Merge pull request #2349 from afkbrb/issue-2103-description-validator

Issue 2103 description validator

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

@@ -156,5 +156,15 @@
     "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/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."
+    "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.",
+    "warnings-messages/item-description-too-long/title": "Description is too long",
+    "warnings-messages/item-description-too-long/body": "Description ({lang}) such as <span class=\"wb-issue-preformat\">{description}</span> on {example_entity} is too long. Its length is {length}, which is more than {max_length}. Descriptions are not full sentences, but small bits of information. In most cases, the proper length is between two and twelve words. See the <a href=\"https://www.wikidata.org/wiki/Help:Description#Length\" target=\"_blank\">manual</a> for more information.",
+    "warnings-messages/item-description-identical-with-label/title": "Description is identical with label",
+    "warnings-messages/item-description-identical-with-label/body": "Both the description ({lang}) and the label ({label_lang}) on {example_entity} are <span class=\"wb-issue-preformat\">{description}</span>. Description are expected to be more specific than labels. See the <a href=\"https://www.wikidata.org/wiki/Help:Description\" target=\"_blank\">manual</a> for more information.",
+    "warnings-messages/item-description-end-by-punctuation-sign/title": "Description ends by punctuation sign",
+    "warnings-messages/item-description-end-by-punctuation-sign/body": "Description ({lang}) such as <span class=\"wb-issue-preformat\">{description}</span> on {example_entity} ends by a punctuation sign \"{punctuation_sign}\". Description are not sentences, so the punctuation sign at the end should be avoided. See the <a href=\"https://www.wikidata.org/wiki/Help:Description#Length\" target=\"_blank\">manual</a> for more information.",
+    "warnings-messages/item-description-begin-with-uppercase/title": "Description begins with uppercase letter",
+    "warnings-messages/item-description-begin-with-uppercase/body": "Description ({lang}) such as <span class=\"wb-issue-preformat\">{description}</span> on {example_entity} begins with uppercase letter \"{uppercase_letter}\". Descriptions begin with a lowercase letter except when uppercase would normally be required or expected. See the <a href=\"https://www.wikidata.org/wiki/Help:Description#Capitalization\" target=\"_blank\">manual</a> for more information.",
+    "warnings-messages/item-description-begin-with-article/title": "Description begins with article (\"a\", \"an\" or \"the\")",
+    "warnings-messages/item-description-begin-with-article/body": "Description ({lang}) such as <span class=\"wb-issue-preformat\">{description}</span> on {example_entity} begins with article \"{article}\". Descriptions should not normally begin with initial articles (\"a\", \"an\", \"the\"). See the <a href=\"https://www.wikidata.org/wiki/Help:Description#No_initial_articles_(a,_an,_the)\" target=\"_blank\">manual</a> for more information."
 }

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

@@ -28,22 +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.*;
-import org.openrefine.wikidata.qa.scrutinizers.DistinctValuesScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.EditScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.EntityTypeScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.FormatScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.InverseConstraintScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.NewItemScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.NoEditsMadeScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.QualifierCompatibilityScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.QuantityScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.RestrictedPositionScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.RestrictedValuesScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.SelfReferentialScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.SingleValueScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.UnsourcedScrutinizer;
-import org.openrefine.wikidata.qa.scrutinizers.WhitespaceScrutinizer;
 import org.openrefine.wikidata.updates.ItemUpdate;
 import org.openrefine.wikidata.updates.scheduler.WikibaseAPIUpdateScheduler;
 import org.openrefine.wikidata.utils.EntityCache;
@@ -82,6 +67,8 @@ public class EditInspector {
         register(new RestrictedValuesScrutinizer());
         register(new EntityTypeScrutinizer());
         register(new CalendarScrutinizer());
+        register(new CommonDescriptionScrutinizer());
+        register(new EnglishDescriptionScrutinizer());
     }
 
     /**

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

@@ -0,0 +1,58 @@
+package org.openrefine.wikidata.qa.scrutinizers;
+
+import org.openrefine.wikidata.qa.QAWarning;
+import org.openrefine.wikidata.updates.ItemUpdate;
+import org.wikidata.wdtk.datamodel.interfaces.MonolingualTextValue;
+
+import java.util.Set;
+
+/**
+ * @author Lu Liu
+ */
+public class CommonDescriptionScrutinizer extends DescriptionScrutinizer {
+
+    public static final String descTooLongType = "item-description-too-long";
+    public static final String descIdenticalWithLabel = "item-description-identical-with-label";
+
+    @Override
+    public void scrutinize(ItemUpdate update, String descText, String lang) {
+        checkLength(update, descText, lang);
+        checkLabel(update, descText, lang);
+    }
+
+    // Descriptions are not full sentences, but small bits of information.
+    // In most cases, the proper length is between two and twelve words.
+    protected void checkLength(ItemUpdate update, String descText, String lang) {
+        final int maxLength = 250;
+        if (descText.length() > maxLength) {
+            QAWarning issue = new QAWarning(descTooLongType, null, QAWarning.Severity.CRITICAL, 1);
+            issue.setProperty("example_entity", update.getItemId());
+            issue.setProperty("description", descText);
+            issue.setProperty("lang", lang);
+            issue.setProperty("length", descText.length());
+            issue.setProperty("max_length", maxLength);
+            addIssue(issue);
+        }
+    }
+
+    // Description are expected to be more specific than labels.
+    protected void checkLabel(ItemUpdate update, String descText, String lang) {
+        Set<MonolingualTextValue> labels = update.getLabels();
+        labels.addAll(update.getLabelsIfNew()); // merge
+        for (MonolingualTextValue label : labels) {
+            String labelText = label.getText();
+            if (labelText == null) continue;
+            labelText = labelText.trim();
+            if (labelText.equals(descText)) {
+                QAWarning issue = new QAWarning(descIdenticalWithLabel, null, QAWarning.Severity.WARNING, 1);
+                issue.setProperty("example_entity", update.getItemId());
+                issue.setProperty("description", descText);
+                issue.setProperty("lang", lang);
+                issue.setProperty("label_lang", label.getLanguageCode());
+                addIssue(issue);
+                break;
+            }
+        }
+    }
+
+}

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

@@ -0,0 +1,30 @@
+package org.openrefine.wikidata.qa.scrutinizers;
+
+import org.openrefine.wikidata.qa.QAWarning;
+import org.openrefine.wikidata.updates.ItemUpdate;
+import org.wikidata.wdtk.datamodel.interfaces.MonolingualTextValue;
+
+import java.util.Set;
+
+/**
+ * @author Lu Liu
+ */
+public abstract class DescriptionScrutinizer extends EditScrutinizer {
+
+    @Override
+    public void scrutinize(ItemUpdate update) {
+        Set<MonolingualTextValue> descriptions = update.getDescriptions();
+        descriptions.addAll(update.getDescriptionsIfNew()); // merge
+        for (MonolingualTextValue description : descriptions) {
+            String descText = description.getText();
+            if (descText == null) continue;
+            descText = descText.trim();
+            if (descText.length() == 0) continue; // avoid NullPointerException
+
+            scrutinize(update, descText, description.getLanguageCode());
+        }
+    }
+
+    public abstract void scrutinize(ItemUpdate update, String descText, String lang);
+
+}

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

@@ -0,0 +1,72 @@
+package org.openrefine.wikidata.qa.scrutinizers;
+
+import org.openrefine.wikidata.qa.QAWarning;
+import org.openrefine.wikidata.updates.ItemUpdate;
+
+/**
+ * @author Lu Liu
+ */
+public class EnglishDescriptionScrutinizer extends DescriptionScrutinizer {
+
+    public static final String descEndsByPunctuationSign = "item-description-end-by-punctuation-sign";
+    public static final String descBeginWithUppercase = "item-description-begin-with-uppercase";
+    public static final String descBeginWithArticle = "item-description-begin-with-article";
+
+    private static final String LANG = "en";
+
+    @Override
+    public void scrutinize(ItemUpdate update, String descText, String lang) {
+        if (!LANG.equalsIgnoreCase(lang)) return;
+
+        checkPunctuationSign(update, descText);
+        checkUppercase(update, descText);
+        checkArticle(update, descText);
+    }
+
+    // Description are not sentences, so the punctuation sign at the end should be avoided.
+    protected void checkPunctuationSign(ItemUpdate update, String descText) {
+        assert descText.length() > 0;
+        final String punctuationSigns = ".?!;:,'\"";
+
+        char last = descText.charAt(descText.length() - 1);
+        if (punctuationSigns.indexOf(last) != -1) {
+            QAWarning issue = new QAWarning(descEndsByPunctuationSign, null, QAWarning.Severity.WARNING, 1);
+            issue.setProperty("example_entity", update.getItemId());
+            issue.setProperty("description", descText);
+            issue.setProperty("lang", LANG);
+            issue.setProperty("punctuation_sign", last);
+            addIssue(issue);
+        }
+    }
+
+    // Descriptions begin with a lowercase letter except when uppercase would normally be required or expected.
+    protected void checkUppercase(ItemUpdate update, String descText) {
+        assert descText.length() > 0;
+
+        char first = descText.charAt(0);
+        if ('A' <= first && first <= 'Z') {
+            QAWarning issue = new QAWarning(descBeginWithUppercase, null, QAWarning.Severity.INFO, 1);
+            issue.setProperty("example_entity", update.getItemId());
+            issue.setProperty("description", descText);
+            issue.setProperty("lang", LANG);
+            issue.setProperty("uppercase_letter", first);
+            addIssue(issue);
+        }
+    }
+
+    // Descriptions should not normally begin with initial articles ("a", "an", "the").
+    protected void checkArticle(ItemUpdate update, String descText) {
+        assert descText.length() > 0;
+
+        String firstWord = descText.split("\\s")[0].toLowerCase();
+        if ("a".equals(firstWord) || "an".equals(firstWord) || "the".equals(firstWord)) {
+            QAWarning issue = new QAWarning(descBeginWithArticle, null, QAWarning.Severity.WARNING, 1);
+            issue.setProperty("example_entity", update.getItemId());
+            issue.setProperty("description", descText);
+            issue.setProperty("lang", LANG);
+            issue.setProperty("article", firstWord);
+            addIssue(issue);
+        }
+    }
+
+}

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

@@ -0,0 +1,74 @@
+package org.openrefine.wikidata.qa.scrutinizers;
+
+import org.openrefine.wikidata.testing.TestingData;
+import org.openrefine.wikidata.updates.ItemUpdate;
+import org.openrefine.wikidata.updates.ItemUpdateBuilder;
+import org.testng.annotations.Test;
+import org.wikidata.wdtk.datamodel.helpers.Datamodel;
+
+public class CommonDescriptionScrutinizerTest extends ScrutinizerTest {
+
+    @Override
+    public EditScrutinizer getScrutinizer() {
+        return new CommonDescriptionScrutinizer();
+    }
+
+    @Test
+    public void testGoodDesc() {
+        String description = "good description";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .build();
+        scrutinize(update);
+        assertNoWarningRaised();
+    }
+
+    @Test
+    public void testTooLong() {
+        String description = "long description long description long description long description "
+                + "long description long description long description long description "
+                + "long description long description long description long description "
+                + "long description long description long description long description";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .build();
+        scrutinize(update);
+        assertWarningsRaised(CommonDescriptionScrutinizer.descTooLongType);
+    }
+
+    @Test
+    public void testIdenticalWithLabel() {
+        String description = "identical with label";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .addLabel(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .build();
+        scrutinize(update);
+        assertWarningsRaised(CommonDescriptionScrutinizer.descIdenticalWithLabel);
+    }
+
+    @Test
+    public void testIdenticalWithLabel1() {
+        String description = "identical with label";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .addLabel(Datamodel.makeMonolingualTextValue("bonjour", "fr"), true)
+                .build();
+        scrutinize(update);
+        assertNoWarningRaised();
+    }
+
+    @Test
+    public void testAwfulDesc() {
+        String description = "long description long description long description long description "
+                + "long description long description long description long description "
+                + "long description long description long description long description "
+                + "long description long description long description long description";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .addLabel(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .build();
+        scrutinize(update);
+        assertWarningsRaised(CommonDescriptionScrutinizer.descTooLongType, CommonDescriptionScrutinizer.descIdenticalWithLabel);
+    }
+}

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

@@ -0,0 +1,67 @@
+package org.openrefine.wikidata.qa.scrutinizers;
+
+import org.openrefine.wikidata.testing.TestingData;
+import org.openrefine.wikidata.updates.ItemUpdate;
+import org.openrefine.wikidata.updates.ItemUpdateBuilder;
+import org.testng.annotations.Test;
+import org.wikidata.wdtk.datamodel.helpers.Datamodel;
+
+public class EnglishDescriptionScrutinizerTest extends ScrutinizerTest {
+
+    @Override
+    public EditScrutinizer getScrutinizer() {
+        return new EnglishDescriptionScrutinizer();
+    }
+
+    @Test
+    public void testGoodDesc() {
+        String description = "good description";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .build();
+        scrutinize(update);
+        assertNoWarningRaised();
+    }
+
+    @Test
+    public void testEndWithPunctuationSign() {
+        String description = "description with punctuationSign.";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), false)
+                .build();
+        scrutinize(update);
+        assertWarningsRaised(EnglishDescriptionScrutinizer.descEndsByPunctuationSign);
+    }
+
+    @Test
+    public void testBeginWithUppercase() {
+        String description = "Begin with uppercase";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .build();
+        scrutinize(update);
+        assertWarningsRaised(EnglishDescriptionScrutinizer.descBeginWithUppercase);
+    }
+
+    @Test
+    public void testBeginWithArticle() {
+        String description = "an article test";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), false)
+                .build();
+        scrutinize(update);
+        assertWarningsRaised(EnglishDescriptionScrutinizer.descBeginWithArticle);
+    }
+
+    @Test
+    public void testAwfulDesc() {
+        String description = "An awful description.";
+        ItemUpdate update = new ItemUpdateBuilder(TestingData.newIdA)
+                .addDescription(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .addLabel(Datamodel.makeMonolingualTextValue(description, "en"), true)
+                .build();
+        scrutinize(update);
+        assertWarningsRaised(EnglishDescriptionScrutinizer.descEndsByPunctuationSign,
+                EnglishDescriptionScrutinizer.descBeginWithUppercase, EnglishDescriptionScrutinizer.descBeginWithArticle);
+    }
+}