aoqi@0: /*
aoqi@0: * Copyright (c) 1997, 2013, Oracle and/or its affiliates. All rights reserved.
aoqi@0: * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
aoqi@0: *
aoqi@0: * This code is free software; you can redistribute it and/or modify it
aoqi@0: * under the terms of the GNU General Public License version 2 only, as
aoqi@0: * published by the Free Software Foundation. Oracle designates this
aoqi@0: * particular file as subject to the "Classpath" exception as provided
aoqi@0: * by Oracle in the LICENSE file that accompanied this code.
aoqi@0: *
aoqi@0: * This code is distributed in the hope that it will be useful, but WITHOUT
aoqi@0: * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
aoqi@0: * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
aoqi@0: * version 2 for more details (a copy is included in the LICENSE file that
aoqi@0: * accompanied this code).
aoqi@0: *
aoqi@0: * You should have received a copy of the GNU General Public License version
aoqi@0: * 2 along with this work; if not, write to the Free Software Foundation,
aoqi@0: * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
aoqi@0: *
aoqi@0: * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
aoqi@0: * or visit www.oracle.com if you need additional information or have any
aoqi@0: * questions.
aoqi@0: */
aoqi@0:
aoqi@0: package com.sun.tools.internal.ws.wsdl.parser;
aoqi@0:
aoqi@0: import com.sun.istack.internal.NotNull;
aoqi@0: import com.sun.tools.internal.ws.util.xml.XmlUtil;
aoqi@0: import com.sun.tools.internal.ws.wscompile.ErrorReceiver;
aoqi@0: import com.sun.tools.internal.ws.wscompile.WsimportOptions;
aoqi@0: import com.sun.tools.internal.ws.wsdl.document.schema.SchemaConstants;
aoqi@0: import com.sun.tools.internal.xjc.reader.internalizer.LocatorTable;
aoqi@0: import com.sun.xml.internal.bind.marshaller.DataWriter;
aoqi@0: import org.w3c.dom.Document;
aoqi@0: import org.w3c.dom.Element;
aoqi@0: import org.w3c.dom.NodeList;
aoqi@0: import org.xml.sax.ContentHandler;
aoqi@0: import org.xml.sax.*;
aoqi@0: import org.xml.sax.helpers.XMLFilterImpl;
aoqi@0:
aoqi@0: import javax.xml.parsers.DocumentBuilder;
aoqi@0: import javax.xml.parsers.DocumentBuilderFactory;
aoqi@0: import javax.xml.parsers.ParserConfigurationException;
aoqi@0: import javax.xml.parsers.SAXParserFactory;
aoqi@0: import javax.xml.transform.Transformer;
aoqi@0: import javax.xml.transform.TransformerException;
aoqi@0: import javax.xml.transform.TransformerFactory;
aoqi@0: import javax.xml.transform.dom.DOMSource;
aoqi@0: import javax.xml.transform.sax.SAXResult;
aoqi@0: import java.io.IOException;
aoqi@0: import java.io.InputStream;
aoqi@0: import java.io.OutputStream;
aoqi@0: import java.io.OutputStreamWriter;
aoqi@0: import java.net.*;
aoqi@0: import java.util.*;
aoqi@0:
aoqi@0: /**
aoqi@0: * @author Vivek Pandey
aoqi@0: */
aoqi@0: public class DOMForest {
aoqi@0: /**
aoqi@0: * To correctly feed documents to a schema parser, we need to remember
aoqi@0: * which documents (of the forest) were given as the root
aoqi@0: * documents, and which of them are read as included/imported
aoqi@0: * documents.
aoqi@0: *
aoqi@0: *
aoqi@0: * Set of system ids as strings.
aoqi@0: */
aoqi@0: protected final Set rootDocuments = new HashSet();
aoqi@0:
aoqi@0: /**
aoqi@0: * Contains wsdl:import(s)
aoqi@0: */
aoqi@0: protected final Set externalReferences = new HashSet();
aoqi@0:
aoqi@0: /**
aoqi@0: * actual data storage map<SystemId,Document>.
aoqi@0: */
aoqi@0: protected final Map core = new HashMap();
aoqi@0: protected final ErrorReceiver errorReceiver;
aoqi@0:
aoqi@0: private final DocumentBuilder documentBuilder;
aoqi@0: private final SAXParserFactory parserFactory;
aoqi@0:
aoqi@0: /**
aoqi@0: * inlined schema elements inside wsdl:type section
aoqi@0: */
aoqi@0: protected final List inlinedSchemaElements = new ArrayList();
aoqi@0:
aoqi@0:
aoqi@0: /**
aoqi@0: * Stores location information for all the trees in this forest.
aoqi@0: */
aoqi@0: public final LocatorTable locatorTable = new LocatorTable();
aoqi@0:
aoqi@0: protected final EntityResolver entityResolver;
aoqi@0: /**
aoqi@0: * Stores all the outer-most <jaxb:bindings> customizations.
aoqi@0: */
aoqi@0: public final Set outerMostBindings = new HashSet();
aoqi@0:
aoqi@0: /**
aoqi@0: * Schema language dependent part of the processing.
aoqi@0: */
aoqi@0: protected final InternalizationLogic logic;
aoqi@0: protected final WsimportOptions options;
aoqi@0:
aoqi@0: public DOMForest(InternalizationLogic logic, @NotNull EntityResolver entityResolver, WsimportOptions options, ErrorReceiver errReceiver) {
aoqi@0: this.options = options;
aoqi@0: this.entityResolver = entityResolver;
aoqi@0: this.errorReceiver = errReceiver;
aoqi@0: this.logic = logic;
aoqi@0: try {
aoqi@0: // secure xml processing can be switched off if input requires it
aoqi@0: boolean secureProcessingEnabled = options == null || !options.disableXmlSecurity;
aoqi@0: DocumentBuilderFactory dbf = XmlUtil.newDocumentBuilderFactory(secureProcessingEnabled);
aoqi@0: dbf.setNamespaceAware(true);
aoqi@0: this.documentBuilder = dbf.newDocumentBuilder();
aoqi@0:
aoqi@0: this.parserFactory = XmlUtil.newSAXParserFactory(secureProcessingEnabled);
aoqi@0: this.parserFactory.setNamespaceAware(true);
aoqi@0: } catch (ParserConfigurationException e) {
aoqi@0: throw new AssertionError(e);
aoqi@0: }
aoqi@0: }
aoqi@0:
aoqi@0: public List getInlinedSchemaElement() {
aoqi@0: return inlinedSchemaElements;
aoqi@0: }
aoqi@0:
aoqi@0: public @NotNull Document parse(InputSource source, boolean root) throws SAXException, IOException {
aoqi@0: if (source.getSystemId() == null)
aoqi@0: throw new IllegalArgumentException();
aoqi@0: return parse(source.getSystemId(), source, root);
aoqi@0: }
aoqi@0:
aoqi@0: /**
aoqi@0: * Parses an XML at the given location (
aoqi@0: * and XMLs referenced by it) into DOM trees
aoqi@0: * and stores them to this forest.
aoqi@0: *
aoqi@0: * @return the parsed DOM document object.
aoqi@0: */
aoqi@0: public Document parse(String systemId, boolean root) throws SAXException, IOException{
aoqi@0:
aoqi@0: systemId = normalizeSystemId(systemId);
aoqi@0:
aoqi@0: InputSource is = null;
aoqi@0:
aoqi@0: // allow entity resolver to find the actual byte stream.
aoqi@0: is = entityResolver.resolveEntity(null, systemId);
aoqi@0: if (is == null)
aoqi@0: is = new InputSource(systemId);
aoqi@0: else {
aoqi@0: resolvedCache.put(systemId, is.getSystemId());
aoqi@0: systemId=is.getSystemId();
aoqi@0: }
aoqi@0:
aoqi@0: if (core.containsKey(systemId)) {
aoqi@0: // this document has already been parsed. Just ignore.
aoqi@0: return core.get(systemId);
aoqi@0: }
aoqi@0:
aoqi@0: if(!root)
aoqi@0: addExternalReferences(systemId);
aoqi@0:
aoqi@0: // but we still use the original system Id as the key.
aoqi@0: return parse(systemId, is, root);
aoqi@0: }
aoqi@0: protected Map resolvedCache = new HashMap();
aoqi@0:
aoqi@0: public Map getReferencedEntityMap() {
aoqi@0: return resolvedCache;
aoqi@0: }
aoqi@0: /**
aoqi@0: * Parses the given document and add it to the DOM forest.
aoqi@0: *
aoqi@0: * @return null if there was a parse error. otherwise non-null.
aoqi@0: */
aoqi@0: private @NotNull Document parse(String systemId, InputSource inputSource, boolean root) throws SAXException, IOException{
aoqi@0: Document dom = documentBuilder.newDocument();
aoqi@0:
aoqi@0: systemId = normalizeSystemId(systemId);
aoqi@0:
aoqi@0: // put into the map before growing a tree, to
aoqi@0: // prevent recursive reference from causing infinite loop.
aoqi@0: core.put(systemId, dom);
aoqi@0:
aoqi@0: dom.setDocumentURI(systemId);
aoqi@0: if (root)
aoqi@0: rootDocuments.add(systemId);
aoqi@0:
aoqi@0: try {
aoqi@0: XMLReader reader = createReader(dom);
aoqi@0:
aoqi@0: InputStream is = null;
aoqi@0: if(inputSource.getByteStream() == null){
aoqi@0: inputSource = entityResolver.resolveEntity(null, systemId);
aoqi@0: }
aoqi@0: reader.parse(inputSource);
aoqi@0: Element doc = dom.getDocumentElement();
aoqi@0: if (doc == null) {
aoqi@0: return null;
aoqi@0: }
aoqi@0: NodeList schemas = doc.getElementsByTagNameNS(SchemaConstants.NS_XSD, "schema");
aoqi@0: for (int i = 0; i < schemas.getLength(); i++) {
aoqi@0: inlinedSchemaElements.add((Element) schemas.item(i));
aoqi@0: }
aoqi@0: } catch (ParserConfigurationException e) {
aoqi@0: errorReceiver.error(e);
aoqi@0: throw new SAXException(e.getMessage());
aoqi@0: }
aoqi@0: resolvedCache.put(systemId, dom.getDocumentURI());
aoqi@0: return dom;
aoqi@0: }
aoqi@0:
aoqi@0: public void addExternalReferences(String ref) {
aoqi@0: if (!externalReferences.contains(ref))
aoqi@0: externalReferences.add(ref);
aoqi@0: }
aoqi@0:
aoqi@0:
aoqi@0: public Set getExternalReferences() {
aoqi@0: return externalReferences;
aoqi@0: }
aoqi@0:
aoqi@0:
aoqi@0:
aoqi@0: public interface Handler extends ContentHandler {
aoqi@0: /**
aoqi@0: * Gets the DOM that was built.
aoqi@0: */
aoqi@0: public Document getDocument();
aoqi@0: }
aoqi@0:
aoqi@0: /**
aoqi@0: * Returns a {@link org.xml.sax.XMLReader} to parse a document into this DOM forest.
aoqi@0: *
aoqi@0: * This version requires that the DOM object to be created and registered
aoqi@0: * to the map beforehand.
aoqi@0: */
aoqi@0: private XMLReader createReader(Document dom) throws SAXException, ParserConfigurationException {
aoqi@0: XMLReader reader = parserFactory.newSAXParser().getXMLReader();
aoqi@0: DOMBuilder dombuilder = new DOMBuilder(dom, locatorTable, outerMostBindings);
aoqi@0: try {
aoqi@0: reader.setProperty("http://xml.org/sax/properties/lexical-handler", dombuilder);
aoqi@0: } catch(SAXException e) {
aoqi@0: errorReceiver.debug(e.getMessage());
aoqi@0: }
aoqi@0:
aoqi@0: ContentHandler handler = new WhitespaceStripper(dombuilder, errorReceiver, entityResolver);
aoqi@0: handler = new VersionChecker(handler, errorReceiver, entityResolver);
aoqi@0:
aoqi@0: // insert the reference finder so that
aoqi@0: // included/imported schemas will be also parsed
aoqi@0: XMLFilterImpl f = logic.createExternalReferenceFinder(this);
aoqi@0: f.setContentHandler(handler);
aoqi@0: if (errorReceiver != null)
aoqi@0: f.setErrorHandler(errorReceiver);
aoqi@0: f.setEntityResolver(entityResolver);
aoqi@0:
aoqi@0: reader.setContentHandler(f);
aoqi@0: if (errorReceiver != null)
aoqi@0: reader.setErrorHandler(errorReceiver);
aoqi@0: reader.setEntityResolver(entityResolver);
aoqi@0: return reader;
aoqi@0: }
aoqi@0:
aoqi@0: private String normalizeSystemId(String systemId) {
aoqi@0: try {
aoqi@0: systemId = new URI(systemId).normalize().toString();
aoqi@0: } catch (URISyntaxException e) {
aoqi@0: // leave the system ID untouched. In my experience URI is often too strict
aoqi@0: }
aoqi@0: return systemId;
aoqi@0: }
aoqi@0:
aoqi@0: boolean isExtensionMode() {
aoqi@0: return options.isExtensionMode();
aoqi@0: }
aoqi@0:
aoqi@0:
aoqi@0: /**
aoqi@0: * Gets the DOM tree associated with the specified system ID,
aoqi@0: * or null if none is found.
aoqi@0: */
aoqi@0: public Document get(String systemId) {
aoqi@0: Document doc = core.get(systemId);
aoqi@0:
aoqi@0: if (doc == null && systemId.startsWith("file:/") && !systemId.startsWith("file://")) {
aoqi@0: // As of JDK1.4, java.net.URL.toExternal method returns URLs like
aoqi@0: // "file:/abc/def/ghi" which is an incorrect file protocol URL according to RFC1738.
aoqi@0: // Some other correctly functioning parts return the correct URLs ("file:///abc/def/ghi"),
aoqi@0: // and this descripancy breaks DOM look up by system ID.
aoqi@0:
aoqi@0: // this extra check solves this problem.
aoqi@0: doc = core.get("file://" + systemId.substring(5));
aoqi@0: }
aoqi@0:
aoqi@0: if (doc == null && systemId.startsWith("file:")) {
aoqi@0: // on Windows, filenames are case insensitive.
aoqi@0: // perform case-insensitive search for improved user experience
aoqi@0: String systemPath = getPath(systemId);
aoqi@0: for (String key : core.keySet()) {
aoqi@0: if (key.startsWith("file:") && getPath(key).equalsIgnoreCase(systemPath)) {
aoqi@0: doc = core.get(key);
aoqi@0: break;
aoqi@0: }
aoqi@0: }
aoqi@0: }
aoqi@0:
aoqi@0: return doc;
aoqi@0: }
aoqi@0:
aoqi@0: /**
aoqi@0: * Strips off the leading 'file:///' portion from an URL.
aoqi@0: */
aoqi@0: private String getPath(String key) {
aoqi@0: key = key.substring(5); // skip 'file:'
aoqi@0: while (key.length() > 0 && key.charAt(0) == '/')
aoqi@0: key = key.substring(1);
aoqi@0: return key;
aoqi@0: }
aoqi@0:
aoqi@0: /**
aoqi@0: * Gets all the system IDs of the documents.
aoqi@0: */
aoqi@0: public String[] listSystemIDs() {
aoqi@0: return core.keySet().toArray(new String[core.keySet().size()]);
aoqi@0: }
aoqi@0:
aoqi@0: /**
aoqi@0: * Gets the system ID from which the given DOM is parsed.
aoqi@0: *
aoqi@0: * Poor-man's base URI.
aoqi@0: */
aoqi@0: public String getSystemId(Document dom) {
aoqi@0: for (Map.Entry e : core.entrySet()) {
aoqi@0: if (e.getValue() == dom)
aoqi@0: return e.getKey();
aoqi@0: }
aoqi@0: return null;
aoqi@0: }
aoqi@0:
aoqi@0: /**
aoqi@0: * Gets the first one (which is more or less random) in {@link #rootDocuments}.
aoqi@0: */
aoqi@0: public String getFirstRootDocument() {
aoqi@0: if(rootDocuments.isEmpty()) return null;
aoqi@0: return rootDocuments.iterator().next();
aoqi@0: }
aoqi@0:
aoqi@0: public Set getRootDocuments() {
aoqi@0: return rootDocuments;
aoqi@0: }
aoqi@0:
aoqi@0: /**
aoqi@0: * Dumps the contents of the forest to the specified stream.
aoqi@0: *
aoqi@0: * This is a debug method. As such, error handling is sloppy.
aoqi@0: */
aoqi@0: public void dump(OutputStream out) throws IOException {
aoqi@0: try {
aoqi@0: // create identity transformer
aoqi@0: // secure xml processing can be switched off if input requires it
aoqi@0: boolean secureProcessingEnabled = options == null || !options.disableXmlSecurity;
aoqi@0: TransformerFactory tf = XmlUtil.newTransformerFactory(secureProcessingEnabled);
aoqi@0: Transformer it = tf.newTransformer();
aoqi@0:
aoqi@0: for (Map.Entry e : core.entrySet()) {
aoqi@0: out.write(("---<< " + e.getKey() + '\n').getBytes());
aoqi@0:
aoqi@0: DataWriter dw = new DataWriter(new OutputStreamWriter(out), null);
aoqi@0: dw.setIndentStep(" ");
aoqi@0: it.transform(new DOMSource(e.getValue()),
aoqi@0: new SAXResult(dw));
aoqi@0:
aoqi@0: out.write("\n\n\n".getBytes());
aoqi@0: }
aoqi@0: } catch (TransformerException e) {
aoqi@0: e.printStackTrace();
aoqi@0: }
aoqi@0: }
aoqi@0:
aoqi@0: }