multilingual rule #192

pom.xml

@@ -5,7 +5,7 @@
   <groupId>de.gwdg.metadataqa</groupId>
   <artifactId>metadata-qa-api</artifactId>
   <packaging>jar</packaging>
-  <version>0.9.5</version>
+  <version>0.9.6-SNAPSHOT</version>
   <name>Metadata Quality Assurance Framework API</name>
   <description>
     A metadata quality assurance framework. It checks some metrics of 

src/main/java/de/gwdg/metadataqa/api/configuration/schema/Rule.java

@@ -53,6 +53,8 @@ public class Rule implements Serializable {
   private Boolean skip = Boolean.FALSE;
   private Boolean debug = Boolean.FALSE;
   private Boolean allowEmptyInstances = Boolean.FALSE;
+  private Boolean multilingual;
+  private Boolean hasLanguageTag;
 
   public String getId() {
     return id;
@@ -533,6 +535,32 @@ public Rule withAllowEmptyInstances(Boolean allowEmptyInstances) {
     return this;
   }
 
+  public Boolean getMultilingual() {
+    return multilingual;
+  }
+
+  public void setIsMultilingual(Boolean multilingual) {
+    this.multilingual = multilingual;
+  }
+
+  public Rule withMultilingual(Boolean multilingual) {
+    this.multilingual = multilingual;
+    return this;
+  }
+
+  public Boolean getHasLanguageTag() {
+    return hasLanguageTag;
+  }
+
+  public void setHasLanguageTag(Boolean hasLanguageTag) {
+    this.hasLanguageTag = hasLanguageTag;
+  }
+
+  public Rule withHasLanguageTag(Boolean hasLanguageTag) {
+    this.hasLanguageTag = hasLanguageTag;
+    return this;
+  }
+
   @JsonIgnore
   public List<String> getRulenames() {
     List<String> excludeFromComparision = List.of("serialVersionUID", "id", "description",

src/main/java/de/gwdg/metadataqa/api/rule/singlefieldchecker/LanguageTagChecker.java

@@ -0,0 +1,49 @@
+package de.gwdg.metadataqa.api.rule.singlefieldchecker;
+
+import de.gwdg.metadataqa.api.counter.FieldCounter;
+import de.gwdg.metadataqa.api.json.DataElement;
+import de.gwdg.metadataqa.api.model.XmlFieldInstance;
+import de.gwdg.metadataqa.api.model.selector.Selector;
+import de.gwdg.metadataqa.api.rule.RuleCheckerOutput;
+import de.gwdg.metadataqa.api.rule.RuleCheckingOutputStatus;
+import de.gwdg.metadataqa.api.rule.RuleCheckingOutputType;
+import de.gwdg.metadataqa.api.uniqueness.UniquenessExtractor;
+
+import java.util.List;
+
+public class LanguageTagChecker extends SingleFieldChecker {
+  private static final long serialVersionUID = 7236047216814906713L;
+  public static final String PREFIX = "languageTag";
+
+  public LanguageTagChecker(DataElement field) {
+    this(field, field.getLabel());
+  }
+
+  public LanguageTagChecker(DataElement field, String header) {
+    super(field, header);
+  }
+
+  @Override
+  public void update(Selector cache, FieldCounter<RuleCheckerOutput> results, RuleCheckingOutputType outputType) {
+    if (isDebug())
+      LOGGER.info(this.getClass() + " " + this.id);
+    var allPassed = true;
+    var isNA = true;
+    List<XmlFieldInstance> instances = cache.get(field);
+    if (instances != null && !instances.isEmpty()) {
+      for (XmlFieldInstance instance : instances) {
+        if (instance.hasLanguage()) {
+          isNA = false;
+          if (isDebug())
+            LOGGER.info("language tag: " + instance.hasLanguage());
+          break;
+        }
+      }
+    }
+
+    addOutput(results, isNA, allPassed, outputType);
+    if (isDebug())
+      LOGGER.info(this.getClass().getSimpleName() + " " + this.id + ") result: " + RuleCheckingOutputStatus.create(isNA, allPassed));
+
+  }
+}

src/main/java/de/gwdg/metadataqa/api/rule/singlefieldchecker/MultilingualChecker.java

@@ -0,0 +1,51 @@
+package de.gwdg.metadataqa.api.rule.singlefieldchecker;
+
+import de.gwdg.metadataqa.api.counter.FieldCounter;
+import de.gwdg.metadataqa.api.json.DataElement;
+import de.gwdg.metadataqa.api.model.XmlFieldInstance;
+import de.gwdg.metadataqa.api.model.selector.Selector;
+import de.gwdg.metadataqa.api.rule.RuleCheckerOutput;
+import de.gwdg.metadataqa.api.rule.RuleCheckingOutputStatus;
+import de.gwdg.metadataqa.api.rule.RuleCheckingOutputType;
+import de.gwdg.metadataqa.api.uniqueness.UniquenessExtractor;
+
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+public class MultilingualChecker extends SingleFieldChecker {
+  private static final long serialVersionUID = 3911175767180059821L;
+  public static final String PREFIX = "multilingual";
+
+  public MultilingualChecker(DataElement field) {
+    this(field, field.getLabel());
+  }
+
+  public MultilingualChecker(DataElement field, String header) {
+    super(field, header);
+  }
+
+  @Override
+  public void update(Selector cache, FieldCounter<RuleCheckerOutput> results, RuleCheckingOutputType outputType) {
+    if (isDebug())
+      LOGGER.info(this.getClass() + " " + this.id);
+    var allPassed = true;
+    var isNA = true;
+    Set<String> languages = new HashSet<>();
+    List<XmlFieldInstance> instances = cache.get(field);
+    if (instances != null && !instances.isEmpty()) {
+      for (XmlFieldInstance instance : instances) {
+        if (instance.hasLanguage()) {
+          languages.add(instance.getLanguage());
+          isNA = false;
+        }
+      }
+    }
+    if (languages.size() < 2)
+      allPassed = false;
+
+    addOutput(results, isNA, allPassed, outputType);
+    if (isDebug())
+      LOGGER.info(this.getClass().getSimpleName() + " " + this.id + ") result: " + RuleCheckingOutputStatus.create(isNA, allPassed));
+  }
+}

src/main/java/de/gwdg/metadataqa/api/schema/SchemaUtils.java

@@ -14,12 +14,14 @@
 import de.gwdg.metadataqa.api.rule.pairchecker.EqualityChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.HasValueChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.ImageDimensionChecker;
+import de.gwdg.metadataqa.api.rule.singlefieldchecker.LanguageTagChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.MaxCountChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.MaxLengthChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.MaxWordsChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.MinCountChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.MinLengthChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.MinWordsChecker;
+import de.gwdg.metadataqa.api.rule.singlefieldchecker.MultilingualChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.NumericValueChecker;
 import de.gwdg.metadataqa.api.rule.singlefieldchecker.PatternChecker;
 import de.gwdg.metadataqa.api.rule.RuleChecker;
@@ -126,6 +128,14 @@ private static List<RuleChecker> processRule(Schema schema, DataElement dataElem
     if (rule.getUnique() != null && rule.getUnique().equals(Boolean.TRUE))
       ruleCheckers.add(new UniquenessChecker(dataElement));
 
+    // TODO
+    if (rule.getMultilingual() != null && rule.getMultilingual().equals(Boolean.TRUE))
+      ruleCheckers.add(new MultilingualChecker(dataElement));
+
+    // TODO
+    if (rule.getHasLanguageTag() != null && rule.getHasLanguageTag().equals(Boolean.TRUE))
+      ruleCheckers.add(new LanguageTagChecker(dataElement));
+
     if (rule.getLessThan() != null)
       pair(schema, ruleCheckers, dataElement, rule.getLessThan(), "LessThan");
 

src/test/java/de/gwdg/metadataqa/api/cli/VersionTest.java

@@ -6,7 +6,7 @@
 
 public class VersionTest {
 
-  private final String EXPECTED_VERSION = "0.9.5";
+  private final String EXPECTED_VERSION = "0.9.6-SNAPSHOT";
 
   @Test
   public void getVersion() {

src/test/java/de/gwdg/metadataqa/api/configuration/SchemaConfigurationTest.java

@@ -350,8 +350,20 @@ public void yaml_contentType() throws FileNotFoundException {
   }
 
   @Test
-  public void yaml_ssLanguageTagged() throws FileNotFoundException {
+  public void yaml_asLanguageTagged() throws FileNotFoundException {
     Schema schema = ConfigurationReader.readSchemaYaml("src/test/resources/configuration/schema/asLanguageTagged.yaml").asSchema();
     assertEquals(true, schema.getPathByLabel("description").isAsLanguageTagged());
   }
+
+  @Test
+  public void yaml_isMultilingual() throws FileNotFoundException {
+    Schema schema = ConfigurationReader.readSchemaYaml("src/test/resources/configuration/schema/rules/isMultilingual.yaml").asSchema();
+    assertEquals(true, schema.getPathByLabel("description").getRules().get(0).getMultilingual());
+  }
+
+  @Test
+  public void yaml_hasLanguaggeTag() throws FileNotFoundException {
+    Schema schema = ConfigurationReader.readSchemaYaml("src/test/resources/configuration/schema/rules/hasLanguageTag.yaml").asSchema();
+    assertEquals(true, schema.getPathByLabel("description").getRules().get(0).getHasLanguageTag());
+  }
 }

src/test/resources/configuration/multilingual.json

@@ -0,0 +1 @@
+{"id":"1","description":[{"@lang":"de","#value":"Portr\u00e4t"},{"@lang":"zh","#value":"\u8096\u50cf"}]}

src/test/resources/configuration/schema/rules/hasLanguageTag.yaml

@@ -0,0 +1,11 @@
+format: json
+fields:
+  - name: about
+    path:  $.['about']
+    rules:
+      - lessThanOrEquals: description
+  - name: description
+    path: $.['description']
+    asLanguageTagged: true
+    rules:
+      - hasLanguageTag: true

src/test/resources/configuration/schema/rules/isMultilingual.yaml

@@ -0,0 +1,11 @@
+format: json
+fields:
+  - name: about
+    path:  $.['about']
+    rules:
+      - lessThanOrEquals: description
+  - name: description
+    path: $.['description']
+    asLanguageTagged: true
+    rules:
+      - isMultilingual: true