Prefix DC identifiers with type if no scheme part is present

This makes a PPN identifier `ppn:12345678` in <dc:identifier> elements.
qucosa · Sep 18, 2015 · f306c1f · f306c1f
1 parent 6e22d5a
commit f306c1f
Show file tree

Hide file tree

Showing 6 changed files with 23 additions and 16 deletions.
diff --git a/src/main/java/org/purl/sword/server/fedora/fileHandlers/METSContainer.java b/src/main/java/org/purl/sword/server/fedora/fileHandlers/METSContainer.java
@@ -33,21 +33,22 @@
 import java.security.NoSuchAlgorithmException;
 import java.util.LinkedList;
 import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 import static org.purl.sword.server.fedora.fedoraObjects.State.INACTIVE;
 
 public class METSContainer {
 
+    public static final Pattern PATTERN = Pattern.compile("^[a-z][a-z0-9\\+\\.\\-]*\\:.*", Pattern.CASE_INSENSITIVE);
+
     private static final String DS_ID_SLUBINFO = "SLUB-INFO";
     private static final String DS_ID_SLUBINFO_LABEL = "SLUB Administrative Metadata";
-
     private static final String DS_ID_QUCOSAXML = "QUCOSA-XML";
     private static final String DS_ID_QUCOSAXML_LABEL = "Pristine Qucosa XML Metadata";
-
     private static final String DS_MODS_MIME_TYPE = "application/mods+xml";
     private static final String DS_ID_MODS = "MODS";
     private static final String DS_ID_MODS_LABEL = "Object Bibliographic Metadata";
-
     private static final String METS_DMDSEC_PREFIX = "/mets:mets/mets:dmdSec";
     private static final String MODS_PREFIX = METS_DMDSEC_PREFIX + "/mets:mdWrap[@MDTYPE='MODS']/mets:xmlData/mods:mods";
 
@@ -196,12 +197,24 @@ private String getPrimaryTitle() {
 
     private List<String> getIdentifiers() {
         try {
-            return XPATH_IDENTIFIERS.selectValues(metsDocument);
+            final List<String> identifiers = new LinkedList<>();
+            final List<Element> elements = XPATH_IDENTIFIERS.selectNodes(metsDocument);
+            for (Element e : elements) {
+                final String type = e.getAttributeValue("type");
+                final String id = e.getTextTrim();
+                identifiers.add(hasProtocol(id) ? id : type + ":" + id);
+            }
+            return identifiers;
         } catch (JDOMException e) {
             return null;
         }
     }
 
+    private boolean hasProtocol(String id) {
+        Matcher matcher = PATTERN.matcher(id);
+        return matcher.matches();
+    }
+
     private Datastream getDatastream(XPathQuery query, String datastreamID, String datastreamLabel) throws SWORDException {
         return getDatastream(query, datastreamID, datastreamLabel, null);
     }

diff --git a/src/main/java/org/purl/sword/server/fedora/fileHandlers/XPathQuery.java b/src/main/java/org/purl/sword/server/fedora/fileHandlers/XPathQuery.java
@@ -44,13 +44,6 @@ public String selectValue(Document doc) throws JDOMException {
         }
     }
 
-    public List<String> selectValues(Document doc) throws JDOMException {
-        final List<Element> els = selectNodes(doc);
-        return new LinkedList<String>() {{
-            for (Element e : els) add(e.getTextTrim());
-        }};
-    }
-
     public Element selectNode(Document doc) throws JDOMException {
         return (Element) xpath.selectSingleNode(doc);
     }

diff --git a/...test/java/org/purl/sword/server/fedora/fileHandlers/QucosaMETSFileHandler_IngestTest.java b/...test/java/org/purl/sword/server/fedora/fileHandlers/QucosaMETSFileHandler_IngestTest.java
@@ -78,8 +78,9 @@ public void dcDatastreamHasIdentifiers() throws Exception {
 
         verify(mockFedoraRepository).ingest(argument.capture());
         DublinCore result = argument.getValue().getDc();
-        assertTrue("Should have identifier", result.getIdentifier().contains("urn:nbn:de:bsz:14-qucosa-32992"));
-        assertTrue("Should have identifier", result.getIdentifier().contains("322202922"));
+
+        assertTrue("Should have URN identifier", result.getIdentifier().contains("urn:nbn:de:bsz:14-qucosa-32992"));
+        assertTrue("Should have PPN identifier", result.getIdentifier().contains("ppn:322202922"));
     }
 
     @Test

diff --git a/src/test/resources/mets_all_references.xml b/src/test/resources/mets_all_references.xml
@@ -30,7 +30,7 @@
                         <mods:subTitle>Das EFRE-Projekt Sächsischer Dokumentenserver</mods:subTitle>
                     </mods:titleInfo>
                     <mods:identifier type="urn">urn:nbn:de:bsz:14-qucosa-32992</mods:identifier>
-                    <mods:identifier type="swb-ppn">322202922</mods:identifier>
+                    <mods:identifier type="ppn">322202922</mods:identifier>
 
                     <mods:relatedItem
                             type="series"

diff --git a/src/test/resources/mets_invalid_file.xml b/src/test/resources/mets_invalid_file.xml
@@ -77,7 +77,7 @@
                     <mods:accessCondition>Dieser Beitrag ist mit Zustimmung des Rechteinhabers frei zugänglich.
                     </mods:accessCondition>
                     <mods:identifier type="urn">urn:nbn:de:bsz:14-qucosa-32992</mods:identifier>
-                    <mods:identifier type="swb-ppn">322202922</mods:identifier>
+                    <mods:identifier type="ppn">322202922</mods:identifier>
                     <mods:relatedItem type="series">
                         <mods:titleInfo>
                             <mods:title>BIS - Das Magazin der Bibliotheken in Sachsen</mods:title>

diff --git a/src/test/resources/mets_ok.xml b/src/test/resources/mets_ok.xml
@@ -61,7 +61,7 @@
                     <mods:accessCondition>Dieser Beitrag ist mit Zustimmung des Rechteinhabers frei zugänglich.
                     </mods:accessCondition>
                     <mods:identifier type="urn">urn:nbn:de:bsz:14-qucosa-32992</mods:identifier>
-                    <mods:identifier type="swb-ppn">322202922</mods:identifier>
+                    <mods:identifier type="ppn">322202922</mods:identifier>
                     <mods:relatedItem type="constituent"
                                       xlink:href="http://nbn-resolving.de/urn:nbn:de:bsz:14-qucosa-32825">
                         <mods:identifier type="urn">urn:nbn:de:bsz:14-qucosa-32825</mods:identifier>