A simple implementation of a feature where an OAI client can be instr…

…ucted to return harvested metadata records unparsed #284
gdcc · Feb 6, 2025 · 8b064ba · 8b064ba
1 parent 9e22f93
commit 8b064ba
Show file tree

Hide file tree

Showing 4 changed files with 55 additions and 10 deletions.
diff --git a/xoai-common/src/main/java/io/gdcc/xoai/model/oaipmh/results/record/Metadata.java b/xoai-common/src/main/java/io/gdcc/xoai/model/oaipmh/results/record/Metadata.java
@@ -77,6 +77,11 @@ public XOAIMetadata getXoaiMetadata() {
         if (element instanceof XOAIMetadata) return (XOAIMetadata) element;
         else return null;
     }
+
+    public String getMetadataAsString() {
+        if (element instanceof EchoElement) return ((EchoElement)element).asUnparsedString();
+        return null;
+    }
 
     /**
      * This is here for Dataverse 4/5 backward compatibility.

diff --git a/xoai-common/src/main/java/io/gdcc/xoai/xml/EchoElement.java b/xoai-common/src/main/java/io/gdcc/xoai/xml/EchoElement.java
@@ -52,6 +52,13 @@ public void write(final XmlWriter writer) throws XmlWriteException {
                     "Cannot write XML when none given (both stream and string null)");
         }
     }
+
+    public String asUnparsedString() {
+        if (xmlString != null) {
+            return xmlString;
+        }
+        return null; 
+    }
 
     private void write(final XmlWriter writer, final InputStream inStream)
             throws XmlWriteException {

diff --git a/xoai-service-provider/src/main/java/io/gdcc/xoai/serviceprovider/model/Context.java b/xoai-service-provider/src/main/java/io/gdcc/xoai/serviceprovider/model/Context.java
@@ -24,6 +24,7 @@ public class Context {
     private final Map<String, Transformer> metadataTransformers = new HashMap<>();
     private String baseUrl;
     private Granularity granularity;
+    private boolean saveUnparsedMetadata = false;
     private OAIClient client;
 
     public Context() {
@@ -90,6 +91,29 @@ public Context withOAIClient(OAIClient client) {
     public OAIClient getClient() {
         return client;
     }
+
+    /**
+     * Should this harvester skip parsing the "metadata" sections of oai records 
+     * in the bodies of GetRecord and ListRecords responses, and cache and make 
+     * them available as unparsed Strings instead.
+     *
+     * @return boolean
+     */
+    public boolean isSaveUnparsedMetadata() {
+        return this.saveUnparsedMetadata;
+    }
+
+    /**
+     * Instruct this harvester not to attempt to parse the "metadata" sections of
+     * oai records in the bodies of GetRecord and ListRecords responses, but
+     * cache and make them available as Strings instead.
+     *
+     * @return A Context
+     */
+    public Context withSaveUnparsedMetadata() {
+        this.saveUnparsedMetadata = true;
+        return this;
+    }
 
     public enum KnownTransformer {
         OAI_DC("to_xoai/oai_dc.xsl");

diff --git a/xoai-service-provider/src/main/java/io/gdcc/xoai/serviceprovider/parsers/RecordParser.java b/xoai-service-provider/src/main/java/io/gdcc/xoai/serviceprovider/parsers/RecordParser.java
@@ -47,18 +47,27 @@ public Record parse(XmlReader reader) throws XmlReaderException {
         if (!record.getHeader().isDeleted()) {
             reader.next(elementName(localPart(equalTo("metadata")))).next(aStartElement());
             String content = reader.retrieveCurrentAsString();
-            ByteArrayInputStream inputStream =
-                    new ByteArrayInputStream(content.getBytes(StandardCharsets.UTF_8));
-            XSLPipeline pipeline =
-                    new XSLPipeline(inputStream, true)
-                            .apply(context.getMetadataTransformer(metadataPrefix));
+            System.out.println("Metadata content: "+content);
+
+            if (this.context.isSaveUnparsedMetadata()) {
+                record.withMetadata(new Metadata(content));
+            } else {
+                ByteArrayInputStream inputStream
+                        = new ByteArrayInputStream(content.getBytes(StandardCharsets.UTF_8));
 
-            if (context.hasTransformer()) pipeline.apply(context.getTransformer());
+                XSLPipeline pipeline
+                        = new XSLPipeline(inputStream, true)
+                                .apply(context.getMetadataTransformer(metadataPrefix));
 
-            try {
-                record.withMetadata(new Metadata(new MetadataParser().parse(pipeline.process())));
-            } catch (TransformerException e) {
-                throw new InternalHarvestException("Unable to process transformer", e);
+                if (context.hasTransformer()) {
+                    pipeline.apply(context.getTransformer());
+                }
+
+                try {
+                    record.withMetadata(new Metadata(new MetadataParser().parse(pipeline.process())));
+                } catch (TransformerException e) {
+                    throw new InternalHarvestException("Unable to process transformer", e);
+                }
             }
         }