src/skrueger/Sitemap.java

package skrueger;

import java.io.FileWriter;
import java.io.StringWriter;
import java.io.Writer;
import java.util.Date;

import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.transform.OutputKeys;
import javax.xml.transform.Result;
import javax.xml.transform.Transformer;
import javax.xml.transform.TransformerFactory;
import javax.xml.transform.dom.DOMSource;
import javax.xml.transform.stream.StreamResult;

import org.apache.log4j.Logger;
import org.w3c.dom.Document;
import org.w3c.dom.Element;

/**
 * This class provides an easy way to list all pages of your site (including
 * generates ones) ina sitemap.xml. See http://sitemaps.org/protocol.php for the
 * XML specification.
 * 
 * <br/>
 * XML output is based on org.w3c.dom
 * 
 * 
 * @author Stefan Tzeggai
 */
public class Sitemap {

        Logger log = Logger.getLogger(Sitemap.class);

        private Document document;

        final private Element urlsetElement;

        /**
         * Returns the number of URLs. May not be more than 50,000 by spec.
         */
        private int size = 0;

        final static String NSURL = "http://www.sitemaps.org/schemas/sitemap/0.9";

        public enum CHANGEFREQ {
                monthly, daily, weekly
        }

        public Sitemap() {

                try {

                        // Create a DOM builder and parse the fragment
                        final DocumentBuilderFactory factory = DocumentBuilderFactory
                                        .newInstance();
                        document = factory.newDocumentBuilder().newDocument();

                        // XML root element
                        urlsetElement = document.createElementNS(NSURL, "urlset");
                        document.appendChild(urlsetElement);

                        // // Linking this XML to the AtlasML Schema
                        // final Attr namespaces = document.createAttributeNS(
                        // "http://www.w3.org/2001/XMLSchema-instance", "schemaLocation");
                        // namespaces
                        // .setValue(NSURL+" http://localhost:"
                        // + Webserver.DEFAULTPORT
                        // + "/skrueger/atlas/resource/AtlasML.xsd");
                        // urlset.setAttributeNode(namespaces);

                } catch (Exception e) {
                        throw new RuntimeException(
                                        "Sitemap org.w3c.xml newDocumentBuilder failed:", e);
                }
        }

        /**
         * 
         * @param locString
         * @param lastmod
         *            when was the page modified the last time? May be
         *            <code>null</code>.
         * @param changefreq
         *            How often does the page change. May be <code>null</code>.
         * @param priority
         *            Priority 0.0 - 1.0. May be <code>null</code>.
         * @return <code>true</code> if the {@link Sitemap} contains less than 50000
         *         urls and the page was added.
         */
        public boolean addUrl(String locString, Date lastmod,
                        CHANGEFREQ changefreq, Double priority) {
                size++;
                if (size >= 50000)
                        return false;

                Element urlElement = document.createElement("url");

                if (locString == null)
                        throw new IllegalArgumentException("location must be provided");
                else {
                        Element e = document.createElement("loc");
                        e.appendChild(document.createTextNode(locString));
                        urlElement.appendChild(e);
                }

                if (lastmod != null) {
                        Element e = document.createElement("lastmod");
                        e.appendChild(document.createTextNode(lastmod.toString()));
                        urlElement.appendChild(e);
                }

                if (changefreq != null) {
                        Element e = document.createElement("changefreq");
                        e.appendChild(document.createTextNode(changefreq.toString()));
                        urlElement.appendChild(e);
                }

                if (priority != null) {
                        Element e = document.createElement("priority");
                        e.appendChild(document.createTextNode(priority.toString()));
                        urlElement.appendChild(e);
                }
                urlsetElement.appendChild(urlElement);
                return true;
        }

        public Document getDocument() {
                return document;
        }

        public String getXmlString() {
                StringWriter stringWriter = new StringWriter();
                outputToWriter(stringWriter);
                return stringWriter.toString();
        }

        /**
         * Can be used to write to a File with {@link FileWriter}.
         */
        public void outputToWriter(Writer stringWriter) {
                try {

                        try { // close outputStreamWriter.close();

                                // ****************************************************************************
                                // Create the XML
                                // ****************************************************************************
                                final Result result = new StreamResult(stringWriter);

                                // with indenting to make it human-readable
                                final TransformerFactory tf = TransformerFactory.newInstance();

                                // TODO Ging mit xerces, geht nicht mehr mit xalan ?! 
                                // tf.setAttribute("indent-number", new Integer(2));

                                final Transformer xformer = tf.newTransformer();
                                xformer.setOutputProperty(OutputKeys.INDENT, "yes");
                                xformer.setOutputProperty(
                                                "{http://xml.apache.org/xalan}indent-amount", "2");

                                // Write the DOM document to the file
                                xformer.transform(new DOMSource(document), result);

                        } finally {
                                stringWriter.close();
                        }

                } catch (Exception e) {
                        log.error("Failed to create sitemap.XML-String", e);
                        throw new RuntimeException(e);
                }
        }

        public int getSize() {
                return size;
        }
}
1	package skrueger;
2
3	import java.io.FileWriter;
4	import java.io.StringWriter;
5	import java.io.Writer;
6	import java.util.Date;
7
8	import javax.xml.parsers.DocumentBuilderFactory;
9	import javax.xml.transform.OutputKeys;
10	import javax.xml.transform.Result;
11	import javax.xml.transform.Transformer;
12	import javax.xml.transform.TransformerFactory;
13	import javax.xml.transform.dom.DOMSource;
14	import javax.xml.transform.stream.StreamResult;
15
16	import org.apache.log4j.Logger;
17	import org.w3c.dom.Document;
18	import org.w3c.dom.Element;
19
20	/**
21	* This class provides an easy way to list all pages of your site (including
22	* generates ones) ina sitemap.xml. See http://sitemaps.org/protocol.php for the
23	* XML specification.
24	*
25	* <br/>
26	* XML output is based on org.w3c.dom
27	*
28	*
29	* @author Stefan Tzeggai
30	*/
31	public class Sitemap {
32
33	Logger log = Logger.getLogger(Sitemap.class);
34
35	private Document document;
36
37	final private Element urlsetElement;
38
39	/**
40	* Returns the number of URLs. May not be more than 50,000 by spec.
41	*/
42	private int size = 0;
43
44	final static String NSURL = "http://www.sitemaps.org/schemas/sitemap/0.9";
45
46	public enum CHANGEFREQ {
47	monthly, daily, weekly
48	}
49
50	public Sitemap() {
51
52	try {
53
54	// Create a DOM builder and parse the fragment
55	final DocumentBuilderFactory factory = DocumentBuilderFactory
56	.newInstance();
57	document = factory.newDocumentBuilder().newDocument();
58
59	// XML root element
60	urlsetElement = document.createElementNS(NSURL, "urlset");
61	document.appendChild(urlsetElement);
62
63	// // Linking this XML to the AtlasML Schema
64	// final Attr namespaces = document.createAttributeNS(
65	// "http://www.w3.org/2001/XMLSchema-instance", "schemaLocation");
66	// namespaces
67	// .setValue(NSURL+" http://localhost:"
68	// + Webserver.DEFAULTPORT
69	// + "/skrueger/atlas/resource/AtlasML.xsd");
70	// urlset.setAttributeNode(namespaces);
71
72	} catch (Exception e) {
73	throw new RuntimeException(
74	"Sitemap org.w3c.xml newDocumentBuilder failed:", e);
75	}
76	}
77
78	/**
79	*
80	* @param locString
81	* @param lastmod
82	* when was the page modified the last time? May be
83	* <code>null</code>.
84	* @param changefreq
85	* How often does the page change. May be <code>null</code>.
86	* @param priority
87	* Priority 0.0 - 1.0. May be <code>null</code>.
88	* @return <code>true</code> if the {@link Sitemap} contains less than 50000
89	* urls and the page was added.
90	*/
91	public boolean addUrl(String locString, Date lastmod,
92	CHANGEFREQ changefreq, Double priority) {
93	size++;
94	if (size >= 50000)
95	return false;
96
97	Element urlElement = document.createElement("url");
98
99	if (locString == null)
100	throw new IllegalArgumentException("location must be provided");
101	else {
102	Element e = document.createElement("loc");
103	e.appendChild(document.createTextNode(locString));
104	urlElement.appendChild(e);
105	}
106
107	if (lastmod != null) {
108	Element e = document.createElement("lastmod");
109	e.appendChild(document.createTextNode(lastmod.toString()));
110	urlElement.appendChild(e);
111	}
112
113	if (changefreq != null) {
114	Element e = document.createElement("changefreq");
115	e.appendChild(document.createTextNode(changefreq.toString()));
116	urlElement.appendChild(e);
117	}
118
119	if (priority != null) {
120	Element e = document.createElement("priority");
121	e.appendChild(document.createTextNode(priority.toString()));
122	urlElement.appendChild(e);
123	}
124	urlsetElement.appendChild(urlElement);
125	return true;
126	}
127
128	public Document getDocument() {
129	return document;
130	}
131
132	public String getXmlString() {
133	StringWriter stringWriter = new StringWriter();
134	outputToWriter(stringWriter);
135	return stringWriter.toString();
136	}
137
138	/**
139	* Can be used to write to a File with {@link FileWriter}.
140	*/
141	public void outputToWriter(Writer stringWriter) {
142	try {
143
144	try { // close outputStreamWriter.close();
145
146	// ****************************************************************************
147	// Create the XML
148	// ****************************************************************************
149	final Result result = new StreamResult(stringWriter);
150
151	// with indenting to make it human-readable
152	final TransformerFactory tf = TransformerFactory.newInstance();
153
154	// TODO Ging mit xerces, geht nicht mehr mit xalan ?!
155	// tf.setAttribute("indent-number", new Integer(2));
156
157	final Transformer xformer = tf.newTransformer();
158	xformer.setOutputProperty(OutputKeys.INDENT, "yes");
159	xformer.setOutputProperty(
160	"{http://xml.apache.org/xalan}indent-amount", "2");
161
162	// Write the DOM document to the file
163	xformer.transform(new DOMSource(document), result);
164
165	} finally {
166	stringWriter.close();
167	}
168
169	} catch (Exception e) {
170	log.error("Failed to create sitemap.XML-String", e);
171	throw new RuntimeException(e);
172	}
173	}
174
175	public int getSize() {
176	return size;
177	}
178	}
Name	Value
svn:eol-style	native
svn:keywords	Id URL
svn:mime-type	text/plain