aboutsummaryrefslogtreecommitdiffstats
path: root/netx/net/sourceforge/jnlp/XmlParser.java
diff options
context:
space:
mode:
Diffstat (limited to 'netx/net/sourceforge/jnlp/XmlParser.java')
-rw-r--r--netx/net/sourceforge/jnlp/XmlParser.java183
1 files changed, 183 insertions, 0 deletions
diff --git a/netx/net/sourceforge/jnlp/XmlParser.java b/netx/net/sourceforge/jnlp/XmlParser.java
new file mode 100644
index 0000000..05080ea
--- /dev/null
+++ b/netx/net/sourceforge/jnlp/XmlParser.java
@@ -0,0 +1,183 @@
+/*
+ Copyright (C) 2013 Red Hat, Inc.
+
+This file is part of IcedTea.
+
+IcedTea is free software; you can redistribute it and/or
+modify it under the terms of the GNU General Public License as published by
+the Free Software Foundation, version 2.
+
+IcedTea is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with IcedTea; see the file COPYING. If not, write to
+the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+02110-1301 USA.
+
+Linking this library statically or dynamically with other modules is
+making a combined work based on this library. Thus, the terms and
+conditions of the GNU General Public License cover the whole
+combination.
+
+As a special exception, the copyright holders of this library give you
+permission to link this library with independent modules to produce an
+executable, regardless of the license terms of these independent
+modules, and to copy and distribute the resulting executable under
+terms of your choice, provided that you also meet, for each linked
+independent module, the terms and conditions of the license of that
+module. An independent module is a module which is not derived from
+or based on this library. If you modify this library, you may extend
+this exception to your version of the library, but you are not
+obligated to do so. If you do not wish to do so, delete this
+exception statement from your version.
+ */
+
+package net.sourceforge.jnlp;
+
+import static net.sourceforge.jnlp.runtime.Translator.R;
+
+import java.io.BufferedInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.InputStreamReader;
+import java.io.PipedInputStream;
+import java.io.PipedOutputStream;
+
+import net.sourceforge.nanoxml.XMLElement;
+
+//import javax.xml.parsers.*; // commented to use right Node
+//import org.w3c.dom.*; // class for using Tiny XML | NanoXML
+//import org.xml.sax.*;
+//import gd.xml.tiny.*;
+
+/**
+ * A gateway to the actual implementation of the parsers.
+ *
+ * Used by net.sourceforge.jnlp.Parser
+ */
+class XMLParser {
+
+ /**
+ * Parses input from an InputStream and returns a Node representing the
+ * root of the parse tree.
+ *
+ * @param input the {@link InputStream} containing the XML
+ * @return a {@link Node} representing the root of the parsed XML
+ * @throws ParseException
+ */
+ public Node getRootNode(InputStream input) throws ParseException {
+
+ try {
+ /* SAX
+ DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
+ factory.setValidating(false);
+ factory.setNamespaceAware(true);
+ DocumentBuilder builder = factory.newDocumentBuilder();
+ builder.setErrorHandler(errorHandler);
+
+ Document doc = builder.parse(input);
+ return doc.getDocumentElement();
+ */
+
+ /* TINY
+ Node document = new Node(TinyParser.parseXML(input));
+ Node jnlpNode = getChildNode(document, "jnlp"); // skip comments
+ */
+
+ //A BufferedInputStream is used to allow marking and reseting
+ //of a stream.
+ BufferedInputStream bs = new BufferedInputStream(input);
+
+ /* NANO */
+ final XMLElement xml = new XMLElement();
+ final PipedInputStream pin = new PipedInputStream();
+ final PipedOutputStream pout = new PipedOutputStream(pin);
+ final InputStreamReader isr = new InputStreamReader(bs, getEncoding(bs));
+ // Clean the jnlp xml file of all comments before passing
+ // it to the parser.
+ new Thread(
+ new Runnable() {
+ public void run() {
+ (new XMLElement()).sanitizeInput(isr, pout);
+ try {
+ pout.close();
+ } catch (IOException ioe) {
+ ioe.printStackTrace();
+ }
+ }
+ }).start();
+ xml.parseFromReader(new InputStreamReader(pin));
+ Node jnlpNode = new Node(xml);
+ return jnlpNode;
+ } catch (Exception ex) {
+ throw new ParseException(R("PBadXML"), ex);
+ }
+ }
+
+ /**
+ * Returns the name of the encoding used in this InputStream.
+ *
+ * @param input the InputStream
+ * @return a String representation of encoding
+ */
+ private static String getEncoding(InputStream input) throws IOException {
+ //Fixme: This only recognizes UTF-8, UTF-16, and
+ //UTF-32, which is enough to parse the prolog portion of xml to
+ //find out the exact encoding (if it exists). The reason being
+ //there could be other encodings, such as ISO 8859 which is 8-bits
+ //but it supports latin characters.
+ //So what needs to be done is to parse the prolog and retrieve
+ //the exact encoding from it.
+
+ int[] s = new int[4];
+ String encoding = "UTF-8";
+
+ //Determine what the first four bytes are and store
+ //them into an int array.
+ input.mark(4);
+ for (int i = 0; i < 4; i++) {
+ s[i] = input.read();
+ }
+ input.reset();
+
+ //Set the encoding base on what the first four bytes of the
+ //inputstream turn out to be (following the information from
+ //www.w3.org/TR/REC-xml/#sec-guessing).
+ if (s[0] == 255) {
+ if (s[1] == 254) {
+ if (s[2] != 0 || s[3] != 0) {
+ encoding = "UnicodeLittle";
+ } else {
+ encoding = "X-UTF-32LE-BOM";
+ }
+ }
+ } else if (s[0] == 254 && s[1] == 255 && (s[2] != 0 ||
+ s[3] != 0)) {
+ encoding = "UTF-16";
+
+ } else if (s[0] == 0 && s[1] == 0 && s[2] == 254 &&
+ s[3] == 255) {
+ encoding = "X-UTF-32BE-BOM";
+
+ } else if (s[0] == 0 && s[1] == 0 && s[2] == 0 &&
+ s[3] == 60) {
+ encoding = "UTF-32BE";
+
+ } else if (s[0] == 60 && s[1] == 0 && s[2] == 0 &&
+ s[3] == 0) {
+ encoding = "UTF-32LE";
+
+ } else if (s[0] == 0 && s[1] == 60 && s[2] == 0 &&
+ s[3] == 63) {
+ encoding = "UTF-16BE";
+ } else if (s[0] == 60 && s[1] == 0 && s[2] == 63 &&
+ s[3] == 0) {
+ encoding = "UTF-16LE";
+ }
+
+ return encoding;
+ }
+}