Add support for XML stream reader

src/main/java/org/entur/netex/NetexParser.java

@@ -6,6 +6,7 @@
 import java.util.Enumeration;
 import java.util.zip.ZipEntry;
 import java.util.zip.ZipFile;
+import javax.xml.stream.XMLStreamReader;
 import org.entur.netex.index.api.NetexEntitiesIndex;
 import org.entur.netex.index.impl.NetexEntitiesIndexImpl;
 import org.entur.netex.loader.NetexXmlParser;
@@ -20,6 +21,22 @@ public class NetexParser {
 
   private final NetexXmlParser xmlParser = new NetexXmlParser();
 
+  private final XmlStreamReaderFactory xmlStreamReaderFactory;
+
+  /**
+   * Create a default NetexParser. The default implementation relies on a SAX parser.
+   */
+  public NetexParser() {
+    xmlStreamReaderFactory = null;
+  }
+
+  /**
+   * Create a NeTexParser that parses a NeTEx document through an XML stream reader (StAX).
+   */
+  public NetexParser(XmlStreamReaderFactory xmlStreamReaderFactory) {
+    this.xmlStreamReaderFactory = xmlStreamReaderFactory;
+  }
+
   /**
    * Parse a NeTEx publication delivery from one or more files in
    * a zip archive
@@ -87,7 +104,15 @@ public NetexEntitiesIndex parse(
 
   private void load(NetexEntitiesIndex index, InputStream inputStream) {
     try {
-      PublicationDeliveryStructure doc = xmlParser.parseXmlDoc(inputStream);
+      PublicationDeliveryStructure doc;
+      if (xmlStreamReaderFactory != null) {
+        XMLStreamReader xmlReader =
+          xmlStreamReaderFactory.createXmlStreamReader(inputStream);
+        doc = xmlParser.parseXmlDoc(xmlReader);
+      } else {
+        doc = xmlParser.parseXmlDoc(inputStream);
+      }
+
       NetexDocumentParser.parseAndPopulateIndex(index, doc);
     } catch (JAXBException e) {
       throw new RuntimeException(e.getMessage(), e);

src/main/java/org/entur/netex/XmlStreamReaderFactory.java

@@ -0,0 +1,11 @@
+package org.entur.netex;
+
+import java.io.InputStream;
+import javax.xml.stream.XMLStreamReader;
+
+/**
+ * Factory for XMLStreamReader. Concrete implementations can be used to ignore elements within the XML document
+ */
+public interface XmlStreamReaderFactory {
+  XMLStreamReader createXmlStreamReader(InputStream inputStream);
+}

src/main/java/org/entur/netex/loader/NetexXmlParser.java

@@ -5,6 +5,7 @@
 import jakarta.xml.bind.JAXBException;
 import jakarta.xml.bind.Unmarshaller;
 import java.io.InputStream;
+import javax.xml.stream.XMLStreamReader;
 import org.rutebanken.netex.model.PublicationDeliveryStructure;
 
 /** Simple wrapper to perform typesafe xml parsing and simple error handling. */
@@ -31,6 +32,20 @@ public PublicationDeliveryStructure parseXmlDoc(InputStream stream)
     return root.getValue();
   }
 
+  /**
+   * Parse an input stream and return the root document type for the given xml file (stream).
+   */
+  public PublicationDeliveryStructure parseXmlDoc(XMLStreamReader reader)
+    throws JAXBException {
+    @SuppressWarnings("unchecked")
+    JAXBElement<PublicationDeliveryStructure> root =
+      (JAXBElement<PublicationDeliveryStructure>) unmarshaller.unmarshal(
+        reader
+      );
+
+    return root.getValue();
+  }
+
   /** factory method for unmarshaller */
   private static Unmarshaller createUnmarshaller() {
     try {