src/skrueger/Sitemap.java

package skrueger;

import java.io.FileWriter;
import java.io.StringWriter;
import java.io.Writer;
import java.net.URL;
import java.net.URLEncoder;
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=",
                        "http://submissions.ask.com/ping?sitemap=",
                        "http://www.google.com/webmasters/tools/ping?sitemap=",
                        "http://api.moreover.com/ping?u=",
                        "http://www.bing.com/webmaster/ping.aspx?siteMap=",
                        "http://search.yahooapis.com/SiteExplorerService/V1/ping?sitemap=" };

        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);

                                urlToSitemap = URLEncoder.encode(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			import java.io.StringWriter;
5			import java.io.Writer;
6	alfonx	1319	import java.net.URL;
7	alfonx	1321	import java.net.URLEncoder;
8	alfonx	1317	import java.util.Date;
9
10			import javax.xml.parsers.DocumentBuilderFactory;
11			import javax.xml.transform.OutputKeys;
12			import javax.xml.transform.Result;
13			import javax.xml.transform.Transformer;
14			import javax.xml.transform.TransformerFactory;
15			import javax.xml.transform.dom.DOMSource;
16			import javax.xml.transform.stream.StreamResult;
17
18			import org.apache.log4j.Logger;
19			import org.w3c.dom.Document;
20			import org.w3c.dom.Element;
21
22			/**
23			* This class provides an easy way to list all pages of your site (including
24			* generates ones) ina sitemap.xml. See http://sitemaps.org/protocol.php for the
25			* XML specification.
26			*
27			* <br/>
28			* XML output is based on org.w3c.dom
29			*
30			*
31			* @author Stefan Tzeggai
32			*/
33			public class Sitemap {
34
35	alfonx	1320	static String[] SEARCHENGINES = new String[] {
36			"http://www.sitemapwriter.com/notify.php?crawler=all&url=",
37			"http://submissions.ask.com/ping?sitemap=",
38			"http://www.google.com/webmasters/tools/ping?sitemap=",
39			"http://api.moreover.com/ping?u=",
40			"http://www.bing.com/webmaster/ping.aspx?siteMap=",
41			"http://search.yahooapis.com/SiteExplorerService/V1/ping?sitemap=" };
42	alfonx	1317
43	alfonx	1319	final static Logger log = Logger.getLogger(Sitemap.class);
44
45	alfonx	1317	private Document document;
46
47			final private Element urlsetElement;
48
49			/**
50			* Returns the number of URLs. May not be more than 50,000 by spec.
51			*/
52			private int size = 0;
53
54			final static String NSURL = "http://www.sitemaps.org/schemas/sitemap/0.9";
55
56			public enum CHANGEFREQ {
57			monthly, daily, weekly
58			}
59
60			public Sitemap() {
61
62			try {
63
64			// Create a DOM builder and parse the fragment
65			final DocumentBuilderFactory factory = DocumentBuilderFactory
66			.newInstance();
67			document = factory.newDocumentBuilder().newDocument();
68
69			// XML root element
70			urlsetElement = document.createElementNS(NSURL, "urlset");
71			document.appendChild(urlsetElement);
72
73			// // Linking this XML to the AtlasML Schema
74			// final Attr namespaces = document.createAttributeNS(
75			// "http://www.w3.org/2001/XMLSchema-instance", "schemaLocation");
76			// namespaces
77			// .setValue(NSURL+" http://localhost:"
78			// + Webserver.DEFAULTPORT
79			// + "/skrueger/atlas/resource/AtlasML.xsd");
80			// urlset.setAttributeNode(namespaces);
81
82			} catch (Exception e) {
83			throw new RuntimeException(
84			"Sitemap org.w3c.xml newDocumentBuilder failed:", e);
85			}
86			}
87
88			/**
89			*
90			* @param locString
91			* @param lastmod
92			* when was the page modified the last time? May be
93			* <code>null</code>.
94			* @param changefreq
95			* How often does the page change. May be <code>null</code>.
96			* @param priority
97			* Priority 0.0 - 1.0. May be <code>null</code>.
98			* @return <code>true</code> if the {@link Sitemap} contains less than 50000
99			* urls and the page was added.
100			*/
101			public boolean addUrl(String locString, Date lastmod,
102			CHANGEFREQ changefreq, Double priority) {
103			size++;
104			if (size >= 50000)
105			return false;
106
107			Element urlElement = document.createElement("url");
108
109			if (locString == null)
110			throw new IllegalArgumentException("location must be provided");
111			else {
112			Element e = document.createElement("loc");
113			e.appendChild(document.createTextNode(locString));
114			urlElement.appendChild(e);
115			}
116
117			if (lastmod != null) {
118			Element e = document.createElement("lastmod");
119			e.appendChild(document.createTextNode(lastmod.toString()));
120			urlElement.appendChild(e);
121			}
122
123			if (changefreq != null) {
124			Element e = document.createElement("changefreq");
125			e.appendChild(document.createTextNode(changefreq.toString()));
126			urlElement.appendChild(e);
127			}
128
129			if (priority != null) {
130			Element e = document.createElement("priority");
131			e.appendChild(document.createTextNode(priority.toString()));
132			urlElement.appendChild(e);
133			}
134			urlsetElement.appendChild(urlElement);
135			return true;
136			}
137
138			public Document getDocument() {
139			return document;
140			}
141
142			public String getXmlString() {
143			StringWriter stringWriter = new StringWriter();
144			outputToWriter(stringWriter);
145			return stringWriter.toString();
146			}
147
148			/**
149			* Can be used to write to a File with {@link FileWriter}.
150			*/
151			public void outputToWriter(Writer stringWriter) {
152			try {
153
154			try { // close outputStreamWriter.close();
155
156			// ****************************************************************************
157			// Create the XML
158			// ****************************************************************************
159			final Result result = new StreamResult(stringWriter);
160
161			// with indenting to make it human-readable
162			final TransformerFactory tf = TransformerFactory.newInstance();
163
164	alfonx	1319	// TODO Ging mit xerces, geht nicht mehr mit xalan ?!
165	alfonx	1317	// tf.setAttribute("indent-number", new Integer(2));
166
167			final Transformer xformer = tf.newTransformer();
168			xformer.setOutputProperty(OutputKeys.INDENT, "yes");
169			xformer.setOutputProperty(
170			"{http://xml.apache.org/xalan}indent-amount", "2");
171
172			// Write the DOM document to the file
173			xformer.transform(new DOMSource(document), result);
174
175			} finally {
176			stringWriter.close();
177			}
178
179			} catch (Exception e) {
180			log.error("Failed to create sitemap.XML-String", e);
181			throw new RuntimeException(e);
182			}
183			}
184
185			public int getSize() {
186			return size;
187			}
188	alfonx	1319
189	alfonx	1320	public static void submitToSearchEngines(String urlToSitemap) {
190	alfonx	1319	for (String se : SEARCHENGINES) {
191			log.info("Submitting " + urlToSitemap + " to " + se);
192			try {
193			URL url = new URL(se + urlToSitemap);
194	alfonx	1320
195	alfonx	1324	urlToSitemap = URLEncoder.encode(urlToSitemap);
196	alfonx	1321
197	alfonx	1319	url.openConnection().connect();
198	alfonx	1320
199	alfonx	1319	} catch (Exception e) {
200			log.error("failed to submit " + urlToSitemap + " to " + se, e);
201			}
202			}
203			}
204	alfonx	1317	}
Name	Value
svn:eol-style	native
svn:keywords	Id URL
svn:mime-type	text/plain