Auftrennen von vermischten deutschen und englischen Titeln

src/main/java/de/landsh/opendata/csw2dcat/MDMetadata2Dataset.java

@@ -39,6 +39,8 @@ import java.util.stream.Collectors;
 public class MDMetadata2Dataset {
 
     private static final String LANGUAGE_DE = "de";
+    private static final String LANGUAGE_EN = "en";
+
     private static final Logger log = LoggerFactory.getLogger(MDMetadata2Dataset.class);
     private static final String XPATH_OPERATION_GET_CAPABILITIES = "gmd:identificationInfo/*/srv:containsOperations/srv:SV_OperationMetadata[srv:operationName/gco:CharacterString/text()='GetCapabilities']";
     static IRIFactory factory = IRIFactory.iriImplementation();
@@ -663,8 +665,17 @@ public class MDMetadata2Dataset {
 
         final String title = getTextOrNull(metadata.selectSingleNode("gmd:identificationInfo/*/gmd:citation/gmd:CI_Citation/gmd:title/gco:CharacterString"));
         if (title != null) {
+
+            if( title.contains("#locale-eng:"))      {
+                // In some source systems, German and English titles are strangely combined in one text.
+                String germanTitle = StringUtils.substringBefore(title, "#locale-eng:");
+                String englishTitle = StringUtils.substringAfter(title, "#locale-eng:");
+                dataset.addLiteral(DCTerms.title, model.createLiteral(germanTitle.trim(), LANGUAGE_DE));
+                dataset.addLiteral(DCTerms.title, model.createLiteral(englishTitle.trim(), LANGUAGE_EN));
+            }   else {
                 dataset.addLiteral(DCTerms.title, model.createLiteral(title.trim(), LANGUAGE_DE));
             }
+        }
 
         final String description = getTextOrNull(metadata.selectSingleNode("gmd:identificationInfo/*/gmd:abstract/gco:CharacterString"));
         if (description != null) {

src/test/java/de/landsh/opendata/csw2dcat/MDMetadata2DatasetTests.java

@@ -1358,4 +1358,19 @@ public class MDMetadata2DatasetTests {
         assertEquals("http://data.europa.eu/bna/c_ac64a52d",
                 result.getPropertyResourceValue(DCATAP.hvdCategory).getURI());
     }
+
+    /**
+     *  In some source systems, German and English titles are strangely combined in one text.
+     */
+    @Test
+    public void convert_mixedGermanEnglishTitles() throws DocumentException, IOException {
+        final Document inputDocument = saxReader.read(getClass().getResourceAsStream("/00729c12-e689-446d-83bd-dd8cfd0eeeb8.xml"));
+        final Resource result = service.convert(inputDocument);
+
+        assertTrue(result.hasProperty( DCTerms.title, ResourceFactory.createLangLiteral(
+                "Robben im Schleswig-Holsteinischen Wattenmeer 1995", "de") ));
+        assertTrue(result.hasProperty( DCTerms.title, ResourceFactory.createLangLiteral(
+                "Seals in the Schleswig-Holstein Wadden Sea in 1995", "en") ));
+    }
+
 }