1
|
/**
|
2
|
* '$RCSfile$'
|
3
|
* Copyright: 2007 Regents of the University of California and the
|
4
|
* National Center for Ecological Analysis and Synthesis
|
5
|
*
|
6
|
* '$Author: jones $'
|
7
|
* '$Date: 2013-05-23 13:26:48 -0700 (Thu, 23 May 2013) $'
|
8
|
* '$Revision: 7766 $'
|
9
|
*
|
10
|
* This program is free software; you can redistribute it and/or modify
|
11
|
* it under the terms of the GNU General Public License as published by
|
12
|
* the Free Software Foundation; either version 2 of the License, or
|
13
|
* (at your option) any later version.
|
14
|
*
|
15
|
* This program is distributed in the hope that it will be useful,
|
16
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
17
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
18
|
* GNU General Public License for more details.
|
19
|
*
|
20
|
* You should have received a copy of the GNU General Public License
|
21
|
* along with this program; if not, write to the Free Software
|
22
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
23
|
*/
|
24
|
|
25
|
package edu.ucsb.nceas.metacat;
|
26
|
|
27
|
import java.io.File;
|
28
|
import java.io.FileOutputStream;
|
29
|
import java.io.IOException;
|
30
|
import java.io.OutputStreamWriter;
|
31
|
import java.io.Writer;
|
32
|
import java.nio.charset.Charset;
|
33
|
import java.sql.PreparedStatement;
|
34
|
import java.sql.ResultSet;
|
35
|
import java.sql.SQLException;
|
36
|
import java.text.SimpleDateFormat;
|
37
|
import java.util.Date;
|
38
|
import java.util.TimerTask;
|
39
|
|
40
|
import org.apache.log4j.Logger;
|
41
|
|
42
|
import edu.ucsb.nceas.metacat.database.DBConnection;
|
43
|
import edu.ucsb.nceas.metacat.database.DBConnectionPool;
|
44
|
import edu.ucsb.nceas.metacat.properties.PropertyService;
|
45
|
import edu.ucsb.nceas.utilities.PropertyNotFoundException;
|
46
|
|
47
|
/**
|
48
|
* A Sitemap represents a document that lists all of the content of the Metacat
|
49
|
* server for use by harvesting spiders that wish to index the contents of the
|
50
|
* Metacat site. It is used to generate an XML representation of all of the URLs
|
51
|
* of the site in order to facilitate indexing of the metacat site by search
|
52
|
* engines.
|
53
|
*
|
54
|
* @author Matt Jones
|
55
|
*/
|
56
|
public class Sitemap extends TimerTask {
|
57
|
|
58
|
private static Logger logMetacat = Logger.getLogger(Sitemap.class);
|
59
|
|
60
|
/**
|
61
|
* Construct a new instance of the Sitemap.
|
62
|
*
|
63
|
* @param directory
|
64
|
* the location to store sitemap files
|
65
|
* @param urlRoot
|
66
|
* the base URL for constructing sitemap URLs
|
67
|
* @param skin
|
68
|
* the format skin to be used in URLs
|
69
|
*/
|
70
|
public Sitemap(File directory, String urlRoot, String skin) {
|
71
|
super();
|
72
|
this.directory = directory;
|
73
|
this.urlRoot = urlRoot;
|
74
|
this.skin = skin;
|
75
|
}
|
76
|
|
77
|
/**
|
78
|
* Execute the timed task when called, in this case by generating the
|
79
|
* sitemap files needed for this Metacat instance.
|
80
|
*/
|
81
|
public void run() {
|
82
|
generateSitemaps();
|
83
|
}
|
84
|
|
85
|
/**
|
86
|
* Generate all of the sitemap files needed to list the URLs from this
|
87
|
* instance of Metacat, using the open sitemap format described here:
|
88
|
* http://www.sitemaps.org/protocol.html
|
89
|
* URLs are written to a single file, unless the maximum number of URLs
|
90
|
* allowed in the sitemap file is exceeded, in which subsequent numbered
|
91
|
* files are created. An index of the sitemaps is also created.
|
92
|
*
|
93
|
* The sitemap index can be registered with search index providers such as
|
94
|
* Google, but beware that it needs to be accessible in a location above the
|
95
|
* mount point for the service URLs. By default the files are placed in
|
96
|
* {context}/sitemaps, but you will need to expose them at {context}/ for
|
97
|
* them to be trusted by Google. See the Sitemaps.org documentation for
|
98
|
* details.
|
99
|
*
|
100
|
* @param directory
|
101
|
* an existing File directory in which to write the sitemaps
|
102
|
* @param urlRoot
|
103
|
* the base URL to use in constructing document URLs
|
104
|
* @param skin
|
105
|
* the name of the skin to be used in formatting metacat
|
106
|
* documents
|
107
|
*/
|
108
|
public void generateSitemaps() {
|
109
|
|
110
|
logMetacat.info("Running the Sitemap task.");
|
111
|
|
112
|
// Test if the passed in File is a directory
|
113
|
if (directory.isDirectory()) {
|
114
|
// Query xml_documents to get list of documents
|
115
|
StringBuffer query = new StringBuffer();
|
116
|
// TODO: make the doctype configurable in the query
|
117
|
String sql =
|
118
|
"SELECT xml_documents.docid, xml_documents.rev " +
|
119
|
"FROM xml_documents, xml_access, identifier " +
|
120
|
"WHERE xml_documents.doctype LIKE 'eml:%' " +
|
121
|
"AND xml_documents.docid = identifier.docid " +
|
122
|
"AND xml_documents.rev = identifier.rev " +
|
123
|
"AND identifier.guid = xml_access.guid " +
|
124
|
"AND xml_access.principal_name = 'public' " +
|
125
|
"AND xml_access.perm_type = 'allow' " +
|
126
|
"order by docid, rev";
|
127
|
query.append(sql);
|
128
|
|
129
|
DBConnection dbConn = null;
|
130
|
int serialNumber = -1;
|
131
|
try {
|
132
|
// Get a database connection from the pool
|
133
|
dbConn = DBConnectionPool
|
134
|
.getDBConnection("Sitemap.generateSitemap()");
|
135
|
serialNumber = dbConn.getCheckOutSerialNumber();
|
136
|
|
137
|
// Execute the query statement
|
138
|
PreparedStatement stmt = dbConn.prepareStatement(query.toString());
|
139
|
stmt.execute();
|
140
|
ResultSet rs = stmt.getResultSet();
|
141
|
|
142
|
// Loop through all of the documents, and write them to a
|
143
|
// sitemap
|
144
|
File sitemapFile = null;
|
145
|
OutputStreamWriter sitemap = null;
|
146
|
int counter = 0;
|
147
|
int fileNumber = 0;
|
148
|
while (rs.next()) {
|
149
|
// Check if a new sitemap file needs to be created
|
150
|
if (counter % MAX_URLS_IN_FILE == 0) {
|
151
|
|
152
|
// if a sitemap file is already open
|
153
|
if (sitemapFile != null && sitemapFile.canWrite()) {
|
154
|
// write the footer and close the file
|
155
|
writeSitemapFooter(sitemap);
|
156
|
}
|
157
|
|
158
|
// Open a new sitemap file for writing
|
159
|
fileNumber++;
|
160
|
sitemapFile = new File(directory, fileRoot + fileNumber
|
161
|
+ ".xml");
|
162
|
sitemap = new OutputStreamWriter(new FileOutputStream(sitemapFile), Charset.forName("UTF-8"));
|
163
|
|
164
|
// Write the sitemap document header for the new file
|
165
|
writeSitemapHeader(sitemap);
|
166
|
}
|
167
|
|
168
|
String separator = PropertyService.getProperty("document.accNumSeparator");
|
169
|
String docid = rs.getString(1) + separator
|
170
|
+ rs.getString(2);
|
171
|
writeSitemapEntry(sitemap, docid);
|
172
|
counter++;
|
173
|
}
|
174
|
stmt.close();
|
175
|
writeSitemapFooter(sitemap);
|
176
|
writeSitemapIndex(fileNumber);
|
177
|
} catch (SQLException e) {
|
178
|
logMetacat.warn("Error while writing to the sitemap file: "
|
179
|
+ e.getMessage());
|
180
|
} catch (IOException ioe) {
|
181
|
logMetacat.warn("Could not open or write to the sitemap file."
|
182
|
+ ioe.getMessage());
|
183
|
} catch (PropertyNotFoundException pnfe) {
|
184
|
logMetacat.warn("Could not retrieve the account number separator."
|
185
|
+ pnfe.getMessage());
|
186
|
} finally {
|
187
|
// Return database connection to the pool
|
188
|
DBConnectionPool.returnDBConnection(dbConn, serialNumber);
|
189
|
}
|
190
|
} else {
|
191
|
logMetacat.warn("Sitemap not created because directory not valid.");
|
192
|
}
|
193
|
}
|
194
|
|
195
|
/**
|
196
|
* Write the header information in a single sitemap file. This includes the
|
197
|
* XML prolog, the root element and namespace declaration, and the elements
|
198
|
* leading up to the first URL entry.
|
199
|
*
|
200
|
* @param sitemap
|
201
|
* the Writer to use for writing the header
|
202
|
* @throws IOException
|
203
|
* if there is a problem writing to the Writer
|
204
|
*/
|
205
|
private void writeSitemapHeader(Writer sitemap) throws IOException {
|
206
|
sitemap.write(PROLOG);
|
207
|
String header = "<urlset xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\"\n" +
|
208
|
"xmlns:sm=\"http://www.sitemaps.org/schemas/sitemap/0.9\"\n" +
|
209
|
"xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\n" +
|
210
|
"xsi:schemaLocation=\"http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd\">\n";
|
211
|
|
212
|
sitemap.write(header);
|
213
|
sitemap.flush();
|
214
|
}
|
215
|
|
216
|
/**
|
217
|
* Write a URL entry to a single sitemap file. This includes the XML markup
|
218
|
* surrounding a particular site URL.
|
219
|
*
|
220
|
* @param sitemap
|
221
|
* the Writer to use for writing the URL
|
222
|
* @param docid
|
223
|
* the identifier to be written in the URL
|
224
|
* @param urlRoot
|
225
|
* the base URL to be used in constructing a URL
|
226
|
* @param skin
|
227
|
* the name of the skin to be used in constructing a URL
|
228
|
* @throws IOException
|
229
|
* if there is a problem writing to the Writer
|
230
|
*/
|
231
|
private void writeSitemapEntry(Writer sitemap, String docid)
|
232
|
throws IOException {
|
233
|
if (sitemap != null && docid != null && urlRoot != null) {
|
234
|
StringBuffer url = new StringBuffer();
|
235
|
url.append(urlRoot);
|
236
|
if (!urlRoot.endsWith("/")) {
|
237
|
url.append("/");
|
238
|
}
|
239
|
url.append(docid);
|
240
|
if (skin != null) {
|
241
|
url.append("/");
|
242
|
url.append(skin);
|
243
|
}
|
244
|
sitemap.write("<url><loc>");
|
245
|
sitemap.write(url.toString());
|
246
|
sitemap.write("</loc>");
|
247
|
// <lastmod>2005-01-01</lastmod>
|
248
|
// <changefreq>monthly</changefreq>
|
249
|
// <priority>0.8</priority>
|
250
|
sitemap.write("</url>");
|
251
|
sitemap.write("\n");
|
252
|
sitemap.flush();
|
253
|
}
|
254
|
}
|
255
|
|
256
|
/**
|
257
|
* Write the footer information in a single sitemap file and close the file.
|
258
|
* This includes the closing tag for the root element.
|
259
|
*
|
260
|
* @param sitemap
|
261
|
* the Writer to use for writing the footer
|
262
|
* @throws IOException
|
263
|
* if there is a problem writing to the Writer
|
264
|
*/
|
265
|
private void writeSitemapFooter(Writer sitemap) throws IOException {
|
266
|
if (sitemap != null)
|
267
|
{
|
268
|
String footer = "</urlset>\n";
|
269
|
sitemap.write(footer);
|
270
|
sitemap.close();
|
271
|
}
|
272
|
}
|
273
|
|
274
|
/**
|
275
|
* Create an index file listing all of the sitemap files that were created.
|
276
|
* @param fileNumber the number of sitemap files that were created.
|
277
|
*/
|
278
|
private void writeSitemapIndex(int fileNumber) {
|
279
|
|
280
|
// Open a new sitemapIndex file for writing
|
281
|
File sitemapIndexFile = null;
|
282
|
OutputStreamWriter sitemapIndex = null;
|
283
|
sitemapIndexFile = new File(directory, indexFilename);
|
284
|
try {
|
285
|
sitemapIndex = new OutputStreamWriter(new FileOutputStream(sitemapIndexFile), Charset.forName("UTF-8"));
|
286
|
|
287
|
// Write the sitemap index header for the new file
|
288
|
sitemapIndex.write(PROLOG);
|
289
|
String header = "<sitemapindex xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\"\n"
|
290
|
+ "xmlns:sm=\"http://www.sitemaps.org/schemas/sitemap/0.9\"\n" + "xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\n"
|
291
|
+ "xsi:schemaLocation=\"http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/siteindex.xsd\">\n";
|
292
|
sitemapIndex.write(header);
|
293
|
sitemapIndex.flush();
|
294
|
|
295
|
// Write out one index entry for each sitemap file
|
296
|
for (int fn = 1; fn <= fileNumber; fn++) {
|
297
|
String filename = fileRoot + fileNumber + ".xml";
|
298
|
writeSitemapIndexEntry(sitemapIndex, filename);
|
299
|
}
|
300
|
|
301
|
// Write the sitemap index footer content
|
302
|
if (sitemapIndex != null) {
|
303
|
String footer = "</sitemapindex>\n";
|
304
|
sitemapIndex.write(footer);
|
305
|
sitemapIndex.close();
|
306
|
}
|
307
|
|
308
|
// Close the index file
|
309
|
if (sitemapIndex != null) {
|
310
|
sitemapIndex.close();
|
311
|
}
|
312
|
|
313
|
} catch (IOException e) {
|
314
|
logMetacat.warn("Could not open or write to the sitemap index file." + e.getMessage());
|
315
|
}
|
316
|
}
|
317
|
|
318
|
/**
|
319
|
* Write a single line of the sitemap index file containing the URL to a specific sitemap file.
|
320
|
* @param sitemapIndex the writer to which the index information is written
|
321
|
* @param filename the name of the index file to be used
|
322
|
* @throws IOException on error writing to the index file
|
323
|
*/
|
324
|
private void writeSitemapIndexEntry(Writer sitemapIndex, String filename)
|
325
|
throws IOException {
|
326
|
if (sitemapIndex != null && filename != null && urlRoot != null) {
|
327
|
StringBuffer url = new StringBuffer();
|
328
|
url.append(urlRoot);
|
329
|
if (!urlRoot.endsWith("/")) {
|
330
|
url.append("/");
|
331
|
}
|
332
|
url.append(filename);
|
333
|
sitemapIndex.write("<sitemap><loc>");
|
334
|
sitemapIndex.write(url.toString());
|
335
|
sitemapIndex.write("</loc>");
|
336
|
Date now = new Date();
|
337
|
SimpleDateFormat fmt = new SimpleDateFormat("yyyy-MM-dd");
|
338
|
sitemapIndex.write("<lastmod>"+ fmt.format(now) +"</lastmod>");
|
339
|
sitemapIndex.write("</sitemap>");
|
340
|
sitemapIndex.write("\n");
|
341
|
sitemapIndex.flush();
|
342
|
}
|
343
|
}
|
344
|
|
345
|
// Member variables
|
346
|
|
347
|
/** The directory in which sitemaps are written. */
|
348
|
private File directory;
|
349
|
|
350
|
/** The root url for constructing sitemap URLs. */
|
351
|
private String urlRoot;
|
352
|
|
353
|
/** The name of the format skin to be used in sitemap URLs. */
|
354
|
private String skin;
|
355
|
|
356
|
/** Maximum number of URLs to write to a single sitemap file */
|
357
|
static final int MAX_URLS_IN_FILE = 25000; // 50,000 according to Google
|
358
|
|
359
|
/** The root name to be used in naming sitemap files. */
|
360
|
static final String fileRoot = "metacat";
|
361
|
|
362
|
/** The name to give to the sitemap index file */
|
363
|
static final String indexFilename = "metacatSitemapIndex.xml";
|
364
|
|
365
|
/** A String constant containing the XML prolog to be written in files. */
|
366
|
static final String PROLOG = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
|
367
|
}
|