src/skrueger/Sitemap.java

package skrueger;

import java.io.FileWriter;
import java.io.IOException;
import java.io.StringWriter;
import java.io.Writer;
import java.net.MalformedURLException;
import java.net.URL;
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 {

        static String[] SEARCHENGINES = new String[] { "http://www.sitemapwriter.com/notify.php?crawler=all&url=" };

        final static 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;
        }

        public static void submitToSearchEngines(String urlToSitemap)  {
                for (String se : SEARCHENGINES) {
                        log.info("Submitting " + urlToSitemap + " to " + se);
                        try {
                                URL url = new URL(se + urlToSitemap);
                                
                                url.openConnection().connect();
                                
                        } catch (Exception e) {
                                log.error("failed to submit " + urlToSitemap + " to " + se, e);
                        }
                }
        }
}
1	alfonx	1317	package skrueger;
2
3			import java.io.FileWriter;
4	alfonx	1319	import java.io.IOException;
5	alfonx	1317	import java.io.StringWriter;
6			import java.io.Writer;
7	alfonx	1319	import java.net.MalformedURLException;
8			import java.net.URL;
9	alfonx	1317	import java.util.Date;
10
11			import javax.xml.parsers.DocumentBuilderFactory;
12			import javax.xml.transform.OutputKeys;
13			import javax.xml.transform.Result;
14			import javax.xml.transform.Transformer;
15			import javax.xml.transform.TransformerFactory;
16			import javax.xml.transform.dom.DOMSource;
17			import javax.xml.transform.stream.StreamResult;
18
19			import org.apache.log4j.Logger;
20			import org.w3c.dom.Document;
21			import org.w3c.dom.Element;
22
23			/**
24			* This class provides an easy way to list all pages of your site (including
25			* generates ones) ina sitemap.xml. See http://sitemaps.org/protocol.php for the
26			* XML specification.
27			*
28			* <br/>
29			* XML output is based on org.w3c.dom
30			*
31			*
32			* @author Stefan Tzeggai
33			*/
34			public class Sitemap {
35
36	alfonx	1319	static String[] SEARCHENGINES = new String[] { "http://www.sitemapwriter.com/notify.php?crawler=all&url=" };
37	alfonx	1317
38	alfonx	1319	final static Logger log = Logger.getLogger(Sitemap.class);
39
40	alfonx	1317	private Document document;
41
42			final private Element urlsetElement;
43
44			/**
45			* Returns the number of URLs. May not be more than 50,000 by spec.
46			*/
47			private int size = 0;
48
49			final static String NSURL = "http://www.sitemaps.org/schemas/sitemap/0.9";
50
51			public enum CHANGEFREQ {
52			monthly, daily, weekly
53			}
54
55			public Sitemap() {
56
57			try {
58
59			// Create a DOM builder and parse the fragment
60			final DocumentBuilderFactory factory = DocumentBuilderFactory
61			.newInstance();
62			document = factory.newDocumentBuilder().newDocument();
63
64			// XML root element
65			urlsetElement = document.createElementNS(NSURL, "urlset");
66			document.appendChild(urlsetElement);
67
68			// // Linking this XML to the AtlasML Schema
69			// final Attr namespaces = document.createAttributeNS(
70			// "http://www.w3.org/2001/XMLSchema-instance", "schemaLocation");
71			// namespaces
72			// .setValue(NSURL+" http://localhost:"
73			// + Webserver.DEFAULTPORT
74			// + "/skrueger/atlas/resource/AtlasML.xsd");
75			// urlset.setAttributeNode(namespaces);
76
77			} catch (Exception e) {
78			throw new RuntimeException(
79			"Sitemap org.w3c.xml newDocumentBuilder failed:", e);
80			}
81			}
82
83			/**
84			*
85			* @param locString
86			* @param lastmod
87			* when was the page modified the last time? May be
88			* <code>null</code>.
89			* @param changefreq
90			* How often does the page change. May be <code>null</code>.
91			* @param priority
92			* Priority 0.0 - 1.0. May be <code>null</code>.
93			* @return <code>true</code> if the {@link Sitemap} contains less than 50000
94			* urls and the page was added.
95			*/
96			public boolean addUrl(String locString, Date lastmod,
97			CHANGEFREQ changefreq, Double priority) {
98			size++;
99			if (size >= 50000)
100			return false;
101
102			Element urlElement = document.createElement("url");
103
104			if (locString == null)
105			throw new IllegalArgumentException("location must be provided");
106			else {
107			Element e = document.createElement("loc");
108			e.appendChild(document.createTextNode(locString));
109			urlElement.appendChild(e);
110			}
111
112			if (lastmod != null) {
113			Element e = document.createElement("lastmod");
114			e.appendChild(document.createTextNode(lastmod.toString()));
115			urlElement.appendChild(e);
116			}
117
118			if (changefreq != null) {
119			Element e = document.createElement("changefreq");
120			e.appendChild(document.createTextNode(changefreq.toString()));
121			urlElement.appendChild(e);
122			}
123
124			if (priority != null) {
125			Element e = document.createElement("priority");
126			e.appendChild(document.createTextNode(priority.toString()));
127			urlElement.appendChild(e);
128			}
129			urlsetElement.appendChild(urlElement);
130			return true;
131			}
132
133			public Document getDocument() {
134			return document;
135			}
136
137			public String getXmlString() {
138			StringWriter stringWriter = new StringWriter();
139			outputToWriter(stringWriter);
140			return stringWriter.toString();
141			}
142
143			/**
144			* Can be used to write to a File with {@link FileWriter}.
145			*/
146			public void outputToWriter(Writer stringWriter) {
147			try {
148
149			try { // close outputStreamWriter.close();
150
151			// ****************************************************************************
152			// Create the XML
153			// ****************************************************************************
154			final Result result = new StreamResult(stringWriter);
155
156			// with indenting to make it human-readable
157			final TransformerFactory tf = TransformerFactory.newInstance();
158
159	alfonx	1319	// TODO Ging mit xerces, geht nicht mehr mit xalan ?!
160	alfonx	1317	// tf.setAttribute("indent-number", new Integer(2));
161
162			final Transformer xformer = tf.newTransformer();
163			xformer.setOutputProperty(OutputKeys.INDENT, "yes");
164			xformer.setOutputProperty(
165			"{http://xml.apache.org/xalan}indent-amount", "2");
166
167			// Write the DOM document to the file
168			xformer.transform(new DOMSource(document), result);
169
170			} finally {
171			stringWriter.close();
172			}
173
174			} catch (Exception e) {
175			log.error("Failed to create sitemap.XML-String", e);
176			throw new RuntimeException(e);
177			}
178			}
179
180			public int getSize() {
181			return size;
182			}
183	alfonx	1319
184			public static void submitToSearchEngines(String urlToSitemap) {
185			for (String se : SEARCHENGINES) {
186			log.info("Submitting " + urlToSitemap + " to " + se);
187			try {
188			URL url = new URL(se + urlToSitemap);
189
190			url.openConnection().connect();
191
192			} catch (Exception e) {
193			log.error("failed to submit " + urlToSitemap + " to " + se, e);
194			}
195			}
196			}
197	alfonx	1317	}
Name	Value
svn:eol-style	native
svn:keywords	Id URL
svn:mime-type	text/plain