adjusted xpath to find GetCapabilities request

Comprehesive tests for a difficult problem with a dataset from Norderstedt.

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

@@ -42,7 +42,7 @@ public class MDMetadata2Dataset {
     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']";
+    private static final String XPATH_OPERATION_GET_CAPABILITIES = "gmd:identificationInfo/*/srv:containsOperations/srv:SV_OperationMetadata[srv:operationName/gco:CharacterString ='GetCapabilities']";
     static IRIFactory factory = IRIFactory.iriImplementation();
     private final Model model;
     private final Map<String, String> keywordMapping = new HashMap<>();
@@ -361,6 +361,19 @@ public class MDMetadata2Dataset {
         metadata.addNamespace("srv", "http://www.isotc211.org/2005/srv");
     }
 
+    static String getServiceEndpoint(Element metadata) {
+        String url;
+        // Is it a service with a comprehensive description of operations?
+        if (metadata.selectSingleNode(XPATH_OPERATION_GET_CAPABILITIES) != null) {
+            final Element operationMetadata = (Element) metadata.selectSingleNode(XPATH_OPERATION_GET_CAPABILITIES);
+            url = repairURL(StringUtils.trim(operationMetadata.selectSingleNode("srv:connectPoint/gmd:CI_OnlineResource/gmd:linkage/gmd:URL").getStringValue()));
+        } else {
+            url = repairURL(getTextOrNull(metadata.selectSingleNode("gmd:identificationInfo[1]/*/gmd:citation/gmd:CI_Citation/gmd:identifier/gmd:MD_Identifier/gmd:code")));
+            log.warn("Using identifier as service endpoint: {}", url);
+        }
+        return url;
+    }
+
     public BridgeSettings getSettings() {
         return settings;
     }
@@ -411,16 +424,7 @@ public class MDMetadata2Dataset {
             return convertMultiFormatDataset(metadata, true);
         }
 
-        final String url;
-
-        // Is it a service with a comprehensive description of operations?
-        if (metadata.selectSingleNode(XPATH_OPERATION_GET_CAPABILITIES) != null) {
-            final Element operationMetadata = (Element) metadata.selectSingleNode(XPATH_OPERATION_GET_CAPABILITIES);
-            url = repairURL(StringUtils.trim(operationMetadata.selectSingleNode("srv:connectPoint/gmd:CI_OnlineResource/gmd:linkage/gmd:URL").getStringValue()));
-        } else {
-            url = repairURL(getTextOrNull(metadata.selectSingleNode("gmd:identificationInfo[1]/*/gmd:citation/gmd:CI_Citation/gmd:identifier/gmd:MD_Identifier/gmd:code")));
-        }
-
+        final String url = getServiceEndpoint(metadata);
         if (!isValidURLandIRI(url)) {
             log.warn("Distribution has an invalid URI {} - ignoring it.", url);
             return Collections.emptySet();
@@ -995,14 +999,7 @@ public class MDMetadata2Dataset {
 
         convertDataResource(metadata, dataService);
 
-        String url;
-        // Is it a service with a comprehensive description of operations?
-        if (metadata.selectSingleNode(XPATH_OPERATION_GET_CAPABILITIES) != null) {
-            final Element operationMetadata = (Element) metadata.selectSingleNode(XPATH_OPERATION_GET_CAPABILITIES);
-            url = repairURL(StringUtils.trim(operationMetadata.selectSingleNode("srv:connectPoint/gmd:CI_OnlineResource/gmd:linkage/gmd:URL").getStringValue()));
-        } else {
-            url = repairURL(getTextOrNull(metadata.selectSingleNode("gmd:identificationInfo[1]/*/gmd:citation/gmd:CI_Citation/gmd:identifier/gmd:MD_Identifier/gmd:code")));
-        }
+        String url = getServiceEndpoint(metadata);
 
         if (StringUtils.contains(StringUtils.lowerCase(url), "request=getcapabilities")) {
             // the GetCapabilites request is linked

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

@@ -34,6 +34,7 @@ public class MDMetadata2DatasetTests {
 
     private static final Property PROPERTY_applicable_Legislation = ResourceFactory.createProperty("http://data.europa.eu/r5r/", "applicableLegislation");
     static Namespace NS_GMD = Namespace.get("gmd", "http://www.isotc211.org/2005/gmd");
+    final Model model = ModelFactory.createDefaultModel();
     private final SAXReader saxReader = new SAXReader();
     private MDMetadata2Dataset service;
     private CswInterface cswInterface;
@@ -85,7 +86,6 @@ public class MDMetadata2DatasetTests {
 
     @Test
     public void testConvert() throws DocumentException, IOException {
-        final Model model = ModelFactory.createDefaultModel();
         model.read(getClass().getResourceAsStream("/f7f90143-c2ad-46b2-934d-93dfd5e0f031.rdf"), null);
         final Resource referenceDataset = model.listResourcesWithProperty(RDF.type, DCAT.Dataset).next();
 
@@ -550,7 +550,6 @@ public class MDMetadata2DatasetTests {
 
         Element element = new SAXReader().read(new StringReader(xml)).getRootElement();
 
-        Model model = ModelFactory.createDefaultModel();
         Resource resource = model.createResource();
 
         MDMetadata2Dataset.addTemporalExtend(resource, element);
@@ -576,7 +575,6 @@ public class MDMetadata2DatasetTests {
 
         Element element = new SAXReader().read(new StringReader(xml)).getRootElement();
 
-        Model model = ModelFactory.createDefaultModel();
         Resource resource = model.createResource();
 
         MDMetadata2Dataset.addTemporalExtend(resource, element);
@@ -725,8 +723,6 @@ public class MDMetadata2DatasetTests {
     @Test
     public void createFoafOrg_invalidEmail() throws DocumentException {
         final Element ciResponsibleParty = saxReader.read(getClass().getResourceAsStream("/CI_ResponsibleParty1.xml")).getRootElement();
-        final Model model = ModelFactory.createDefaultModel();
-
         MDMetadata2Dataset.createFoafOrg(ciResponsibleParty, model);
         model.write(Writer.nullWriter(), "RDF/XML-ABBREV");
     }
@@ -848,7 +844,6 @@ public class MDMetadata2DatasetTests {
 
     @Test
     void addBoundingBox_null() throws DocumentException {
-        Model model = ModelFactory.createDefaultModel();
         Resource resource = model.createResource();
         MDMetadata2Dataset.addBoundingBox(resource, null);
         assertFalse(model.listStatements().hasNext());
@@ -865,7 +860,6 @@ public class MDMetadata2DatasetTests {
                 "</gmd:EX_GeographicBoundingBox>";
 
         Element element = new SAXReader().read(new StringReader(xml)).getRootElement();
-        Model model = ModelFactory.createDefaultModel();
         Resource resource = model.createResource();
 
         MDMetadata2Dataset.addBoundingBox(resource, element);
@@ -905,7 +899,6 @@ public class MDMetadata2DatasetTests {
                 "</gmd:EX_GeographicBoundingBox>";
 
         Element element = new SAXReader().read(new StringReader(xml)).getRootElement();
-        Model model = ModelFactory.createDefaultModel();
         Resource resource = model.createResource();
 
         MDMetadata2Dataset.addBoundingBox(resource, element);
@@ -943,7 +936,6 @@ public class MDMetadata2DatasetTests {
                 "</gmd:EX_GeographicBoundingBox>";
 
         Element element = new SAXReader().read(new StringReader(xml)).getRootElement();
-        Model model = ModelFactory.createDefaultModel();
         Resource resource = model.createResource();
 
         MDMetadata2Dataset.addBoundingBox(resource, element);
@@ -981,7 +973,6 @@ public class MDMetadata2DatasetTests {
                 "</gmd:EX_GeographicBoundingBox>";
 
         Element element = new SAXReader().read(new StringReader(xml)).getRootElement();
-        Model model = ModelFactory.createDefaultModel();
         Resource resource = model.createResource();
 
         MDMetadata2Dataset.addBoundingBox(resource, element);
@@ -1457,4 +1448,64 @@ public class MDMetadata2DatasetTests {
                 result.getPropertyResourceValue(DCATAP.hvdCategory).getURI());
     }
 
+    @Test
+    public void getServiceEndpoint_norderstedtWFS() throws DocumentException {
+        final Element inputDocument = saxReader.read(getClass().getResourceAsStream("/norderstedt_wfs.xml")).getRootElement();
+        String result = MDMetadata2Dataset.getServiceEndpoint(inputDocument);
+        assertEquals("https://geoservice.norderstedt.de/geoserver/pub/wfs?SERVICE=WFS&version=2.0.0&request=GetCapabilities", result);
+    }
+
+    @Test
+    public void createDataService_norderstedtWMS() throws DocumentException {
+        final Element input = saxReader.read(getClass().getResourceAsStream("/norderstedt_wms.xml")).getRootElement();
+        Resource dataService = service.createDataService(input, model);
+        assertEquals("https://geoservice.norderstedt.de/geoserver/pub/soz_Wohlfahrt/wms", dataService.getPropertyResourceValue(DCAT.endpointURL).getURI());
+    }
+
+    @Test
+    public void convertServiceToDistributions_norderstedtWMS() throws DocumentException {
+        final Element input = saxReader.read(getClass().getResourceAsStream("/norderstedt_wms.xml")).getRootElement();
+        Collection<Resource> distributions = service.convertServiceToDistributions(input);
+        assertEquals(1, distributions.size());
+        Resource distribution = distributions.iterator().next();
+
+        assertEquals("https://geoservice.norderstedt.de/geoserver/pub/soz_Wohlfahrt/wms?SERVICE=WMS&version=1.3.0&request=GetCapabilities", distribution.getPropertyResourceValue(DCAT.accessURL).getURI());
+    }
+
+    @Test
+    public void getServiceEndpoint_norderstedtWMS() throws DocumentException {
+        final Element inputDocument = saxReader.read(getClass().getResourceAsStream("/norderstedt_wms.xml")).getRootElement();
+        String result = MDMetadata2Dataset.getServiceEndpoint(inputDocument);
+        assertEquals("https://geoservice.norderstedt.de/geoserver/pub/soz_Wohlfahrt/wms?SERVICE=WMS&version=1.3.0&request=GetCapabilities", result);
+    }
+
+    @Test
+    public void convert_norderstedtDataset() throws DocumentException, IOException {
+        final Element wmsElement = saxReader.read(getClass().getResourceAsStream("/norderstedt_wms.xml")).getRootElement();
+        final Element wfsElement = saxReader.read(getClass().getResourceAsStream("/norderstedt_wfs.xml")).getRootElement();
+        final Element input = saxReader.read(getClass().getResourceAsStream("/norderstedt_dataset.xml")).getRootElement();
+
+        Mockito.when(cswInterface.findOperatesOn("https://registry.gdi-de.org/id/de.sh/ea7599f9-0134-4a3f-ab46-5116addc1bb3")).thenReturn(Arrays.asList(wfsElement, wmsElement));
+
+        Resource dataset = service.convert(input);
+
+        final Map<String, Resource> distributionMap = collectDistributions(dataset);
+        assertEquals(4, distributionMap.size());
+
+        Resource distributionWMS = distributionMap.get("https://geoservice.norderstedt.de/geoserver/pub/soz_Wohlfahrt/wms?SERVICE=WMS&version=1.3.0&request=GetCapabilities");
+        Resource distributionWFS = distributionMap.get("https://geoservice.norderstedt.de/geoserver/pub/wfs?SERVICE=WFS&version=2.0.0&request=GetCapabilities");
+        Resource distributionCSV = distributionMap.get("https://geoservice.norderstedt.de/geoserver/pub/ows?service=WFS&version=1.0.0&request=GetFeature&typeName=pub:soz_Wohlfahrt&outputFormat=CSV");
+        Resource distributionSHP = distributionMap.get("https://geoservice.norderstedt.de/geoserver/pub/ows?service=WFS&version=1.0.0&request=GetFeature&typeName=pub:soz_Wohlfahrt&outputFormat=SHAPE-ZIP");
+
+        assertNotNull(distributionCSV);
+        assertNotNull(distributionSHP);
+        assertNotNull(distributionWFS);
+        assertNotNull(distributionWMS);
+
+        assertEquals("http://publications.europa.eu/resource/authority/file-type/WFS_SRVC",
+                distributionWFS.getPropertyResourceValue(DCTerms.format).getURI());
+        assertEquals("http://publications.europa.eu/resource/authority/file-type/WMS_SRVC",
+                distributionWMS.getPropertyResourceValue(DCTerms.format).getURI());
+    }
+
 }