Merge pull request #4578 from evolvedbinary/feature/preserve-xml-declaration

Preserve XML Declaration

Author: reinhapa

exist-core/pom.xml

@@ -271,7 +271,7 @@
             <groupId>org.exist-db.thirdparty.xerces</groupId>
             <artifactId>xercesImpl</artifactId>
             <version>2.12.2</version>
-            <classifier>xml-schema-1.1</classifier>
+            <classifier>jdk14-xml-schema-1.1</classifier>
             <exclusions>
                 <exclusion> <!-- conflicts with Java 17's javax.xml module -->
                     <groupId>xml-apis</groupId>
@@ -679,6 +679,7 @@
                                 <!--
                                     Already under LGPL 2.1, but with a different Copyright
                                 -->
+                                <exclude>src/main/java/org/exist/dom/persistent/XMLDeclarationImpl.java</exclude>
                                 <exclude>src/main/java/org/exist/resolver/ResolverFactory.java</exclude>
                                 <exclude>src/main/java/org/exist/resolver/XercesXmlResolverAdapter.java</exclude>
                                 <exclude>src/main/java/org/exist/util/UTF8.java</exclude>
@@ -831,6 +832,7 @@ The original license statement is also included below.]]></preamble>
                             -->
                             <header>${project.parent.relativePath}/FDB-backport-LGPL-21-ONLY-license.template.txt</header>
                             <includes>
+                                <include>src/main/java/org/exist/dom/persistent/XMLDeclarationImpl.java</include>
                                 <include>src/main/java/org/exist/resolver/ResolverFactory.java</include>
                                 <include>src/main/java/org/exist/resolver/XercesXmlResolverAdapter.java</include>
                                 <include>src/test/java/org/exist/storage/MoveCollectionTest.java</include>

exist-core/src/main/java/org/exist/Indexer.java

@@ -26,6 +26,7 @@
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
 import org.exist.collections.CollectionConfiguration;
+import org.exist.dom.QName;
 import org.exist.dom.persistent.AttrImpl;
 import org.exist.dom.persistent.CDATASectionImpl;
 import org.exist.dom.persistent.CommentImpl;
@@ -34,9 +35,9 @@
 import org.exist.dom.persistent.ElementImpl;
 import org.exist.dom.persistent.NodeHandle;
 import org.exist.dom.persistent.ProcessingInstructionImpl;
-import org.exist.dom.QName;
 import org.exist.dom.persistent.StoredNode;
 import org.exist.dom.persistent.TextImpl;
+import org.exist.dom.persistent.XMLDeclarationImpl;
 import org.exist.indexing.StreamListener;
 import org.exist.indexing.StreamListener.ReindexMode;
 import org.exist.storage.DBBroker;
@@ -62,6 +63,8 @@
 import org.xml.sax.SAXParseException;
 import org.xml.sax.ext.LexicalHandler;
 
+import javax.annotation.Nullable;
+
 /**
  * Parses a given input document via SAX, stores it to the database and handles
  * index-creation.
@@ -542,6 +545,12 @@ public void startDocument() {
         nodeFactoryInstanceCnt = 1;
     }
 
+    @Override
+    public void declaration(@Nullable final String version, @Nullable final String encoding, @Nullable final String standalone) throws SAXException {
+        final XMLDeclarationImpl xmlDecl = new XMLDeclarationImpl(version, encoding, standalone);
+        document.setXmlDeclaration(xmlDecl);
+    }
+
     final boolean hasNormAttribute(final Attributes attributes) {
         for (int i = 0; i < attributes.getLength(); i++) {
             if(attributes.getLocalName(i).equals("norm")) {

exist-core/src/main/java/org/exist/client/InteractiveClient.java

@@ -89,6 +89,8 @@
 import se.softhouse.jargo.ArgumentException;
 
 import static java.nio.charset.StandardCharsets.UTF_8;
+import static javax.xml.transform.OutputKeys.OMIT_XML_DECLARATION;
+import static org.exist.storage.serializers.EXistOutputKeys.OMIT_ORIGINAL_XML_DECLARATION;
 import static org.exist.storage.serializers.EXistOutputKeys.OUTPUT_DOCTYPE;
 import static org.xmldb.api.base.ResourceType.XML_RESOURCE;
 
@@ -143,6 +145,8 @@ public class InteractiveClient {
         DEFAULT_PROPERTIES.setProperty(USER, USER_DEFAULT);
         DEFAULT_PROPERTIES.setProperty(EDITOR, EDIT_CMD);
         DEFAULT_PROPERTIES.setProperty(INDENT, "true");
+        DEFAULT_PROPERTIES.setProperty(OMIT_XML_DECLARATION, "no");
+        DEFAULT_PROPERTIES.setProperty(OMIT_ORIGINAL_XML_DECLARATION, "no");
         DEFAULT_PROPERTIES.setProperty(OUTPUT_DOCTYPE, "true");
         DEFAULT_PROPERTIES.setProperty(ENCODING, ENCODING_DEFAULT.name());
         DEFAULT_PROPERTIES.setProperty(COLORS, "false");

exist-core/src/main/java/org/exist/collections/triggers/DocumentTriggers.java

@@ -40,6 +40,8 @@
 import org.xml.sax.SAXParseException;
 import org.xml.sax.ext.LexicalHandler;
 
+import javax.annotation.Nullable;
+
 /**
  * @author <a href="mailto:[email protected]">Dmitriy Shabanov</a>
  *
@@ -144,6 +146,11 @@ public void startDocument() throws SAXException {
         contentHandler.startDocument();
     }
 
+    @Override
+    public void declaration(@Nullable final String version, @Nullable final String encoding, @Nullable final String standalone) throws SAXException {
+        contentHandler.declaration(version, encoding, standalone);
+    }
+
     @Override
     public void endDocument() throws SAXException {
         contentHandler.endDocument();

exist-core/src/main/java/org/exist/collections/triggers/SAXTrigger.java

@@ -36,6 +36,8 @@
 import org.xml.sax.SAXParseException;
 import org.xml.sax.ext.LexicalHandler;
 
+import javax.annotation.Nullable;
+
 /**
  * Abstract default implementation of a Trigger. This implementation just
  * forwards all SAX events to the output content handler.
@@ -89,6 +91,13 @@ public void startDocument() throws SAXException {
             nextContentHandler.startDocument();
     }
 
+    @Override
+    public void declaration(@Nullable final String version, @Nullable final String encoding, @Nullable final String standalone) throws SAXException {
+        if (nextContentHandler != null) {
+            nextContentHandler.declaration(version, encoding, standalone);
+        }
+    }
+
     @Override
     public void endDocument() throws SAXException {
         if (nextContentHandler != null)

exist-core/src/main/java/org/exist/dom/memtree/DocumentBuilderReceiver.java

@@ -122,6 +122,11 @@ public void endDocument() throws SAXException {
         builder.endDocument();
     }
 
+    @Override
+    public void declaration(final String version, final String encoding, final String standalone) throws SAXException {
+        // NOTE(AR) in-memory documents do not support XML Declaration
+    }
+
     @Override
     public void startPrefixMapping(final String prefix, final String namespaceURI) throws SAXException {
         if(prefix == null || prefix.length() == 0) {

exist-core/src/main/java/org/exist/dom/persistent/DocumentImpl.java

@@ -77,6 +77,9 @@ public class DocumentImpl extends NodeImpl<DocumentImpl> implements Resource, Do
 
     //public static final byte DOCUMENT_NODE_SIGNATURE = 0x0F;
 
+    public static final byte NO_XMLDECL = 0;
+    public static final byte HAS_XMLDECL = 1;
+
     public static final byte NO_DOCTYPE = 0;
     public static final byte HAS_DOCTYPE = 1;
 
@@ -140,6 +143,11 @@ public class DocumentImpl extends NodeImpl<DocumentImpl> implements Resource, Do
      */
     protected int userLock = 0;
 
+    /**
+     * The document's XML Declaration - if specified.
+     */
+    @Nullable private XMLDeclarationImpl xmlDecl = null;
+
     /**
      * The document's doctype declaration - if specified.
      */
@@ -203,7 +211,7 @@ public DocumentImpl(final Expression expression, final BrokerPool pool, @Nullabl
             final XmldbURI fileURI) {
         this(expression, pool, collection, docId, fileURI,
                 PermissionFactory.getDefaultResourcePermission(pool.getSecurityManager()), 0, null,
-                System.currentTimeMillis(), null, null, null);
+                System.currentTimeMillis(), null, null, null, null);
     }
 
     /**
@@ -225,7 +233,7 @@ public DocumentImpl(final int docId, final DocumentImpl prevDoc) {
      */
     public DocumentImpl(final Expression expression, final int docId, final DocumentImpl prevDoc) {
         this(expression, prevDoc.pool, prevDoc.collection, docId, prevDoc.fileURI, prevDoc.permissions.copy(), 0, null,
-                System.currentTimeMillis(), null, null, null);
+                System.currentTimeMillis(), null, null, null, null);
     }
 
     /**
@@ -242,13 +250,40 @@ public DocumentImpl(final Expression expression, final int docId, final Document
      * @param lastModified the last modified time of the document, or null to use the {@code created} time
      * @param mimeType the media type of the document, or null for application/xml
      * @param docType the document type, or null
+     *
+     * @deprecated Use {@link DocumentImpl(BrokerPool, Collection, int, XmldbURI, Permission, int, long[], long, Long, String, XMLDeclarationImpl, DocumentType)}
+     */
+    @Deprecated
+    public DocumentImpl(final BrokerPool pool, @Nullable final Collection collection,
+                        final int docId, final XmldbURI fileURI, final Permission permissions,
+                        final int children, @Nullable final long[] childAddress,
+                        final long created, @Nullable final Long lastModified, @Nullable final String mimeType,
+                        @Nullable final DocumentType docType) {
+        this(null, pool, collection, docId, fileURI, permissions, children, childAddress, created, lastModified, mimeType, null, docType);
+    }
+
+    /**
+     * Creates a new persistent Document instance to replace an existing document instance.
+     *
+     * @param pool The broker pool
+     * @param collection The Collection which holds this document
+     * @param docId the id of the document
+     * @param fileURI The name of the document
+     * @param permissions the permissions of the document
+     * @param children the number of children that the document has
+     * @param childAddress the addresses of the child nodes
+     * @param created the created time of the document
+     * @param lastModified the last modified time of the document, or null to use the {@code created} time
+     * @param mimeType the media type of the document, or null for application/xml
+     * @param xmlDecl the XML Declaration, or null
+     * @param docType the document type, or null
      */
     public DocumentImpl(final BrokerPool pool, @Nullable final Collection collection,
             final int docId, final XmldbURI fileURI, final Permission permissions,
             final int children, @Nullable final long[] childAddress,
             final long created, @Nullable final Long lastModified, @Nullable final String mimeType,
-            @Nullable final DocumentType docType) {
-        this(null, pool, collection, docId, fileURI, permissions, children, childAddress, created, lastModified, mimeType, docType);
+            @Nullable final XMLDeclarationImpl xmlDecl, @Nullable final DocumentType docType) {
+        this(null, pool, collection, docId, fileURI, permissions, children, childAddress, created, lastModified, mimeType, xmlDecl, docType);
     }
 
     /**
@@ -266,12 +301,40 @@ public DocumentImpl(final BrokerPool pool, @Nullable final Collection collection
      * @param lastModified the last modified time of the document, or null to use the {@code created} time
      * @param mimeType the media type of the document, or null for application/xml
      * @param docType the document type, or null
+     *
+     * @deprecated Use {@link DocumentImpl(Expression, BrokerPool, Collection, int ,XmldbURI, Permission, int, long[], long, Long, String, XMLDeclarationImpl, DocumentType)}
      */
+    @Deprecated
     public DocumentImpl(final Expression expression, final BrokerPool pool, @Nullable final Collection collection,
             final int docId, final XmldbURI fileURI, final Permission permissions,
             final int children, @Nullable final long[] childAddress,
             final long created, @Nullable final Long lastModified, @Nullable final String mimeType,
             @Nullable final DocumentType docType) {
+        this(expression, pool, collection, docId, fileURI, permissions, children, childAddress, created, lastModified, mimeType, null, docType);
+    }
+
+    /**
+     * Creates a new persistent Document instance to replace an existing document instance.
+     *
+     * @param expression the expression from which the document instance derives
+     * @param pool The broker pool
+     * @param collection The Collection which holds this document
+     * @param docId the id of the document
+     * @param fileURI The name of the document
+     * @param permissions the permissions of the document
+     * @param children the number of children that the document has
+     * @param childAddress the addresses of the child nodes
+     * @param created the created time of the document
+     * @param lastModified the last modified time of the document, or null to use the {@code created} time
+     * @param mimeType the media type of the document, or null for application/xml
+     * @param xmlDecl the XML Declaration, or null
+     * @param docType the document type, or null
+     */
+    public DocumentImpl(final Expression expression, final BrokerPool pool, @Nullable final Collection collection,
+            final int docId, final XmldbURI fileURI, final Permission permissions,
+            final int children, @Nullable final long[] childAddress,
+            final long created, @Nullable final Long lastModified, @Nullable final String mimeType,
+            @Nullable final XMLDeclarationImpl xmlDecl, @Nullable final DocumentType docType) {
         super(expression);
         this.pool = pool;
 
@@ -286,6 +349,7 @@ public DocumentImpl(final Expression expression, final BrokerPool pool, @Nullabl
         this.created = created;
         this.lastModified = lastModified == null ? created : lastModified;
         this.mimeType = mimeType == null ?  MimeType.XML_TYPE.getName() : mimeType;
+        this.xmlDecl = xmlDecl;
         this.docType = docType;
 
         //inherit the group to the resource if current collection is setGid
@@ -790,6 +854,12 @@ void writeDocumentAttributes(final SymbolTable symbolTable, final VariableByteOu
         ostream.writeInt(symbolTable.getMimeTypeId(mimeType));
         ostream.writeInt(pageCount);
         ostream.writeInt(userLock);
+        if (xmlDecl != null) {
+            ostream.writeByte(HAS_XMLDECL);
+            xmlDecl.write(ostream);
+        } else {
+            ostream.writeByte(NO_XMLDECL);
+        }
         if (docType != null) {
             ostream.writeByte(HAS_DOCTYPE);
             ((DocumentTypeImpl) docType).write(ostream);
@@ -833,6 +903,12 @@ public static DocumentImpl read(final BrokerPool pool, final VariableByteInput i
         final String mimeType = pool.getSymbols().getMimeType(mimeTypeSymbolsIndex);
         final int pageCount = istream.readInt();
         final int userLock = istream.readInt();
+        final XMLDeclarationImpl xmlDecl;
+        if (istream.readByte() == HAS_XMLDECL) {
+            xmlDecl = XMLDeclarationImpl.read(istream);
+        } else {
+            xmlDecl = null;
+        }
         final DocumentTypeImpl docType;
         if (istream.readByte() == HAS_DOCTYPE) {
             docType = DocumentTypeImpl.read(istream);
@@ -846,7 +922,7 @@ public static DocumentImpl read(final BrokerPool pool, final VariableByteInput i
             lockToken = null;
         }
 
-        final DocumentImpl doc = new DocumentImpl(null, pool, null, docId, fileURI, permissions, children, childAddress, created, lastModified, mimeType, docType);
+        final DocumentImpl doc = new DocumentImpl(null, pool, null, docId, fileURI, permissions, children, childAddress, created, lastModified, mimeType, xmlDecl, docType);
         doc.pageCount = pageCount;
         doc.userLock = userLock;
         doc.lockToken = lockToken;
@@ -1027,6 +1103,21 @@ protected NodeList findElementsByTagName(final NodeHandle root, final QName qnam
         return NodeSet.EMPTY_SET;
     }
 
+    /**
+     * The method <code>setXmlDeclaration</code>
+     *
+     * @param xmlDecl a <code>XMLDeclarationImpl</code> value
+     */
+    @EnsureContainerLocked(mode=WRITE_LOCK)
+    public void setXmlDeclaration(final XMLDeclarationImpl xmlDecl) {
+        this.xmlDecl = xmlDecl;
+    }
+
+    @EnsureContainerLocked(mode=READ_LOCK)
+    public @Nullable XMLDeclarationImpl getXmlDeclaration() {
+        return xmlDecl;
+    }
+
     /************************************************
      *
      * NodeImpl methods