1
|
/**
|
2
|
* '$RCSfile$'
|
3
|
* Copyright: 2004 University of New Mexico and the
|
4
|
* Regents of the University of California
|
5
|
*
|
6
|
* '$Author: daigle $'
|
7
|
* '$Date: 2008-07-28 17:07:03 -0700 (Mon, 28 Jul 2008) $'
|
8
|
* '$Revision: 4175 $'
|
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.harvesterClient;
|
26
|
|
27
|
import com.oreilly.servlet.MailMessage;
|
28
|
import java.io.IOException;
|
29
|
import java.io.PrintStream;
|
30
|
import java.sql.Connection;
|
31
|
import java.sql.DriverManager;
|
32
|
import java.sql.ResultSet;
|
33
|
import java.sql.SQLException;
|
34
|
import java.sql.SQLWarning;
|
35
|
import java.sql.Statement;
|
36
|
import java.util.ArrayList;
|
37
|
import java.text.SimpleDateFormat;
|
38
|
import java.util.Date;
|
39
|
|
40
|
|
41
|
import edu.ucsb.nceas.metacat.client.Metacat;
|
42
|
import edu.ucsb.nceas.metacat.client.MetacatFactory;
|
43
|
import edu.ucsb.nceas.metacat.client.MetacatInaccessibleException;
|
44
|
import edu.ucsb.nceas.metacat.service.PropertyService;
|
45
|
import edu.ucsb.nceas.metacat.service.ServiceException;
|
46
|
import edu.ucsb.nceas.metacat.util.SystemUtil;
|
47
|
import edu.ucsb.nceas.utilities.PropertyNotFoundException;
|
48
|
|
49
|
/**
|
50
|
* Harvester is the main class for the Harvester application. The main
|
51
|
* method creates a single Harvester object which drives the application.
|
52
|
*
|
53
|
* @author costa
|
54
|
*
|
55
|
*/
|
56
|
public class Harvester {
|
57
|
|
58
|
/*
|
59
|
* Class fields
|
60
|
*/
|
61
|
private static final String CONFIG_DIR = "../../build/war/WEB-INF";
|
62
|
private static final String CONFIG_DIR_TEST = "./build/war/WEB-INF";
|
63
|
// private static final String CONFIG_NAME = "metacat.properties";
|
64
|
public static final String filler = "*";
|
65
|
private static boolean keepRunning = true;
|
66
|
public static final String marker =
|
67
|
"*****************************************************************************";
|
68
|
// public static PropertyService propertyService = null;
|
69
|
private static String schemaLocation = null;
|
70
|
|
71
|
|
72
|
/*
|
73
|
* Class methods
|
74
|
*/
|
75
|
|
76
|
|
77
|
/**
|
78
|
* Constructor. Creates a new instance of Harvester.
|
79
|
*/
|
80
|
public Harvester() {
|
81
|
}
|
82
|
|
83
|
|
84
|
/**
|
85
|
* Loads Harvester options from a configuration file.
|
86
|
*/
|
87
|
public static void loadProperties(boolean test) {
|
88
|
String configDir = test ? CONFIG_DIR_TEST : CONFIG_DIR;
|
89
|
|
90
|
try {
|
91
|
PropertyService.getInstance(configDir);
|
92
|
}
|
93
|
catch (ServiceException e) {
|
94
|
System.out.println("Error in loading properties: " + e.getMessage());
|
95
|
}
|
96
|
}
|
97
|
|
98
|
|
99
|
/**
|
100
|
* Harvester main method.
|
101
|
*
|
102
|
* @param args
|
103
|
* the command line arguments
|
104
|
* @throws SAXException
|
105
|
* @throws IOException
|
106
|
* @throws ParserConfigurationException
|
107
|
*/
|
108
|
public static void main(String[] args) {
|
109
|
Integer delayDefault = new Integer(0); // Default number of hours delay
|
110
|
int delay = delayDefault.intValue(); // Delay in hours before first
|
111
|
// harvest
|
112
|
Integer d; // Used for determining delay
|
113
|
long delta; // endTime - startTime
|
114
|
long endTime; // time that a harvest completes
|
115
|
Harvester harvester; // object for a single harvest run
|
116
|
Integer maxHarvestsDefault = new Integer(0); // Default max harvests
|
117
|
int maxHarvests = maxHarvestsDefault.intValue(); // Max number of
|
118
|
// harvests
|
119
|
Integer mh; // used in determining max harvests
|
120
|
int nHarvests = 0; // counts the number of harvest runs
|
121
|
final long oneHour = (60 * 60 * 1000); // milliseconds in one hour
|
122
|
Integer periodDefault = new Integer(24); // Default hours between
|
123
|
// harvests
|
124
|
int period = periodDefault.intValue(); // Hours between harvests
|
125
|
Integer p; // Used in determining the period
|
126
|
long startTime; // time that a harvest run starts
|
127
|
boolean test = false; // Passed to loadOption()
|
128
|
|
129
|
if (args.length > 0) {
|
130
|
schemaLocation = args[0];
|
131
|
System.err.println("schemaLocation: " + schemaLocation);
|
132
|
|
133
|
try {
|
134
|
Thread.sleep(10000);
|
135
|
} catch (InterruptedException e) {
|
136
|
e.printStackTrace();
|
137
|
}
|
138
|
}
|
139
|
|
140
|
System.out.println(marker);
|
141
|
System.out.println("Starting Harvester");
|
142
|
Harvester.loadProperties(test);
|
143
|
|
144
|
// Parse the delay property. Use default if necessary.
|
145
|
try {
|
146
|
d = Integer.valueOf(PropertyService.getProperty("harvester.delay"));
|
147
|
delay = d.intValue();
|
148
|
} catch (NumberFormatException e) {
|
149
|
System.out.println("NumberFormatException: Error parsing delay: "
|
150
|
+ e.getMessage());
|
151
|
System.out.println("Defaulting to delay=" + delayDefault);
|
152
|
delay = delayDefault.intValue();
|
153
|
} catch (PropertyNotFoundException pnfe) {
|
154
|
System.out.println("PropertyNotFoundException: Error finding delay: "
|
155
|
+ pnfe.getMessage());
|
156
|
System.out.println("Defaulting to delay=" + delayDefault);
|
157
|
delay = delayDefault.intValue();
|
158
|
}
|
159
|
|
160
|
// Parse the maxHarvests property. Use default if necessary.
|
161
|
try {
|
162
|
mh = Integer.valueOf(PropertyService.getProperty("harvester.maxHarvests"));
|
163
|
maxHarvests = mh.intValue();
|
164
|
} catch (NumberFormatException e) {
|
165
|
System.out.println("NumberFormatException: Error parsing maxHarvests: "
|
166
|
+ e.getMessage());
|
167
|
System.out.println("Defaulting to maxHarvests=" + maxHarvestsDefault);
|
168
|
maxHarvests = maxHarvestsDefault.intValue();
|
169
|
} catch (PropertyNotFoundException pnfe) {
|
170
|
System.out.println("PropertyNotFoundException: Error finding maxHarvests: "
|
171
|
+ pnfe.getMessage());
|
172
|
System.out.println("Defaulting to maxHarvests=" + maxHarvestsDefault);
|
173
|
maxHarvests = maxHarvestsDefault.intValue();
|
174
|
}
|
175
|
|
176
|
// Parse the period property. Use default if necessary.
|
177
|
try {
|
178
|
p = Integer.valueOf(PropertyService.getProperty("harvester.period"));
|
179
|
period = p.intValue();
|
180
|
} catch (NumberFormatException e) {
|
181
|
System.out.println("NumberFormatException: Error parsing period: "
|
182
|
+ e.getMessage());
|
183
|
System.out.println("Defaulting to period=" + periodDefault);
|
184
|
period = periodDefault.intValue();
|
185
|
} catch (PropertyNotFoundException pnfe) {
|
186
|
System.out.println("PropertyNotFoundException: Error finding period: "
|
187
|
+ pnfe.getMessage());
|
188
|
System.out.println("Defaulting to period=" + periodDefault);
|
189
|
period = periodDefault.intValue();
|
190
|
}
|
191
|
|
192
|
// Sleep for delay number of hours prior to starting first harvest
|
193
|
if (delay > 0) {
|
194
|
try {
|
195
|
System.out.print("First harvest will begin in " + delay);
|
196
|
if (delay == 1) {
|
197
|
System.out.println(" hour.");
|
198
|
} else {
|
199
|
System.out.println(" hours.");
|
200
|
}
|
201
|
Thread.sleep(delay * oneHour);
|
202
|
} catch (InterruptedException e) {
|
203
|
System.err.println("InterruptedException: " + e.getMessage());
|
204
|
System.exit(1);
|
205
|
}
|
206
|
}
|
207
|
|
208
|
// Repeat a new harvest once every period number of hours, until we reach
|
209
|
// the maximum number of harvests, or indefinitely if maxHarvests <= 0.
|
210
|
// Subtract delta from the time period so
|
211
|
// that each harvest will start at a fixed interval.
|
212
|
//
|
213
|
while (keepRunning && ((nHarvests < maxHarvests) || (maxHarvests <= 0))) {
|
214
|
nHarvests++;
|
215
|
startTime = System.currentTimeMillis();
|
216
|
harvester = new Harvester(); // New object for this
|
217
|
// harvest
|
218
|
harvester.startup(nHarvests, maxHarvests); // Start up Harvester
|
219
|
harvester.readHarvestSiteSchedule(); // Read the database table
|
220
|
harvester.harvest(); // Harvest the documents
|
221
|
harvester.shutdown(); // Shut down Harvester
|
222
|
endTime = System.currentTimeMillis();
|
223
|
delta = endTime - startTime;
|
224
|
|
225
|
if ((nHarvests < maxHarvests) || (maxHarvests <= 0)) {
|
226
|
try {
|
227
|
System.out.println("Next harvest will begin in " +
|
228
|
period + " hours.");
|
229
|
Thread.sleep((period * oneHour) - delta);
|
230
|
}
|
231
|
catch (InterruptedException e) {
|
232
|
System.err.println("InterruptedException: " + e.getMessage());
|
233
|
System.exit(1);
|
234
|
}
|
235
|
}
|
236
|
}
|
237
|
}
|
238
|
|
239
|
|
240
|
/**
|
241
|
* Set the keepRunning flag. If set to false, the main program will end
|
242
|
* the while loop that keeps harvester running every period number of hours.
|
243
|
* The static method is intended to be called from the HarvesterServlet class
|
244
|
* which creates a thread to run Harvester. When the thread is destroyed, the
|
245
|
* thread's destroy() method calls Harvester.setKeepRunning(false).
|
246
|
*
|
247
|
* @param keepRunning
|
248
|
*/
|
249
|
static void setKeepRunning(boolean keepRunning) {
|
250
|
Harvester.keepRunning = keepRunning;
|
251
|
}
|
252
|
|
253
|
|
254
|
/*
|
255
|
* Object fields
|
256
|
*/
|
257
|
|
258
|
/** Database connection */
|
259
|
private Connection conn = null;
|
260
|
|
261
|
/** Used during development to determine whether to connect to metacat
|
262
|
* Sometimes it's useful to test parts of the code without actually
|
263
|
* connecting to Metacat.
|
264
|
*/
|
265
|
private boolean connectToMetacat;
|
266
|
|
267
|
/** Highest DETAIL_LOG_ID primary key in the HARVEST_DETAIL_LOG table */
|
268
|
private int detailLogID;
|
269
|
|
270
|
/** Email address of the Harvester Administrator */
|
271
|
String harvesterAdministrator;
|
272
|
|
273
|
/** Highest HARVEST_LOG_ID primary key in the HARVEST_LOG table */
|
274
|
private int harvestLogID;
|
275
|
|
276
|
/** End time of this harvest session */
|
277
|
private Date harvestEndTime;
|
278
|
|
279
|
/** List of HarvestLog objects. Stores log entries for report generation. */
|
280
|
private ArrayList harvestLogList = new ArrayList();
|
281
|
|
282
|
/** List of HarvestSiteSchedule objects */
|
283
|
private ArrayList harvestSiteScheduleList = new ArrayList();
|
284
|
|
285
|
/** Start time of this harvest session */
|
286
|
private Date harvestStartTime;
|
287
|
|
288
|
/** Number of days to save log records. Any that are older are purged. */
|
289
|
int logPeriod;
|
290
|
|
291
|
/** Metacat client object */
|
292
|
Metacat metacat;
|
293
|
|
294
|
/** SMTP server for sending mail messages */
|
295
|
String smtpServer;
|
296
|
|
297
|
/** The timestamp for this harvest run. Used for output only. */
|
298
|
String timestamp;
|
299
|
|
300
|
|
301
|
/*
|
302
|
* Object methods
|
303
|
*/
|
304
|
|
305
|
/**
|
306
|
* Creates a new HarvestLog object and adds it to the harvestLogList.
|
307
|
*
|
308
|
* @param status the status of the harvest operation
|
309
|
* @param message the message text of the harvest operation
|
310
|
* @param harvestOperationCode the harvest operation code
|
311
|
* @param siteScheduleID the siteScheduleID for which this operation was
|
312
|
* performed. 0 indicates that the operation did not
|
313
|
* involve a particular harvest site.
|
314
|
* @param harvestDocument the associated HarvestDocument object. May be null.
|
315
|
* @param errorMessage additional error message pertaining to document
|
316
|
* error.
|
317
|
*/
|
318
|
void addLogEntry(int status,
|
319
|
String message,
|
320
|
String harvestOperationCode,
|
321
|
int siteScheduleID,
|
322
|
HarvestDocument harvestDocument,
|
323
|
String errorMessage
|
324
|
) {
|
325
|
HarvestLog harvestLog;
|
326
|
int harvestLogID = getHarvestLogID();
|
327
|
int detailLogID;
|
328
|
|
329
|
/* If there is no associated harvest document, call the basic constructor;
|
330
|
* else call the extended constructor.
|
331
|
*/
|
332
|
if (harvestDocument == null) {
|
333
|
harvestLog = new HarvestLog(this, conn, harvestLogID, harvestStartTime,
|
334
|
status, message, harvestOperationCode,
|
335
|
siteScheduleID);
|
336
|
}
|
337
|
else {
|
338
|
detailLogID = getDetailLogID();
|
339
|
harvestLog = new HarvestLog(this, conn, harvestLogID, detailLogID,
|
340
|
harvestStartTime, status, message,
|
341
|
harvestOperationCode, siteScheduleID,
|
342
|
harvestDocument, errorMessage);
|
343
|
}
|
344
|
|
345
|
harvestLogList.add(harvestLog);
|
346
|
}
|
347
|
|
348
|
|
349
|
public void closeConnection() {
|
350
|
try {
|
351
|
// Close the database connection
|
352
|
System.out.println("Closing the database connection.");
|
353
|
conn.close();
|
354
|
}
|
355
|
catch (SQLException e) {
|
356
|
System.out.println("Database access failed " + e);
|
357
|
}
|
358
|
}
|
359
|
|
360
|
|
361
|
/**
|
362
|
* Determines whether Harvester should attempt to connect to Metacat.
|
363
|
* Used during development and testing.
|
364
|
*
|
365
|
* @return true if Harvester should connect, otherwise false
|
366
|
*/
|
367
|
boolean connectToMetacat () {
|
368
|
return connectToMetacat;
|
369
|
}
|
370
|
|
371
|
|
372
|
/**
|
373
|
* Normalizes text prior to insertion into the HARVEST_LOG or
|
374
|
* HARVEST_DETAIL_LOG tables. In particular, replaces the single quote
|
375
|
* character with the double quote character. This prevents SQL errors
|
376
|
* involving words that contain single quotes. Also removes \n and \r
|
377
|
* characters from the text.
|
378
|
*
|
379
|
* @param text the original string
|
380
|
* @return a string containing the normalized text
|
381
|
*/
|
382
|
public String dequoteText(String text) {
|
383
|
char c;
|
384
|
StringBuffer stringBuffer = new StringBuffer();
|
385
|
|
386
|
for (int i = 0; i < text.length(); i++) {
|
387
|
c = text.charAt(i);
|
388
|
switch (c) {
|
389
|
case '\'':
|
390
|
stringBuffer.append('\"');
|
391
|
break;
|
392
|
case '\r':
|
393
|
case '\n':
|
394
|
break;
|
395
|
default:
|
396
|
stringBuffer.append(c);
|
397
|
break;
|
398
|
}
|
399
|
}
|
400
|
|
401
|
return stringBuffer.toString();
|
402
|
}
|
403
|
|
404
|
/**
|
405
|
* Returns a connection to the database. Opens the connection if a connection
|
406
|
* has not already been made previously.
|
407
|
*
|
408
|
* @return conn the database Connection object
|
409
|
*/
|
410
|
public Connection getConnection() {
|
411
|
String dbDriver = "";
|
412
|
String defaultDB = null;
|
413
|
String password = null;
|
414
|
String user = null;
|
415
|
SQLWarning warn;
|
416
|
|
417
|
if (conn == null) {
|
418
|
try {
|
419
|
dbDriver = PropertyService.getProperty("database.driver");
|
420
|
defaultDB = PropertyService.getProperty("database.connectionURI");
|
421
|
password = PropertyService.getProperty("database.password");
|
422
|
user = PropertyService.getProperty("database.user");
|
423
|
} catch (PropertyNotFoundException pnfe) {
|
424
|
System.out.println("Can't find property " + pnfe);
|
425
|
System.exit(1);
|
426
|
}
|
427
|
|
428
|
// Load the jdbc driver
|
429
|
try {
|
430
|
Class.forName(dbDriver);
|
431
|
}
|
432
|
catch (ClassNotFoundException e) {
|
433
|
System.out.println("Can't load driver " + e);
|
434
|
System.exit(1);
|
435
|
}
|
436
|
|
437
|
// Make the database connection
|
438
|
try {
|
439
|
System.out.println("Getting connection to Harvester tables");
|
440
|
conn = DriverManager.getConnection(defaultDB, user, password);
|
441
|
|
442
|
// If a SQLWarning object is available, print its warning(s).
|
443
|
// There may be multiple warnings chained.
|
444
|
warn = conn.getWarnings();
|
445
|
|
446
|
if (warn != null) {
|
447
|
while (warn != null) {
|
448
|
System.out.println("SQLState: " + warn.getSQLState());
|
449
|
System.out.println("Message: " + warn.getMessage());
|
450
|
System.out.println("Vendor: " + warn.getErrorCode());
|
451
|
System.out.println("");
|
452
|
warn = warn.getNextWarning();
|
453
|
}
|
454
|
}
|
455
|
}
|
456
|
catch (SQLException e) {
|
457
|
System.out.println("Database access failed " + e);
|
458
|
System.exit(1);
|
459
|
}
|
460
|
}
|
461
|
|
462
|
return conn;
|
463
|
}
|
464
|
|
465
|
|
466
|
/**
|
467
|
* Gets the current value of the detailLogID for storage as a primary key in
|
468
|
* the DETAIL_LOG_ID field of the HARVEST_DETAIL_LOG table.
|
469
|
*
|
470
|
* @return the current value of the detailLogID
|
471
|
*/
|
472
|
public int getDetailLogID() {
|
473
|
int currentValue = detailLogID;
|
474
|
|
475
|
detailLogID++;
|
476
|
return currentValue;
|
477
|
}
|
478
|
|
479
|
|
480
|
/**
|
481
|
* Gets the current value of the harvestLogID for storage as a primary key in
|
482
|
* the HARVEST_LOG_ID field of the HARVEST_LOG table.
|
483
|
*
|
484
|
* @return the current value of the detailLogID
|
485
|
*/
|
486
|
public int getHarvestLogID() {
|
487
|
int currentValue = harvestLogID;
|
488
|
|
489
|
harvestLogID++;
|
490
|
return currentValue;
|
491
|
}
|
492
|
|
493
|
|
494
|
/**
|
495
|
* Gets the maximum value of an integer field from a table.
|
496
|
*
|
497
|
* @param tableName the database table name
|
498
|
* @param fieldName the field name of the integer field in the table
|
499
|
* @return the maximum integer stored in the fieldName field of tableName
|
500
|
*/
|
501
|
private int getMaxValue(String tableName, String fieldName) {
|
502
|
int maxValue = 0;
|
503
|
int fieldValue;
|
504
|
String query = "SELECT " + fieldName + " FROM " + tableName;
|
505
|
Statement stmt;
|
506
|
|
507
|
try {
|
508
|
stmt = conn.createStatement();
|
509
|
ResultSet rs = stmt.executeQuery(query);
|
510
|
|
511
|
while (rs.next()) {
|
512
|
fieldValue = rs.getInt(fieldName);
|
513
|
maxValue = Math.max(maxValue, fieldValue);
|
514
|
}
|
515
|
|
516
|
stmt.close();
|
517
|
}
|
518
|
catch(SQLException ex) {
|
519
|
System.out.println("SQLException: " + ex.getMessage());
|
520
|
}
|
521
|
|
522
|
return maxValue;
|
523
|
}
|
524
|
|
525
|
|
526
|
/**
|
527
|
* Gets the minimum value of an integer field from a table.
|
528
|
*
|
529
|
* @param tableName the database table name
|
530
|
* @param fieldName the field name of the integer field in the table
|
531
|
* @return the minimum integer stored in the fieldName field of tableName
|
532
|
*/
|
533
|
private int getMinValue(String tableName, String fieldName) {
|
534
|
int minValue = 0;
|
535
|
int fieldValue;
|
536
|
String query = "SELECT " + fieldName + " FROM " + tableName;
|
537
|
Statement stmt;
|
538
|
|
539
|
try {
|
540
|
stmt = conn.createStatement();
|
541
|
ResultSet rs = stmt.executeQuery(query);
|
542
|
|
543
|
while (rs.next()) {
|
544
|
fieldValue = rs.getInt(fieldName);
|
545
|
|
546
|
if (minValue == 0) {
|
547
|
minValue = fieldValue;
|
548
|
}
|
549
|
else {
|
550
|
minValue = Math.min(minValue, fieldValue);
|
551
|
}
|
552
|
}
|
553
|
|
554
|
stmt.close();
|
555
|
}
|
556
|
catch(SQLException ex) {
|
557
|
System.out.println("SQLException: " + ex.getMessage());
|
558
|
}
|
559
|
|
560
|
return minValue;
|
561
|
}
|
562
|
|
563
|
|
564
|
/**
|
565
|
* For every Harvest site schedule in the database, harvest the
|
566
|
* documents for that site if they are due to be harvested.
|
567
|
*
|
568
|
* @throws SAXException
|
569
|
* @throws IOException
|
570
|
* @throws ParserConfigurationException
|
571
|
*/
|
572
|
private void harvest() {
|
573
|
HarvestSiteSchedule harvestSiteSchedule;
|
574
|
|
575
|
for (int i = 0; i < harvestSiteScheduleList.size(); i++) {
|
576
|
harvestSiteSchedule = (HarvestSiteSchedule)harvestSiteScheduleList.get(i);
|
577
|
|
578
|
if (Harvester.schemaLocation != null) {
|
579
|
harvestSiteSchedule.setSchemaLocation(Harvester.schemaLocation);
|
580
|
}
|
581
|
|
582
|
harvestSiteSchedule.harvestDocumentList();
|
583
|
}
|
584
|
}
|
585
|
|
586
|
|
587
|
/**
|
588
|
* Initializes the detailLogID and harvestLogID values to their current
|
589
|
* maximums + 1.
|
590
|
*/
|
591
|
public void initLogIDs() {
|
592
|
detailLogID = getMaxValue("HARVEST_DETAIL_LOG", "DETAIL_LOG_ID") + 1;
|
593
|
harvestLogID = getMaxValue("HARVEST_LOG", "HARVEST_LOG_ID") + 1;
|
594
|
}
|
595
|
|
596
|
|
597
|
/**
|
598
|
* Prints the header of the harvest report.
|
599
|
*
|
600
|
* @param out the PrintStream object to print to
|
601
|
* @param siteScheduleID the siteScheduleId of the HarvestSiteSchedule. Will
|
602
|
* have a value of 0 if no particular site is involved,
|
603
|
* which indicates that the report is being prepared
|
604
|
* for the Harvester Administrator rather than for a
|
605
|
* particular Site Contact.
|
606
|
*/
|
607
|
void printHarvestHeader(PrintStream out, int siteScheduleID) {
|
608
|
HarvestLog harvestLog;
|
609
|
int logSiteScheduleID;
|
610
|
int nErrors = 0;
|
611
|
String phrase;
|
612
|
|
613
|
for (int i = 0; i < harvestLogList.size(); i++) {
|
614
|
harvestLog = (HarvestLog) harvestLogList.get(i);
|
615
|
logSiteScheduleID = harvestLog.getSiteScheduleID();
|
616
|
|
617
|
if ((siteScheduleID == 0) || (siteScheduleID == logSiteScheduleID)) {
|
618
|
if (harvestLog.isErrorEntry()) {
|
619
|
nErrors++;
|
620
|
}
|
621
|
}
|
622
|
}
|
623
|
|
624
|
out.println(marker);
|
625
|
out.println(filler);
|
626
|
out.println("* METACAT HARVESTER REPORT: " + timestamp);
|
627
|
out.println(filler);
|
628
|
|
629
|
if (nErrors > 0) {
|
630
|
phrase = (nErrors == 1) ? " ERROR WAS " : " ERRORS WERE ";
|
631
|
out.println("* A TOTAL OF " + nErrors + phrase + "DETECTED.");
|
632
|
out.println("* Please see the log entries below for additonal details.");
|
633
|
}
|
634
|
else {
|
635
|
out.println("* NO ERRORS WERE DETECTED DURING THIS HARVEST.");
|
636
|
}
|
637
|
|
638
|
out.println(filler);
|
639
|
out.println(marker);
|
640
|
}
|
641
|
|
642
|
|
643
|
/**
|
644
|
* Prints harvest log entries for this harvest run. Entries may be filtered
|
645
|
* for a particular site, or all entries may be printed.
|
646
|
*
|
647
|
* @param out the PrintStream object to write to
|
648
|
* @param maxCodeLevel the maximum code level that should be printed,
|
649
|
* e.g. "warning". Any log entries higher than this
|
650
|
* level will not be printed.
|
651
|
* @param siteScheduleID if greater than 0, indicates that the log
|
652
|
* entry should only be printed for a particular site
|
653
|
* as identified by its siteScheduleID. if 0, then
|
654
|
* print output for all sites.
|
655
|
*/
|
656
|
void printHarvestLog(PrintStream out, String maxCodeLevel, int siteScheduleID
|
657
|
) {
|
658
|
HarvestLog harvestLog;
|
659
|
int logSiteScheduleID;
|
660
|
int nErrors = 0;
|
661
|
String phrase;
|
662
|
|
663
|
out.println("");
|
664
|
out.println(marker);
|
665
|
out.println(filler);
|
666
|
out.println("* LOG ENTRIES");
|
667
|
out.println(filler);
|
668
|
out.println(marker);
|
669
|
|
670
|
for (int i = 0; i < harvestLogList.size(); i++) {
|
671
|
harvestLog = (HarvestLog) harvestLogList.get(i);
|
672
|
logSiteScheduleID = harvestLog.getSiteScheduleID();
|
673
|
if ((siteScheduleID == 0) || (siteScheduleID == logSiteScheduleID)) {
|
674
|
harvestLog.printOutput(out, maxCodeLevel);
|
675
|
}
|
676
|
}
|
677
|
}
|
678
|
|
679
|
|
680
|
/**
|
681
|
* Prints the site schedule data for a given site.
|
682
|
*
|
683
|
* @param out the PrintStream to write to
|
684
|
* @param siteScheduleID the primary key in the HARVEST_SITE_SCHEDULE table
|
685
|
*/
|
686
|
void printHarvestSiteSchedule(PrintStream out, int siteScheduleID) {
|
687
|
HarvestSiteSchedule harvestSiteSchedule;
|
688
|
|
689
|
for (int i = 0; i < harvestSiteScheduleList.size(); i++) {
|
690
|
harvestSiteSchedule = (HarvestSiteSchedule)harvestSiteScheduleList.get(i);
|
691
|
if (harvestSiteSchedule.siteScheduleID == siteScheduleID) {
|
692
|
harvestSiteSchedule.printOutput(out);
|
693
|
}
|
694
|
}
|
695
|
}
|
696
|
|
697
|
|
698
|
/**
|
699
|
* Prunes old records from the HARVEST_LOG table. Records are removed if
|
700
|
* their HARVEST_DATE is older than a given number of days, as stored in the
|
701
|
* logPeriod object field. First deletes records from the HARVEST_DETAIL_LOG
|
702
|
* table that reference the to-be-pruned entries in the HARVEST_LOG table.
|
703
|
*/
|
704
|
private void pruneHarvestLog() {
|
705
|
long currentTime = harvestStartTime.getTime(); // time in milliseconds
|
706
|
Date dateLastLog; // Prune everything prior to this date
|
707
|
String deleteString;
|
708
|
String deleteStringDetailLog;
|
709
|
long delta;
|
710
|
final long millisecondsPerDay = (1000 * 60 * 60 * 24);
|
711
|
int recordsDeleted;
|
712
|
int recordsDeletedDetail = 0;
|
713
|
SimpleDateFormat simpleDateFormat = new SimpleDateFormat("dd-MMM-yyyy");
|
714
|
String dateString;
|
715
|
ResultSet rs;
|
716
|
String selectString;
|
717
|
Statement stmt;
|
718
|
long timeLastLog = 0;
|
719
|
SQLWarning warn;
|
720
|
|
721
|
delta = logPeriod * millisecondsPerDay;
|
722
|
deleteString = "DELETE FROM HARVEST_LOG WHERE HARVEST_DATE < ";
|
723
|
selectString="SELECT HARVEST_LOG_ID FROM HARVEST_LOG WHERE HARVEST_DATE < ";
|
724
|
deleteStringDetailLog =
|
725
|
"DELETE FROM HARVEST_DETAIL_LOG WHERE HARVEST_LOG_ID = ";
|
726
|
timeLastLog = currentTime - delta;
|
727
|
dateLastLog = new Date(timeLastLog);
|
728
|
dateString = "'" + simpleDateFormat.format(dateLastLog) + "'";
|
729
|
deleteString += dateString;
|
730
|
selectString += dateString;
|
731
|
|
732
|
try {
|
733
|
System.out.println(
|
734
|
"Pruning log entries from HARVEST_DETAIL_LOG and HARVEST_LOG:");
|
735
|
|
736
|
/* Get the list of entries that need to be pruned from the HARVEST_LOG
|
737
|
* table.
|
738
|
*/
|
739
|
stmt = conn.createStatement();
|
740
|
rs = stmt.executeQuery(selectString);
|
741
|
warn = rs.getWarnings();
|
742
|
|
743
|
if (warn != null) {
|
744
|
System.out.println("\n---Warning---\n");
|
745
|
|
746
|
while (warn != null) {
|
747
|
System.out.println("Message: " + warn.getMessage());
|
748
|
System.out.println("SQLState: " + warn.getSQLState());
|
749
|
System.out.print("Vendor error code: ");
|
750
|
System.out.println(warn.getErrorCode());
|
751
|
System.out.println("");
|
752
|
warn = warn.getNextWarning();
|
753
|
}
|
754
|
}
|
755
|
|
756
|
/* Delete any entries from the HARVEST_DETAIL_LOG which reference
|
757
|
* HARVEST_LOG_IDs that are about to be pruned. HARVEST_DETAIL_LOG must
|
758
|
* be pruned first because its records have a child relationship to those
|
759
|
* in HARVEST_LOG.
|
760
|
*/
|
761
|
while (rs.next()) {
|
762
|
harvestLogID = rs.getInt("HARVEST_LOG_ID");
|
763
|
stmt = conn.createStatement();
|
764
|
recordsDeleted = stmt.executeUpdate(deleteStringDetailLog +
|
765
|
harvestLogID);
|
766
|
recordsDeletedDetail += recordsDeleted;
|
767
|
stmt.close();
|
768
|
}
|
769
|
|
770
|
/* Now prune entries from the HARVEST_LOG table using a single update.
|
771
|
*/
|
772
|
stmt = conn.createStatement();
|
773
|
recordsDeleted = stmt.executeUpdate(deleteString);
|
774
|
stmt.close();
|
775
|
|
776
|
System.out.println(" " + recordsDeletedDetail +
|
777
|
" records deleted from HARVEST_DETAIL_LOG");
|
778
|
System.out.println(" " + recordsDeleted +
|
779
|
" records deleted from HARVEST_LOG");
|
780
|
}
|
781
|
catch (SQLException e) {
|
782
|
System.out.println("SQLException: " + e.getMessage());
|
783
|
}
|
784
|
}
|
785
|
|
786
|
|
787
|
/**
|
788
|
* Reads the HARVEST_SITE_SCHEDULE table in the database, creating
|
789
|
* a HarvestSiteSchedule object for each row in the table.
|
790
|
*/
|
791
|
private void readHarvestSiteSchedule() {
|
792
|
HarvestSiteSchedule harvestSiteSchedule;
|
793
|
ResultSet rs;
|
794
|
SQLWarning warn;
|
795
|
Statement stmt;
|
796
|
|
797
|
String contactEmail;
|
798
|
String dateLastHarvest;
|
799
|
String dateNextHarvest;
|
800
|
String documentListURL;
|
801
|
String ldapDN;
|
802
|
String ldapPwd;
|
803
|
int siteScheduleID;
|
804
|
String unit;
|
805
|
int updateFrequency;
|
806
|
|
807
|
try {
|
808
|
// Read the HARVEST_SITE_SCHEDULE table
|
809
|
stmt = conn.createStatement();
|
810
|
rs = stmt.executeQuery("SELECT * FROM HARVEST_SITE_SCHEDULE");
|
811
|
warn = rs.getWarnings();
|
812
|
|
813
|
if (warn != null) {
|
814
|
System.out.println("\n---Warning---\n");
|
815
|
|
816
|
while (warn != null) {
|
817
|
System.out.println("Message: " + warn.getMessage());
|
818
|
System.out.println("SQLState: " + warn.getSQLState());
|
819
|
System.out.print("Vendor error code: ");
|
820
|
System.out.println(warn.getErrorCode());
|
821
|
System.out.println("");
|
822
|
warn = warn.getNextWarning();
|
823
|
}
|
824
|
}
|
825
|
|
826
|
while (rs.next()) {
|
827
|
siteScheduleID = rs.getInt("SITE_SCHEDULE_ID");
|
828
|
documentListURL = rs.getString("DOCUMENTLISTURL");
|
829
|
ldapDN = rs.getString("LDAPDN");
|
830
|
ldapPwd = rs.getString("LDAPPWD");
|
831
|
dateNextHarvest = rs.getString("DATENEXTHARVEST");
|
832
|
dateLastHarvest = rs.getString("DATELASTHARVEST");
|
833
|
updateFrequency = rs.getInt("UPDATEFREQUENCY");
|
834
|
unit = rs.getString("UNIT");
|
835
|
contactEmail = rs.getString("CONTACT_EMAIL");
|
836
|
|
837
|
warn = rs.getWarnings();
|
838
|
|
839
|
if (warn != null) {
|
840
|
System.out.println("\n---Warning---\n");
|
841
|
|
842
|
while (warn != null) {
|
843
|
System.out.println("Message: " + warn.getMessage());
|
844
|
System.out.println("SQLState: " + warn.getSQLState());
|
845
|
System.out.print("Vendor error code: ");
|
846
|
System.out.println(warn.getErrorCode());
|
847
|
System.out.println("");
|
848
|
warn = warn.getNextWarning();
|
849
|
}
|
850
|
}
|
851
|
|
852
|
harvestSiteSchedule = new HarvestSiteSchedule(this,
|
853
|
siteScheduleID,
|
854
|
documentListURL,
|
855
|
ldapDN,
|
856
|
ldapPwd,
|
857
|
dateNextHarvest,
|
858
|
dateLastHarvest,
|
859
|
updateFrequency,
|
860
|
unit,
|
861
|
contactEmail
|
862
|
);
|
863
|
harvestSiteScheduleList.add(harvestSiteSchedule);
|
864
|
}
|
865
|
|
866
|
rs.close();
|
867
|
stmt.close();
|
868
|
}
|
869
|
catch (SQLException e) {
|
870
|
System.out.println("Database access failed " + e);
|
871
|
System.exit(1);
|
872
|
}
|
873
|
|
874
|
}
|
875
|
|
876
|
|
877
|
/**
|
878
|
* Sends a report to the Harvester Administrator. The report prints each log
|
879
|
* entry pertaining to this harvest run.
|
880
|
*
|
881
|
* @param maxCodeLevel the maximum code level that should be printed,
|
882
|
* e.g. "warning". Any log entries higher than this
|
883
|
* level will not be printed.
|
884
|
*/
|
885
|
void reportToAdministrator(String maxCodeLevel) {
|
886
|
PrintStream body;
|
887
|
String from = harvesterAdministrator;
|
888
|
String[] fromArray;
|
889
|
MailMessage msg;
|
890
|
int siteScheduleID = 0;
|
891
|
String subject = "Report from Metacat Harvester: " + timestamp;
|
892
|
String to = harvesterAdministrator;
|
893
|
|
894
|
if (!to.equals("")) {
|
895
|
System.out.println("Sending report to Harvester Administrator at address "
|
896
|
+ harvesterAdministrator);
|
897
|
|
898
|
try {
|
899
|
msg = new MailMessage(smtpServer);
|
900
|
|
901
|
if (from.indexOf(',') > 0) {
|
902
|
fromArray = from.split(",");
|
903
|
|
904
|
for (int i = 0; i < fromArray.length; i++) {
|
905
|
if (i == 0) {
|
906
|
msg.from(fromArray[i]);
|
907
|
}
|
908
|
|
909
|
msg.to(fromArray[i]);
|
910
|
}
|
911
|
}
|
912
|
else if (from.indexOf(';') > 0) {
|
913
|
fromArray = from.split(";");
|
914
|
|
915
|
for (int i = 0; i < fromArray.length; i++) {
|
916
|
if (i == 0) {
|
917
|
msg.from(fromArray[i]);
|
918
|
}
|
919
|
|
920
|
msg.to(fromArray[i]);
|
921
|
}
|
922
|
}
|
923
|
else {
|
924
|
msg.from(from);
|
925
|
msg.to(to);
|
926
|
}
|
927
|
|
928
|
msg.setSubject(subject);
|
929
|
body = msg.getPrintStream();
|
930
|
printHarvestHeader(body, siteScheduleID);
|
931
|
printHarvestLog(body, maxCodeLevel, siteScheduleID);
|
932
|
msg.sendAndClose();
|
933
|
}
|
934
|
catch (IOException e) {
|
935
|
System.out.println("There was a problem sending email to " + to);
|
936
|
System.out.println("IOException: " + e.getMessage());
|
937
|
}
|
938
|
}
|
939
|
}
|
940
|
|
941
|
|
942
|
/**
|
943
|
* Sets the harvest start time for this harvest run.
|
944
|
*
|
945
|
* @param date
|
946
|
*/
|
947
|
public void setHarvestStartTime(Date date) {
|
948
|
harvestStartTime = date;
|
949
|
}
|
950
|
|
951
|
|
952
|
/**
|
953
|
* Shuts down Harvester. Performs cleanup operations such as logging out
|
954
|
* of Metacat and disconnecting from the database.
|
955
|
*/
|
956
|
private void shutdown() {
|
957
|
String maxCodeLevel = "debug"; // Print all log entries from level 1
|
958
|
// ("error") to level 5 ("debug")
|
959
|
int siteScheduleID = 0;
|
960
|
|
961
|
// Log shutdown operation
|
962
|
System.out.println("Shutting Down Harvester");
|
963
|
addLogEntry(0, "Shutting Down Harvester", "harvester.HarvesterShutdown", 0, null, "");
|
964
|
pruneHarvestLog();
|
965
|
closeConnection();
|
966
|
// Print log to standard output and then email the Harvester administrator
|
967
|
printHarvestLog(System.out, maxCodeLevel, siteScheduleID);
|
968
|
reportToAdministrator(maxCodeLevel); // Send a copy to harvester admin
|
969
|
}
|
970
|
|
971
|
|
972
|
/**
|
973
|
* Initializes Harvester at startup. Connects to the database and to Metacat.
|
974
|
*
|
975
|
* @param nHarvests the nth harvest
|
976
|
* @param maxHarvests the maximum number of harvests that this process
|
977
|
* can run
|
978
|
*/
|
979
|
private void startup(int nHarvests, int maxHarvests) {
|
980
|
Boolean ctm;
|
981
|
Integer lp;
|
982
|
String metacatURL;
|
983
|
Date now = new Date();
|
984
|
|
985
|
timestamp = now.toString();
|
986
|
System.out.println(Harvester.marker);
|
987
|
System.out.print(timestamp + ": Starting Next Harvest");
|
988
|
if (maxHarvests > 0) {
|
989
|
System.out.print(" (" + nHarvests + "/" + maxHarvests + ")");
|
990
|
}
|
991
|
System.out.print("\n");
|
992
|
try {
|
993
|
ctm = Boolean.valueOf(PropertyService.getProperty("harvester.connectToMetacat"));
|
994
|
connectToMetacat = ctm.booleanValue();
|
995
|
harvesterAdministrator = PropertyService
|
996
|
.getProperty("harvester.administrator");
|
997
|
smtpServer = PropertyService.getProperty("harvester.smtpServer");
|
998
|
|
999
|
lp = Integer.valueOf(PropertyService.getProperty("harvester.logPeriod"));
|
1000
|
logPeriod = lp.intValue();
|
1001
|
} catch (NumberFormatException e) {
|
1002
|
System.err.println("NumberFormatException: Error parsing logPeriod "
|
1003
|
+ logPeriod + e.getMessage());
|
1004
|
System.err.println("Defaulting to logPeriod of 90 days");
|
1005
|
logPeriod = 90;
|
1006
|
} catch (PropertyNotFoundException pnfe) {
|
1007
|
System.out.println("PropertyNotFoundException: Error getting property: "
|
1008
|
+ pnfe.getMessage());
|
1009
|
return;
|
1010
|
}
|
1011
|
|
1012
|
conn = getConnection();
|
1013
|
initLogIDs();
|
1014
|
setHarvestStartTime(now);
|
1015
|
// Log startup operation
|
1016
|
addLogEntry(0, "Starting Up Harvester", "harvester.HarvesterStartup", 0, null, "");
|
1017
|
|
1018
|
if (connectToMetacat()) {
|
1019
|
try {
|
1020
|
metacatURL = SystemUtil.getServletURL();
|
1021
|
System.out.println("Connecting to Metacat: " + metacatURL);
|
1022
|
metacat = MetacatFactory.createMetacatConnection(metacatURL);
|
1023
|
} catch (MetacatInaccessibleException e) {
|
1024
|
System.out.println("Metacat connection failed." + e.getMessage());
|
1025
|
} catch (Exception e) {
|
1026
|
System.out.println("Metacat connection failed." + e.getMessage());
|
1027
|
}
|
1028
|
}
|
1029
|
}
|
1030
|
|
1031
|
}
|