html tags are removed from extras as well as description field

git-svn-id: https://svn.d4science.research-infrastructures.eu/gcube/trunk/data-catalogue/grsf-publisher-ws@135101 82a268e6-3cf1-43bd-a215-b396298e98cf

pom.xml

@@ -44,9 +44,17 @@
 		<webappDirectory>${project.build.directory}/${project.build.finalName}</webappDirectory>
 		<distroDirectory>distro</distroDirectory>
 		<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
+		<jsoup.version>1.10.1</jsoup.version>
 	</properties>
 
 	<dependencies>
+		<!-- jsoup HTML parser library @ http://jsoup.org/ -->
+		<dependency>
+			<groupId>org.jsoup</groupId>
+			<artifactId>jsoup</artifactId>
+			<version>${jsoup.version}</version>
+			<scope>compile</scope>
+		</dependency>
 		<!-- SmartGears -->
 		<dependency>
 			<groupId>org.gcube.core</groupId>

src/main/java/org/gcube/data_catalogue/grsf_publish_ws/services/GrsfPublisherFisheryService.java

@@ -223,7 +223,7 @@ public class GrsfPublisherFisheryService {
 										record.getMaintainer(), 
 										record.getMaintainerContact(), 
 										version, 
-										record.getDescription(), 
+										HelperMethods.removeHTML(record.getDescription()), 
 										license, 
 										tags, 
 										customFields, 

src/main/java/org/gcube/data_catalogue/grsf_publish_ws/services/GrsfPublisherStockService.java

@@ -213,7 +213,7 @@ public class GrsfPublisherStockService {
 										record.getMaintainer(), 
 										record.getMaintainerContact(), 
 										version, 
-										record.getDescription(), 
+										HelperMethods.removeHTML(record.getDescription()), 
 										license, 
 										tags, 
 										customFields, 

src/main/java/org/gcube/data_catalogue/grsf_publish_ws/utils/HelperMethods.java

@@ -30,6 +30,7 @@ import org.gcube.datacatalogue.ckanutillibrary.DataCatalogue;
 import org.gcube.datacatalogue.ckanutillibrary.DataCatalogueFactory;
 import org.gcube.datacatalogue.ckanutillibrary.DataCatalogueImpl;
 import org.gcube.datacatalogue.ckanutillibrary.models.ResourceBean;
+import org.jsoup.Jsoup;
 import org.slf4j.LoggerFactory;
 
 import eu.trentorise.opendata.jackan.internal.org.apache.http.HttpResponse;
@@ -214,19 +215,23 @@ public abstract class HelperMethods {
 										elementsToConsider = Math.min(elementsToConsider, TIME_SERIES_TAKE_LAST_VALUES);
 
 										for (int i = (asList.size() - elementsToConsider); i < asList.size(); i++) {
-											logger.debug(asList.get(i).toString().trim());
+											// trim and remove html
-											valuesForKey.add(asList.get(i).toString().trim());
+											String clean = HelperMethods.removeHTML(asList.get(i).toString().trim()); 
+											logger.debug(clean);
+											valuesForKey.add(clean);
 										}
 
 									}else{	
 										for (int i = 0; i < elementsToConsider; i++) {
-											logger.debug(asList.get(i).toString().trim());
+											String clean = HelperMethods.removeHTML(asList.get(i).toString().trim()); 
-											valuesForKey.add(asList.get(i).toString().trim());
+											logger.debug(clean);
+											valuesForKey.add(clean);
 										}
 									}
 								}
 							}else{
-								valuesForKey.add(f.toString().trim());
+								String clean = HelperMethods.removeHTML(f.toString().trim()); 
+								valuesForKey.add(clean);
 							}
 
 							// add to the map
@@ -352,7 +357,7 @@ public abstract class HelperMethods {
 
 			return res;
 		}catch(Exception e){
-			logger.error("error while performing post method " + e.toString());
+			logger.error("error while performing get method " + e.toString());
 		}
 
 		return null;
@@ -522,4 +527,16 @@ public abstract class HelperMethods {
 		}
 		return result;
 	}
+	
+	/**
+	 * Strip out HTML code
+	 * @param html
+	 * @return
+	 */
+	public static String removeHTML(String html) {
+		logger.debug("Incoming text is " + html);
+		String withoutHTML = Jsoup.parse(html).text();
+		logger.debug("Without html is " + withoutHTML);
+	    return withoutHTML;
+	}
 }

src/test/java/org/gcube/data_catalogue/grsf_publish_ws/GcoreEndpointGRSFService.java

@@ -0,0 +1,77 @@
+/**
+ *
+ */
+package org.gcube.data_catalogue.grsf_publish_ws;
+
+import static org.gcube.resources.discovery.icclient.ICFactory.client;
+import static org.gcube.resources.discovery.icclient.ICFactory.queryFor;
+
+import java.util.List;
+
+import org.gcube.common.resources.gcore.GCoreEndpoint;
+import org.gcube.common.scope.api.ScopeProvider;
+import org.gcube.resources.discovery.client.api.DiscoveryClient;
+import org.gcube.resources.discovery.client.queries.api.SimpleQuery;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Retrieve the GRSF Service endpoint in the Infrastructure.
+ * @author Costantino Perciante at ISTI-CNR (costantino.perciante@isti.cnr.it)
+ */
+public class GcoreEndpointGRSFService {
+
+	private static final String resource = "jersey-servlet";
+	private static final String serviceName = "GRSFPublisher";
+	private static final String serviceClass = "Data-Catalogue";
+
+	private static Logger logger = LoggerFactory.getLogger(GcoreEndpointGRSFService.class);
+
+	/**
+	 * Instantiates a new gcore endpoint reader.
+	 * @param scope the scope
+	 * @throws Exception the exception
+	 */
+	public static String getServiceUrl(String scope) throws Exception {
+
+		if(scope == null || scope.isEmpty())
+			throw new IllegalArgumentException();
+
+		String oldScope = ScopeProvider.instance.get();
+
+		try{
+
+			logger.info("set scope "+scope);
+			ScopeProvider.instance.set(scope);
+
+			SimpleQuery query = queryFor(GCoreEndpoint.class);
+			query.addCondition(String.format("$resource/Profile/ServiceClass/text() eq '%s'",serviceClass));
+			query.addCondition("$resource/Profile/DeploymentData/Status/text() eq 'ready'");
+			query.addCondition(String.format("$resource/Profile/ServiceName/text() eq '%s'",serviceName));
+			query.setResult("$resource/Profile/AccessPoint/RunningInstanceInterfaces//Endpoint[@EntryName/string() eq \""+resource+"\"]/text()");
+
+			logger.debug("submitting quey "+query.toString());
+
+			DiscoveryClient<String> client = client();
+			List<String> endpoints = client.submit(query);
+			if (endpoints == null || endpoints.isEmpty()) throw new Exception("Cannot retrieve the GCoreEndpoint serviceName: "+serviceName +", serviceClass: " +serviceClass +", in scope: "+scope);
+
+			String urlFound = endpoints.get(0);
+			if(urlFound==null)
+				throw new Exception("Endpoint:"+resource+", is null for serviceName: "+serviceName +", serviceClass: " +serviceClass +", in scope: "+scope);
+
+			logger.info("found entyname "+urlFound+" for ckanResource: "+resource);
+
+			urlFound = urlFound.replaceFirst(":80", "").replace("http", "https");
+			return urlFound;
+
+		}catch(Exception e){
+			String error = "An error occurred during GCoreEndpoint discovery, serviceName: "+serviceName +", serviceClass: " +serviceClass +", in scope: "+scope +".";
+			logger.error(error, e);
+			throw new Exception(error);
+		}finally{
+			if(oldScope != null && !oldScope.equals(scope))
+				ScopeProvider.instance.set(oldScope);
+		}
+	}
+}

src/test/java/org/gcube/data_catalogue/grsf_publish_ws/JTests.java

@@ -42,7 +42,6 @@ import org.gcube.data_catalogue.grsf_publish_ws.utils.groups.Status;
 import org.gcube.data_catalogue.grsf_publish_ws.utils.groups.Type_Fishery;
 import org.gcube.datacatalogue.ckanutillibrary.DataCatalogue;
 import org.gcube.datacatalogue.ckanutillibrary.DataCatalogueFactory;
-import org.junit.Test;
 import org.slf4j.LoggerFactory;
 
 import com.fasterxml.jackson.core.JsonProcessingException;
@@ -506,4 +505,20 @@ public class JTests {
 		return result;
 	}
 
+	//@Test
+	public void GRSFServiceUrl() throws Exception{
+		String context = "/gcube/devNext/NextNext";
+		String url = GcoreEndpointGRSFService.getServiceUrl(context);
+		logger.debug("Url is " + url);
+
+	}
+	
+	//@Test
+	public void removeHTML() throws Exception{
+		
+		String toTest = "<p>Based on the results of the MP operation for 2015&ndash;17 in its 2013 meeting and the outcome of the review of exceptional circumstances in its 2015 meeting, the ESC recommended that there is no need to revise the Extended Commission&rsquo;s 2013 TAC decision regarding the TAC for 2016&ndash;17. The recommended annual TAC for the years 2016-2017 is 14,647.4t.</p>";
+		HelperMethods.removeHTML(toTest);
+
+	}
+
 }