[CST-5303] refactoring of wos live import

dspace-api/src/main/java/org/dspace/importer/external/metadatamapping/contributor/SimpleConcatContributor.java

@@ -7,14 +7,20 @@
  */
 package org.dspace.importer.external.metadatamapping.contributor;
 
+import java.util.ArrayList;
 import java.util.Collection;
 import java.util.LinkedList;
 import java.util.List;
 
 import org.apache.commons.lang3.StringUtils;
+import org.apache.logging.log4j.LogManager;
+import org.apache.logging.log4j.Logger;
 import org.dspace.importer.external.metadatamapping.MetadatumDTO;
 import org.jdom2.Element;
 import org.jdom2.Namespace;
+import org.jdom2.filter.Filters;
+import org.jdom2.xpath.XPathExpression;
+import org.jdom2.xpath.XPathFactory;
 
 /**
  * This contributor is able to concat multi value.
@@ -25,16 +31,26 @@ import org.jdom2.Namespace;
  */
 public class SimpleConcatContributor extends SimpleXpathMetadatumContributor {
 
+    private final static Logger log = LogManager.getLogger();
+
     @Override
-    public Collection<MetadatumDTO> contributeMetadata(Element element) {
+    public Collection<MetadatumDTO> contributeMetadata(Element t) {
         List<MetadatumDTO> values = new LinkedList<>();
         StringBuilder text = new StringBuilder();
+        List<Namespace> namespaces = new ArrayList<Namespace>();
         for (String ns : prefixToNamespaceMapping.keySet()) {
-            List<Element> nodes = element.getChildren(query, Namespace.getNamespace(ns));
+            namespaces.add(Namespace.getNamespace(prefixToNamespaceMapping.get(ns), ns));
-            for (Element el : nodes) {
+        }
-                if (StringUtils.isNotBlank(el.getValue())) {
+        XPathExpression<Object> xpath = XPathFactory.instance().compile(query, Filters.fpassthrough(), null,namespaces);
+        List<Object> nodes = xpath.evaluate(t);
+        for (Object el : nodes) {
+            if (el instanceof Element) {
+                Element element = (Element) el;
+                if (StringUtils.isNotBlank(element.getText())) {
                     text.append(element.getText());
                 }
+            } else {
+                log.warn("node of type: " + el.getClass());
             }
         }
         if (StringUtils.isNotBlank(text.toString())) {

dspace-api/src/main/java/org/dspace/importer/external/metadatamapping/contributor/SimpleMultiplePathContributor.java

@@ -7,13 +7,19 @@
  */
 package org.dspace.importer.external.metadatamapping.contributor;
 
+import java.util.ArrayList;
 import java.util.Collection;
 import java.util.LinkedList;
 import java.util.List;
 
+import org.apache.logging.log4j.LogManager;
+import org.apache.logging.log4j.Logger;
 import org.dspace.importer.external.metadatamapping.MetadatumDTO;
 import org.jdom2.Element;
 import org.jdom2.Namespace;
+import org.jdom2.filter.Filters;
+import org.jdom2.xpath.XPathExpression;
+import org.jdom2.xpath.XPathFactory;
 
 /**
  * This contributor can perform research on multi-paths
@@ -22,6 +28,8 @@ import org.jdom2.Namespace;
  */
 public class SimpleMultiplePathContributor extends SimpleXpathMetadatumContributor {
 
+    private final static Logger log = LogManager.getLogger();
+
     private List<String> paths;
 
     public SimpleMultiplePathContributor() {}
@@ -31,13 +39,21 @@ public class SimpleMultiplePathContributor extends SimpleXpathMetadatumContribut
     }
 
     @Override
-    public Collection<MetadatumDTO> contributeMetadata(Element element) {
+    public Collection<MetadatumDTO> contributeMetadata(Element t) {
         List<MetadatumDTO> values = new LinkedList<>();
         for (String path : this.paths) {
+            List<Namespace> namespaces = new ArrayList<Namespace>();
             for (String ns : prefixToNamespaceMapping.keySet()) {
-                List<Element> nodes = element.getChildren(path, Namespace.getNamespace(ns));
+                namespaces.add(Namespace.getNamespace(prefixToNamespaceMapping.get(ns), ns));
-                for (Element el : nodes) {
+            }
-                    values.add(metadataFieldMapping.toDCValue(field, el.getValue()));
+            XPathExpression<Object> xpath = XPathFactory.instance().compile(path, Filters.fpassthrough(), null,
+                    namespaces);
+            List<Object> nodes = xpath.evaluate(t);
+            for (Object el : nodes) {
+                if (el instanceof Element) {
+                    values.add(metadataFieldMapping.toDCValue(field, ((Element) el).getText()));
+                } else {
+                    log.warn("node of type: " + el.getClass());
                 }
             }
         }

dspace-api/src/main/java/org/dspace/importer/external/metadatamapping/contributor/SimpleXpathMetadatumAndAttributeContributor.java

@@ -7,15 +7,21 @@
  */
 package org.dspace.importer.external.metadatamapping.contributor;
 
+import java.util.ArrayList;
 import java.util.Collection;
 import java.util.LinkedList;
 import java.util.List;
 
 import org.apache.commons.lang3.StringUtils;
+import org.apache.logging.log4j.LogManager;
+import org.apache.logging.log4j.Logger;
 import org.dspace.core.Constants;
 import org.dspace.importer.external.metadatamapping.MetadatumDTO;
 import org.jdom2.Element;
 import org.jdom2.Namespace;
+import org.jdom2.filter.Filters;
+import org.jdom2.xpath.XPathExpression;
+import org.jdom2.xpath.XPathFactory;
 
 /**
  * This contributor checks for each node returned for the supplied path
@@ -26,20 +32,32 @@ import org.jdom2.Namespace;
  */
 public class SimpleXpathMetadatumAndAttributeContributor extends SimpleXpathMetadatumContributor {
 
+    private final static Logger log = LogManager.getLogger();
+
     private String attribute;
 
     @Override
-    public Collection<MetadatumDTO> contributeMetadata(Element element) {
+    public Collection<MetadatumDTO> contributeMetadata(Element t) {
         List<MetadatumDTO> values = new LinkedList<>();
+        List<Namespace> namespaces = new ArrayList<Namespace>();
         for (String ns : prefixToNamespaceMapping.keySet()) {
-            List<Element> nodes = element.getChildren(query, Namespace.getNamespace(ns));
+            namespaces.add(Namespace.getNamespace(prefixToNamespaceMapping.get(ns), ns));
-            for (Element el : nodes) {
+        }
-                String attributeValue = el.getAttributeValue(this.attribute);
+        XPathExpression<Object> xpath = XPathFactory.instance().compile(query, Filters.fpassthrough(), null,
+                namespaces);
+        List<Object> nodes = xpath.evaluate(t);
+        for (Object el : nodes) {
+            if (el instanceof Element) {
+                Element element = (Element) el;
+                String attributeValue = element.getAttributeValue(this.attribute);
                 if (StringUtils.isNotBlank(attributeValue)) {
                     values.add(metadataFieldMapping.toDCValue(this.field, attributeValue));
                 } else {
-                    values.add(metadataFieldMapping.toDCValue(this.field, Constants.PLACEHOLDER_PARENT_METADATA_VALUE));
+                    values.add(metadataFieldMapping.toDCValue(this.field,
+                            Constants.PLACEHOLDER_PARENT_METADATA_VALUE));
                 }
+            } else {
+                log.warn("node of type: " + el.getClass());
             }
         }
         return values;

dspace-api/src/main/java/org/dspace/importer/external/metadatamapping/contributor/WosAttribute2ValueContributor.java

@@ -7,6 +7,7 @@
  */
 package org.dspace.importer.external.metadatamapping.contributor;
 
+import java.util.ArrayList;
 import java.util.Collection;
 import java.util.LinkedList;
 import java.util.List;
@@ -14,14 +15,16 @@ import java.util.Map;
 import javax.annotation.Resource;
 
 import org.apache.commons.lang3.StringUtils;
+import org.apache.logging.log4j.LogManager;
+import org.apache.logging.log4j.Logger;
 import org.dspace.importer.external.metadatamapping.MetadataFieldConfig;
 import org.dspace.importer.external.metadatamapping.MetadataFieldMapping;
 import org.dspace.importer.external.metadatamapping.MetadatumDTO;
-import org.jaxen.JaxenException;
 import org.jdom2.Element;
 import org.jdom2.Namespace;
-import org.slf4j.Logger;
+import org.jdom2.filter.Filters;
-import org.slf4j.LoggerFactory;
+import org.jdom2.xpath.XPathExpression;
+import org.jdom2.xpath.XPathFactory;
 
 /**
  * This contributor checks for each node returned for the given path if the node contains "this.attribute"
@@ -33,7 +36,7 @@ import org.slf4j.LoggerFactory;
  */
 public class WosAttribute2ValueContributor implements MetadataContributor<Element> {
 
-    private static final Logger log = LoggerFactory.getLogger(WosAttribute2ValueContributor.class);
+    private final static Logger log = LogManager.getLogger();
 
     private String query;
 
@@ -60,24 +63,28 @@ public class WosAttribute2ValueContributor implements MetadataContributor<Elemen
     }
 
     @Override
-    public Collection<MetadatumDTO> contributeMetadata(Element element) {
+    public Collection<MetadatumDTO> contributeMetadata(Element t) {
         List<MetadatumDTO> values = new LinkedList<>();
-        try {
+        List<Namespace> namespaces = new ArrayList<Namespace>();
-            for (String ns : prefixToNamespaceMapping.keySet()) {
+        for (String ns : prefixToNamespaceMapping.keySet()) {
-                List<Element> nodes = element.getChildren(query, Namespace.getNamespace(ns));
+            namespaces.add(Namespace.getNamespace(prefixToNamespaceMapping.get(ns), ns));
-                for (Element el : nodes) {
-                    String attributeValue = el.getAttributeValue(this.attribute);
-                    setField(attributeValue, element, values);
-                }
-            }
-            return values;
-        } catch (JaxenException e) {
-            log.warn(query, e);
-            throw new RuntimeException(e);
         }
+        XPathExpression<Object> xpath = XPathFactory.instance().compile(query, Filters.fpassthrough(), null,
+                namespaces);
+        List<Object> nodes = xpath.evaluate(t);
+        for (Object el : nodes) {
+            if (el instanceof Element) {
+                Element element = (Element) el;
+                String attributeValue = element.getAttributeValue(this.attribute);
+                setField(attributeValue, element, values);
+            } else {
+                log.warn("node of type: " + el.getClass());
+            }
+        }
+        return values;
     }
 
-    private void setField(String attributeValue, Element el, List<MetadatumDTO> values) throws JaxenException {
+    private void setField(String attributeValue, Element el, List<MetadatumDTO> values) {
         for (String id : attributeValue2metadata.keySet()) {
             if (StringUtils.equals(id, attributeValue)) {
                 if (this.firstChild) {

dspace-api/src/main/java/org/dspace/importer/external/metadatamapping/contributor/WosIdentifierRidContributor.java

@@ -7,19 +7,22 @@
  */
 package org.dspace.importer.external.metadatamapping.contributor;
 
+import java.util.ArrayList;
 import java.util.Collection;
 import java.util.LinkedList;
 import java.util.List;
 import java.util.Objects;
 
 import org.apache.commons.lang3.StringUtils;
+import org.apache.logging.log4j.LogManager;
+import org.apache.logging.log4j.Logger;
 import org.dspace.core.Constants;
 import org.dspace.importer.external.metadatamapping.MetadatumDTO;
-import org.jaxen.JaxenException;
 import org.jdom2.Element;
 import org.jdom2.Namespace;
-import org.slf4j.Logger;
+import org.jdom2.filter.Filters;
-import org.slf4j.LoggerFactory;
+import org.jdom2.xpath.XPathExpression;
+import org.jdom2.xpath.XPathFactory;
 
 /**
  * 
@@ -27,30 +30,33 @@ import org.slf4j.LoggerFactory;
  */
 public class WosIdentifierRidContributor extends SimpleXpathMetadatumContributor {
 
-    private static final Logger log = LoggerFactory.getLogger(WosIdentifierRidContributor.class);
+    private final static Logger log = LogManager.getLogger();
 
     @Override
-    public Collection<MetadatumDTO> contributeMetadata(Element element) {
+    public Collection<MetadatumDTO> contributeMetadata(Element t) {
         List<MetadatumDTO> values = new LinkedList<>();
-        try {
+        List<Namespace> namespaces = new ArrayList<Namespace>();
-            for (String ns : prefixToNamespaceMapping.keySet()) {
+        for (String ns : prefixToNamespaceMapping.keySet()) {
-                List<Element> nodes = element.getChildren(query, Namespace.getNamespace(ns));
+            namespaces.add(Namespace.getNamespace(prefixToNamespaceMapping.get(ns), ns));
-                for (Element el : nodes) {
-                    // Element element2 = el.getFirstChildWithName("name");
-                    if (Objects.nonNull(element)) {
-                        String type = element.getAttributeValue("role");
-                        setIdentyfire(type, element, values);
-                    }
-                }
-            }
-            return values;
-        } catch (JaxenException e) {
-            log.error(query, e);
-            throw new RuntimeException(e);
         }
+        XPathExpression<Object> xpath = XPathFactory.instance().compile(query, Filters.fpassthrough(), null,
+                namespaces);
+        List<Object> nodes = xpath.evaluate(t);
+        for (Object el : nodes) {
+            if (el instanceof Element) {
+                Element element = ((Element) el).getChild("name");
+                if (Objects.nonNull(element)) {
+                    String type = element.getAttributeValue("role");
+                    setIdentyfire(type, element, values);
+                }
+            } else {
+                log.warn("node of type: " + el.getClass());
+            }
+        }
+        return values;
     }
 
-    private void setIdentyfire(String type, Element el, List<MetadatumDTO> values) throws JaxenException {
+    private void setIdentyfire(String type, Element el, List<MetadatumDTO> values) {
         if (StringUtils.equals("researcher_id", type)) {
             String value = el.getAttributeValue("r_id");
             if (StringUtils.isNotBlank(value)) {

dspace-server-webapp/src/test/java/org/dspace/app/rest/AbstractLiveImportIntegrationTest.java

@@ -32,8 +32,8 @@ public class AbstractLiveImportIntegrationTest extends AbstractControllerIntegra
         ImportRecord  secondImported = recordsImported.iterator().next();
         ImportRecord  first2match = recordsImported.iterator().next();
         ImportRecord  second2match = recordsImported.iterator().next();
-        boolean checkFirstRecord = firstImported.getValueList().containsAll(first2match.getValueList());
+        boolean checkFirstRecord = first2match.getValueList().containsAll(firstImported.getValueList());
-        boolean checkSecondRecord = secondImported.getValueList().containsAll(second2match.getValueList());
+        boolean checkSecondRecord = second2match.getValueList().containsAll(secondImported.getValueList());
         return checkFirstRecord && checkSecondRecord;
     }