <a href=\"https://www.wikidata.org/wiki/Wikidata:Tools/OpenRefine/Editing/Schema_alignment#Dates\" target=\"_blank\">manual</a> to specify the appropriate calendar for your dates."
 }

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

@@ -28,6 +28,7 @@ import java.util.List;
 import java.util.Map;
 import java.util.stream.Collectors;
 
+import org.openrefine.wikidata.qa.scrutinizers.CalendarScrutinizer;
 import org.openrefine.wikidata.qa.scrutinizers.DistinctValuesScrutinizer;
 import org.openrefine.wikidata.qa.scrutinizers.EditScrutinizer;
 import org.openrefine.wikidata.qa.scrutinizers.EntityTypeScrutinizer;
@@ -79,6 +80,7 @@ public class EditInspector {
         register(new QuantityScrutinizer());
         register(new RestrictedValuesScrutinizer());
         register(new EntityTypeScrutinizer());
+        register(new CalendarScrutinizer());
     }
 
     /**

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

@@ -0,0 +1,31 @@
+package org.openrefine.wikidata.qa.scrutinizers;
+
+import org.openrefine.wikidata.qa.QAWarning;
+import org.wikidata.wdtk.datamodel.helpers.Datamodel;
+import org.wikidata.wdtk.datamodel.interfaces.TimeValue;
+import org.wikidata.wdtk.datamodel.interfaces.Value;
+
+public class CalendarScrutinizer extends ValueScrutinizer {
+	
+    public static final String earlyGregorianDateType = "early-gregorian-date";
+	
+	public static final TimeValue earliestGregorian = Datamodel.makeTimeValue(
+			1582, (byte)10, (byte)15, (byte)0, (byte)0, (byte)0, (byte)11, 0, 0, 0, TimeValue.CM_GREGORIAN_PRO);
+
+	@Override
+	public void scrutinize(Value value) {
+		if(TimeValue.class.isInstance(value)) {
+			TimeValue time = (TimeValue)value;
+			if(time.getPreferredCalendarModel().equals(earliestGregorian.getPreferredCalendarModel()) &&
+			   time.getPrecision() >= 10 &&
+			  (time.getYear() < earliestGregorian.getYear() ||
+			   time.getYear() == earliestGregorian.getYear() && time.getMonth() < earliestGregorian.getMonth() ||
+			   time.getYear() == earliestGregorian.getYear() && time.getMonth() == earliestGregorian.getMonth() && time.getDay() < earliestGregorian.getDay())) {
+				QAWarning warning = new QAWarning(earlyGregorianDateType, null, QAWarning.Severity.WARNING, 1);
+		        warning.setProperty("example_year", Long.toString(time.getYear()));
+		        addIssue(warning);
+			}
+		}
+	}
+
+}

extensions/wikidata/tests/src/org/openrefine/wikidata/qa/scrutinizers/CalendarScrutinizerTest.java

@@ -0,0 +1,37 @@
+package org.openrefine.wikidata.qa.scrutinizers;
+
+import org.testng.annotations.Test;
+import org.wikidata.wdtk.datamodel.helpers.Datamodel;
+import org.wikidata.wdtk.datamodel.interfaces.TimeValue;
+
+public class CalendarScrutinizerTest extends ValueScrutinizerTest {
+
+	@Override
+	public EditScrutinizer getScrutinizer() {
+		return new CalendarScrutinizer();
+	}
+	
+    @Test
+    public void testScrutinizeRecentValue() {
+    	scrutinize(Datamodel.makeTimeValue(1978L, (byte)3, (byte)4, (byte)0, (byte)0, (byte)0, 11, TimeValue.CM_GREGORIAN_PRO));
+    	assertNoWarningRaised();
+    }
+    
+    @Test
+    public void testScrutinizeCloseValue() {
+    	scrutinize(Datamodel.makeTimeValue(1582L, (byte)10, (byte)17, (byte)0, (byte)0, (byte)0, 11, TimeValue.CM_GREGORIAN_PRO));
+    	assertNoWarningRaised();
+    }
+    
+    @Test
+    public void testScrutinizeEarlyYear() {
+    	scrutinize(Datamodel.makeTimeValue(1400L, (byte)1, (byte)1, (byte)0, (byte)0, (byte)0, (byte)9, 0, 0, 0, TimeValue.CM_GREGORIAN_PRO));
+    	assertNoWarningRaised();
+    }
+    
+    @Test
+    public void testScrutinizeEarlyDay() {
+    	scrutinize(Datamodel.makeTimeValue(1440L, (byte)10, (byte)17, (byte)0, (byte)0, (byte)0, 11, TimeValue.CM_GREGORIAN_PRO));
+    	assertWarningsRaised(CalendarScrutinizer.earlyGregorianDateType);
+    }
+}