| 1 |
2094
|
jones
|
/**
|
| 2 |
|
|
* '$RCSfile$'
|
| 3 |
|
|
* Copyright: 2004 University of New Mexico and the
|
| 4 |
|
|
* Regents of the University of California
|
| 5 |
2022
|
costa
|
*
|
| 6 |
2094
|
jones
|
* '$Author$'
|
| 7 |
|
|
* '$Date$'
|
| 8 |
|
|
* '$Revision$'
|
| 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 |
2022
|
costa
|
*/
|
| 24 |
|
|
|
| 25 |
|
|
package edu.ucsb.nceas.metacat.harvesterClient;
|
| 26 |
|
|
|
| 27 |
2086
|
costa
|
import com.oreilly.servlet.MailMessage;
|
| 28 |
2031
|
costa
|
import java.io.FileNotFoundException;
|
| 29 |
|
|
import java.io.IOException;
|
| 30 |
|
|
import java.io.InputStream;
|
| 31 |
|
|
import java.io.InputStreamReader;
|
| 32 |
2086
|
costa
|
import java.io.PrintStream;
|
| 33 |
2031
|
costa
|
import java.io.Reader;
|
| 34 |
|
|
import java.net.MalformedURLException;
|
| 35 |
|
|
import java.net.URL;
|
| 36 |
2022
|
costa
|
import java.sql.Connection;
|
| 37 |
|
|
import java.sql.SQLException;
|
| 38 |
|
|
import java.sql.Statement;
|
| 39 |
2031
|
costa
|
import java.text.DateFormat;
|
| 40 |
|
|
import java.text.ParseException;
|
| 41 |
|
|
import java.text.SimpleDateFormat;
|
| 42 |
|
|
import java.util.ArrayList;
|
| 43 |
|
|
import java.util.Date;
|
| 44 |
|
|
import javax.xml.parsers.ParserConfigurationException;
|
| 45 |
|
|
import org.xml.sax.Attributes;
|
| 46 |
|
|
import org.xml.sax.ContentHandler;
|
| 47 |
|
|
import org.xml.sax.ErrorHandler;
|
| 48 |
|
|
import org.xml.sax.InputSource;
|
| 49 |
|
|
import org.xml.sax.SAXException;
|
| 50 |
|
|
import org.xml.sax.SAXParseException;
|
| 51 |
|
|
import org.xml.sax.XMLReader;
|
| 52 |
|
|
import org.xml.sax.helpers.DefaultHandler;
|
| 53 |
|
|
import org.xml.sax.helpers.XMLReaderFactory;
|
| 54 |
2022
|
costa
|
|
| 55 |
2031
|
costa
|
import edu.ucsb.nceas.metacat.client.Metacat;
|
| 56 |
|
|
import edu.ucsb.nceas.metacat.client.MetacatException;
|
| 57 |
|
|
import edu.ucsb.nceas.metacat.client.MetacatInaccessibleException;
|
| 58 |
2022
|
costa
|
|
| 59 |
|
|
|
| 60 |
|
|
/**
|
| 61 |
|
|
* HarvestSiteSchedule manages a single entry in the HARVEST_SITE_SCHEDULE
|
| 62 |
|
|
* table, determining when and how to harvest the documents for a given site.
|
| 63 |
|
|
*
|
| 64 |
|
|
* @author costa
|
| 65 |
|
|
*/
|
| 66 |
2139
|
costa
|
public class HarvestSiteSchedule {
|
| 67 |
2022
|
costa
|
|
| 68 |
|
|
private String contactEmail;
|
| 69 |
|
|
private String dateLastHarvest;
|
| 70 |
|
|
private String dateNextHarvest;
|
| 71 |
|
|
private long delta;
|
| 72 |
|
|
private String documentListURL;
|
| 73 |
|
|
private Harvester harvester;
|
| 74 |
2031
|
costa
|
private ArrayList harvestDocumentList = new ArrayList();
|
| 75 |
2022
|
costa
|
private String harvestSiteEndTime;
|
| 76 |
|
|
private String harvestSiteStartTime;
|
| 77 |
|
|
private String ldapDN;
|
| 78 |
2031
|
costa
|
private String ldapPwd;
|
| 79 |
2022
|
costa
|
final private long millisecondsPerDay = (1000 * 60 * 60 * 24);
|
| 80 |
2139
|
costa
|
private String schemaLocation =
|
| 81 |
|
|
"eml://ecoinformatics.org/harvestList ../../lib/harvester/harvestList.xsd";
|
| 82 |
2031
|
costa
|
int siteScheduleID;
|
| 83 |
2022
|
costa
|
private String unit;
|
| 84 |
|
|
private int updateFrequency;
|
| 85 |
|
|
|
| 86 |
|
|
/**
|
| 87 |
|
|
* Creates a new instance of HarvestSiteSchedule. Initialized with the data
|
| 88 |
|
|
* that was read from a single row in the HARVEST_SITE_SCHEDULE table.
|
| 89 |
|
|
*
|
| 90 |
|
|
* @param harvester the parent Harvester object
|
| 91 |
|
|
* @param siteScheduleID the value of the SITE_SCHEDULE_ID field
|
| 92 |
|
|
* @param documentListURL the value of the DOCUMENTLISTURL field
|
| 93 |
|
|
* @param ldapDN the value of the LDAPDN field
|
| 94 |
2031
|
costa
|
* @param ldapPwd the value of the LDAPPASSWORD field
|
| 95 |
2022
|
costa
|
* @param dateNextHarvest the value of the DATENEXTHARVEST field
|
| 96 |
|
|
* @param dateLastHarvest the value of the DATELASTHARVEST field
|
| 97 |
|
|
* @param updateFrequency the value of the UPDATEFREQUENCY field
|
| 98 |
|
|
* @param unit the value of the UNIT field
|
| 99 |
|
|
* @param contactEmail the value of the CONTACT_EMAIL field
|
| 100 |
|
|
*/
|
| 101 |
|
|
public HarvestSiteSchedule(
|
| 102 |
|
|
Harvester harvester,
|
| 103 |
|
|
int siteScheduleID,
|
| 104 |
|
|
String documentListURL,
|
| 105 |
|
|
String ldapDN,
|
| 106 |
2031
|
costa
|
String ldapPwd,
|
| 107 |
2022
|
costa
|
String dateNextHarvest,
|
| 108 |
|
|
String dateLastHarvest,
|
| 109 |
|
|
int updateFrequency,
|
| 110 |
|
|
String unit,
|
| 111 |
|
|
String contactEmail
|
| 112 |
|
|
)
|
| 113 |
|
|
{
|
| 114 |
|
|
this.harvester = harvester;
|
| 115 |
|
|
this.siteScheduleID = siteScheduleID;
|
| 116 |
|
|
this.documentListURL = documentListURL;
|
| 117 |
|
|
this.ldapDN = ldapDN;
|
| 118 |
2031
|
costa
|
this.ldapPwd = ldapPwd;
|
| 119 |
2022
|
costa
|
this.dateNextHarvest = dateNextHarvest;
|
| 120 |
|
|
this.dateLastHarvest = dateLastHarvest;
|
| 121 |
|
|
this.updateFrequency = updateFrequency;
|
| 122 |
|
|
this.unit = unit;
|
| 123 |
|
|
this.contactEmail = contactEmail;
|
| 124 |
|
|
|
| 125 |
|
|
// Calculate the value of delta, the number of milliseconds between the
|
| 126 |
|
|
// last harvest date and the next harvest date.
|
| 127 |
|
|
delta = updateFrequency * millisecondsPerDay;
|
| 128 |
|
|
|
| 129 |
|
|
if (unit.equals("weeks")) {
|
| 130 |
|
|
delta *= 7;
|
| 131 |
|
|
}
|
| 132 |
|
|
else if (unit.equals("months")) {
|
| 133 |
|
|
delta *= 30;
|
| 134 |
|
|
}
|
| 135 |
|
|
}
|
| 136 |
|
|
|
| 137 |
|
|
|
| 138 |
|
|
/**
|
| 139 |
2060
|
costa
|
* Updates the DATELASTHARVEST and DATENEXTHARVEST values of the
|
| 140 |
|
|
* HARVEST_SITE_SCHEDULE table after a harvest operation has completed.
|
| 141 |
|
|
* Calculates the date of the next harvest based on today's date and the
|
| 142 |
|
|
* update frequency.
|
| 143 |
2022
|
costa
|
*/
|
| 144 |
2060
|
costa
|
private void dbUpdateHarvestDates() {
|
| 145 |
2380
|
costa
|
Connection conn;
|
| 146 |
2022
|
costa
|
long currentTime; // Current time in milliseconds
|
| 147 |
|
|
Date dateNextHarvest; // Date of next harvest
|
| 148 |
|
|
String lastHarvest;
|
| 149 |
|
|
String nextHarvest;
|
| 150 |
|
|
Date now = new Date();
|
| 151 |
|
|
SimpleDateFormat simpleDateFormat = new SimpleDateFormat("dd-MMM-yyyy");
|
| 152 |
2380
|
costa
|
Statement stmt;
|
| 153 |
2022
|
costa
|
long timeNextHarvest;
|
| 154 |
|
|
|
| 155 |
2139
|
costa
|
conn = harvester.getConnection();
|
| 156 |
2022
|
costa
|
now = new Date();
|
| 157 |
2031
|
costa
|
currentTime = now.getTime();
|
| 158 |
2022
|
costa
|
timeNextHarvest = currentTime + delta;
|
| 159 |
|
|
dateNextHarvest = new Date(timeNextHarvest);
|
| 160 |
|
|
nextHarvest = "'" + simpleDateFormat.format(dateNextHarvest) + "'";
|
| 161 |
|
|
lastHarvest = "'" + simpleDateFormat.format(now) + "'";
|
| 162 |
|
|
|
| 163 |
2380
|
costa
|
try {
|
| 164 |
|
|
stmt = conn.createStatement();
|
| 165 |
|
|
stmt.executeUpdate(
|
| 166 |
2367
|
costa
|
"UPDATE HARVEST_SITE_SCHEDULE SET DATENEXTHARVEST = " +
|
| 167 |
2031
|
costa
|
nextHarvest +
|
| 168 |
|
|
" WHERE SITE_SCHEDULE_ID = " +
|
| 169 |
|
|
siteScheduleID);
|
| 170 |
2380
|
costa
|
stmt.executeUpdate(
|
| 171 |
2367
|
costa
|
"UPDATE HARVEST_SITE_SCHEDULE SET DATELASTHARVEST = " +
|
| 172 |
2031
|
costa
|
lastHarvest +
|
| 173 |
|
|
" WHERE SITE_SCHEDULE_ID = " +
|
| 174 |
|
|
siteScheduleID);
|
| 175 |
2380
|
costa
|
stmt.close();
|
| 176 |
|
|
}
|
| 177 |
2022
|
costa
|
catch(SQLException e) {
|
| 178 |
2380
|
costa
|
System.out.println("SQLException: " + e.getMessage());
|
| 179 |
|
|
}
|
| 180 |
2022
|
costa
|
}
|
| 181 |
|
|
|
| 182 |
|
|
|
| 183 |
|
|
/**
|
| 184 |
|
|
* Boolean to determine whether this site is currently due for its next
|
| 185 |
|
|
* harvest.
|
| 186 |
|
|
*
|
| 187 |
|
|
* @retrun true if due for harvest, otherwise false
|
| 188 |
|
|
*/
|
| 189 |
2139
|
costa
|
public boolean dueForHarvest() {
|
| 190 |
2022
|
costa
|
boolean dueForHarvest = false;
|
| 191 |
2303
|
costa
|
// DateFormat dateFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss.S");
|
| 192 |
|
|
DateFormat dateFormat = new SimpleDateFormat("yyyy-MM-dd");
|
| 193 |
2022
|
costa
|
Date now = new Date();
|
| 194 |
|
|
Date dnh; // Date of next harvest
|
| 195 |
|
|
long currentTime = now.getTime(); // Current time in milliseconds
|
| 196 |
|
|
long timeNextHarvest = 0;
|
| 197 |
|
|
|
| 198 |
|
|
try {
|
| 199 |
2036
|
costa
|
dnh = dateFormat.parse(dateNextHarvest);
|
| 200 |
|
|
timeNextHarvest = dnh.getTime();
|
| 201 |
2022
|
costa
|
|
| 202 |
|
|
if (timeNextHarvest < currentTime) {
|
| 203 |
|
|
dueForHarvest = true;
|
| 204 |
2031
|
costa
|
System.out.println("Due for harvest: " + documentListURL);
|
| 205 |
2022
|
costa
|
}
|
| 206 |
|
|
else {
|
| 207 |
2031
|
costa
|
System.out.println("Not due for harvest: " + documentListURL);
|
| 208 |
2022
|
costa
|
}
|
| 209 |
|
|
}
|
| 210 |
|
|
catch (ParseException e) {
|
| 211 |
2031
|
costa
|
System.out.println("Error parsing date: " + e.getMessage());
|
| 212 |
2022
|
costa
|
}
|
| 213 |
|
|
|
| 214 |
2031
|
costa
|
return dueForHarvest;
|
| 215 |
2022
|
costa
|
}
|
| 216 |
2139
|
costa
|
|
| 217 |
2022
|
costa
|
|
| 218 |
2139
|
costa
|
/**
|
| 219 |
|
|
* Accessor method for the schemaLocation field.
|
| 220 |
|
|
*
|
| 221 |
|
|
* @return schemaLocation the schema location string
|
| 222 |
|
|
*/
|
| 223 |
|
|
public String getSchemaLocation() {
|
| 224 |
|
|
return schemaLocation;
|
| 225 |
|
|
}
|
| 226 |
2022
|
costa
|
|
| 227 |
2139
|
costa
|
|
| 228 |
2022
|
costa
|
/**
|
| 229 |
|
|
* Harvests each document in the site document list.
|
| 230 |
|
|
*
|
| 231 |
|
|
* @throws SAXException
|
| 232 |
|
|
* @throws IOException
|
| 233 |
|
|
* @throws ParserConfigurationException
|
| 234 |
|
|
*/
|
| 235 |
|
|
public void harvestDocumentList() {
|
| 236 |
|
|
HarvestDocument harvestDocument;
|
| 237 |
2036
|
costa
|
boolean success;
|
| 238 |
2022
|
costa
|
|
| 239 |
|
|
if (dueForHarvest()) {
|
| 240 |
|
|
try {
|
| 241 |
2139
|
costa
|
success = parseHarvestList();
|
| 242 |
2036
|
costa
|
|
| 243 |
|
|
/* If the document list was validated, then proceed with harvesting
|
| 244 |
|
|
* the documents
|
| 245 |
|
|
*/
|
| 246 |
|
|
if (success) {
|
| 247 |
|
|
metacatLogin();
|
| 248 |
2022
|
costa
|
|
| 249 |
2036
|
costa
|
for (int i = 0; i < harvestDocumentList.size(); i++) {
|
| 250 |
|
|
harvestDocument = (HarvestDocument) harvestDocumentList.get(i);
|
| 251 |
2022
|
costa
|
|
| 252 |
2036
|
costa
|
if (harvestDocument != null) {
|
| 253 |
|
|
harvestDocument.harvestDocument();
|
| 254 |
|
|
}
|
| 255 |
2022
|
costa
|
}
|
| 256 |
2036
|
costa
|
|
| 257 |
|
|
metacatLogout();
|
| 258 |
2060
|
costa
|
dbUpdateHarvestDates(); // Update the schedule
|
| 259 |
2022
|
costa
|
}
|
| 260 |
|
|
}
|
| 261 |
|
|
catch (ParserConfigurationException e) {
|
| 262 |
2031
|
costa
|
System.out.println("ParserConfigurationException: " + e.getMessage());
|
| 263 |
2022
|
costa
|
}
|
| 264 |
|
|
|
| 265 |
2155
|
costa
|
reportToSiteContact();
|
| 266 |
2022
|
costa
|
}
|
| 267 |
|
|
}
|
| 268 |
|
|
|
| 269 |
|
|
|
| 270 |
|
|
/**
|
| 271 |
2031
|
costa
|
* Login to Metacat using the ldapDN and ldapPwd
|
| 272 |
2022
|
costa
|
*/
|
| 273 |
2139
|
costa
|
public void metacatLogin() {
|
| 274 |
2022
|
costa
|
Metacat metacat = harvester.metacat;
|
| 275 |
2139
|
costa
|
String response;
|
| 276 |
2022
|
costa
|
|
| 277 |
|
|
if (harvester.connectToMetacat()) {
|
| 278 |
|
|
try {
|
| 279 |
|
|
System.out.println("Logging in to Metacat: " + ldapDN);
|
| 280 |
2139
|
costa
|
response = metacat.login(ldapDN, ldapPwd);
|
| 281 |
2022
|
costa
|
//System.out.println("Metacat login response: " + response);
|
| 282 |
|
|
}
|
| 283 |
|
|
catch (MetacatInaccessibleException e) {
|
| 284 |
|
|
System.out.println("Metacat login failed." + e.getMessage());
|
| 285 |
|
|
}
|
| 286 |
|
|
catch (Exception e) {
|
| 287 |
|
|
System.out.println("Metacat login failed." + e.getMessage());
|
| 288 |
|
|
}
|
| 289 |
2031
|
costa
|
}
|
| 290 |
2022
|
costa
|
}
|
| 291 |
|
|
|
| 292 |
|
|
|
| 293 |
|
|
/**
|
| 294 |
|
|
* Logout from Metacat
|
| 295 |
|
|
*/
|
| 296 |
|
|
private void metacatLogout() {
|
| 297 |
|
|
Metacat metacat = harvester.metacat;
|
| 298 |
|
|
|
| 299 |
|
|
if (harvester.connectToMetacat()) {
|
| 300 |
|
|
try {
|
| 301 |
|
|
// Log out from the Metacat session
|
| 302 |
|
|
System.out.println("Logging out from Metacat");
|
| 303 |
|
|
metacat.logout();
|
| 304 |
|
|
}
|
| 305 |
|
|
catch (MetacatInaccessibleException e) {
|
| 306 |
|
|
System.out.println("Metacat inaccessible: " + e.getMessage());
|
| 307 |
|
|
}
|
| 308 |
|
|
catch (MetacatException e) {
|
| 309 |
|
|
System.out.println("Metacat exception: " + e.getMessage());
|
| 310 |
|
|
}
|
| 311 |
|
|
}
|
| 312 |
|
|
}
|
| 313 |
|
|
|
| 314 |
|
|
|
| 315 |
|
|
/**
|
| 316 |
2139
|
costa
|
* Parses the site harvest list XML file to find out which documents to
|
| 317 |
|
|
* harvest.
|
| 318 |
2036
|
costa
|
*
|
| 319 |
|
|
* @return true if successful, otherwise false
|
| 320 |
2022
|
costa
|
*/
|
| 321 |
2139
|
costa
|
public boolean parseHarvestList()
|
| 322 |
2031
|
costa
|
throws ParserConfigurationException {
|
| 323 |
|
|
DocumentListHandler documentListHandler = new DocumentListHandler();
|
| 324 |
|
|
InputStream inputStream;
|
| 325 |
|
|
InputStreamReader inputStreamReader;
|
| 326 |
2139
|
costa
|
String schemaLocation = getSchemaLocation();
|
| 327 |
2036
|
costa
|
boolean success = false;
|
| 328 |
2031
|
costa
|
URL url;
|
| 329 |
|
|
|
| 330 |
|
|
try {
|
| 331 |
|
|
url = new URL(documentListURL);
|
| 332 |
|
|
inputStream = url.openStream();
|
| 333 |
2060
|
costa
|
harvester.addLogEntry(0,
|
| 334 |
|
|
"Retrieved: " + documentListURL,
|
| 335 |
2139
|
costa
|
"GetHarvestListSuccess",
|
| 336 |
2060
|
costa
|
siteScheduleID,
|
| 337 |
|
|
null,
|
| 338 |
|
|
"");
|
| 339 |
2031
|
costa
|
inputStreamReader = new InputStreamReader(inputStream);
|
| 340 |
|
|
documentListHandler.runParser(inputStreamReader, schemaLocation);
|
| 341 |
2060
|
costa
|
harvester.addLogEntry(0,
|
| 342 |
|
|
"Validated: " + documentListURL,
|
| 343 |
2139
|
costa
|
"ValidateHarvestListSuccess",
|
| 344 |
2060
|
costa
|
siteScheduleID,
|
| 345 |
|
|
null,
|
| 346 |
|
|
"");
|
| 347 |
2036
|
costa
|
success = true;
|
| 348 |
2031
|
costa
|
}
|
| 349 |
|
|
catch (MalformedURLException e){
|
| 350 |
|
|
harvester.addLogEntry(1, "MalformedURLException: " + e.getMessage(),
|
| 351 |
2139
|
costa
|
"GetHarvestListError", siteScheduleID, null, "");
|
| 352 |
2031
|
costa
|
}
|
| 353 |
|
|
catch (FileNotFoundException e) {
|
| 354 |
|
|
harvester.addLogEntry(1, "FileNotFoundException: " + e.getMessage(),
|
| 355 |
2139
|
costa
|
"GetHarvestListError", siteScheduleID, null, "");
|
| 356 |
2031
|
costa
|
}
|
| 357 |
|
|
catch (SAXException e) {
|
| 358 |
|
|
harvester.addLogEntry(1, "SAXException: " + e.getMessage(),
|
| 359 |
2367
|
costa
|
"ValidateHarvestListError", siteScheduleID, null, "");
|
| 360 |
2031
|
costa
|
}
|
| 361 |
|
|
catch (ClassNotFoundException e) {
|
| 362 |
|
|
harvester.addLogEntry(1, "ClassNotFoundException: " + e.getMessage(),
|
| 363 |
2367
|
costa
|
"ValidateHarvestListError", siteScheduleID, null, "");
|
| 364 |
2031
|
costa
|
}
|
| 365 |
|
|
catch (IOException e) {
|
| 366 |
|
|
harvester.addLogEntry(1, "IOException: " + e.getMessage(),
|
| 367 |
2139
|
costa
|
"GetHarvestListError", siteScheduleID, null, "");
|
| 368 |
2031
|
costa
|
}
|
| 369 |
2036
|
costa
|
|
| 370 |
|
|
return success;
|
| 371 |
2022
|
costa
|
}
|
| 372 |
|
|
|
| 373 |
|
|
|
| 374 |
|
|
/**
|
| 375 |
|
|
* Prints the data that is stored in this HarvestSiteSchedule object.
|
| 376 |
2086
|
costa
|
*
|
| 377 |
|
|
* @param out the PrintStream to write to
|
| 378 |
2022
|
costa
|
*/
|
| 379 |
2139
|
costa
|
public void printOutput(PrintStream out) {
|
| 380 |
2086
|
costa
|
out.println("* siteScheduleID: " + siteScheduleID);
|
| 381 |
|
|
out.println("* documentListURL: " + documentListURL);
|
| 382 |
|
|
out.println("* ldapDN: " + ldapDN);
|
| 383 |
|
|
out.println("* dateNextHarvest: " + dateNextHarvest);
|
| 384 |
|
|
out.println("* dateLastHarvest: " + dateLastHarvest);
|
| 385 |
|
|
out.println("* updateFrequency: " + updateFrequency);
|
| 386 |
|
|
out.println("* unit: " + unit);
|
| 387 |
|
|
out.println("* contactEmail: " + contactEmail);
|
| 388 |
2022
|
costa
|
}
|
| 389 |
|
|
|
| 390 |
2155
|
costa
|
/**
|
| 391 |
|
|
* Reports a summary of the site harvest. Includes the following:
|
| 392 |
|
|
* A list of documents that were successfully inserted.
|
| 393 |
|
|
* A list of documents that were successfully updated.
|
| 394 |
|
|
* A list of documents that could not be accessed at the site.
|
| 395 |
|
|
* A list of documents that could not be uploaded to Metacat.
|
| 396 |
|
|
* A list of documents that were already found in Metacat.
|
| 397 |
|
|
*
|
| 398 |
|
|
* @param out the PrintStream to write to
|
| 399 |
|
|
*/
|
| 400 |
|
|
void printSiteSummary(PrintStream out) {
|
| 401 |
|
|
HarvestDocument harvestDocument;
|
| 402 |
|
|
int nAccessError = 0;
|
| 403 |
|
|
int nInserted = 0;
|
| 404 |
|
|
int nMetacatHasIt = 0;
|
| 405 |
|
|
int nUpdated = 0;
|
| 406 |
|
|
int nUploadError = 0;
|
| 407 |
|
|
|
| 408 |
|
|
for (int i = 0; i < harvestDocumentList.size(); i++) {
|
| 409 |
|
|
harvestDocument = (HarvestDocument) harvestDocumentList.get(i);
|
| 410 |
|
|
|
| 411 |
|
|
if (harvestDocument != null) {
|
| 412 |
|
|
if (harvestDocument.accessError) { nAccessError++; }
|
| 413 |
|
|
if (harvestDocument.inserted) { nInserted++; }
|
| 414 |
|
|
if (harvestDocument.metacatHasIt) { nMetacatHasIt++; }
|
| 415 |
|
|
if (harvestDocument.updated) { nUpdated++; }
|
| 416 |
|
|
if (harvestDocument.uploadError) { nUploadError++; }
|
| 417 |
|
|
}
|
| 418 |
|
|
}
|
| 419 |
|
|
|
| 420 |
|
|
if (nInserted > 0) {
|
| 421 |
|
|
printSiteSummaryHeader(out);
|
| 422 |
|
|
out.println("* The following document(s) were successfully inserted:");
|
| 423 |
|
|
for (int i = 0; i < harvestDocumentList.size(); i++) {
|
| 424 |
|
|
harvestDocument = (HarvestDocument) harvestDocumentList.get(i);
|
| 425 |
|
|
if (harvestDocument != null) {
|
| 426 |
|
|
if (harvestDocument.inserted) {
|
| 427 |
|
|
harvestDocument.prettyPrint(out);
|
| 428 |
|
|
}
|
| 429 |
|
|
}
|
| 430 |
|
|
}
|
| 431 |
|
|
printSiteSummaryTrailer(out);
|
| 432 |
|
|
}
|
| 433 |
2022
|
costa
|
|
| 434 |
2155
|
costa
|
if (nUpdated > 0) {
|
| 435 |
|
|
printSiteSummaryHeader(out);
|
| 436 |
|
|
out.println("* The following document(s) were successfully updated:");
|
| 437 |
|
|
for (int i = 0; i < harvestDocumentList.size(); i++) {
|
| 438 |
|
|
harvestDocument = (HarvestDocument) harvestDocumentList.get(i);
|
| 439 |
|
|
if (harvestDocument != null) {
|
| 440 |
|
|
if (harvestDocument.updated) {
|
| 441 |
|
|
harvestDocument.prettyPrint(out);
|
| 442 |
|
|
}
|
| 443 |
|
|
}
|
| 444 |
|
|
}
|
| 445 |
|
|
printSiteSummaryTrailer(out);
|
| 446 |
|
|
}
|
| 447 |
|
|
|
| 448 |
|
|
if (nAccessError > 0) {
|
| 449 |
|
|
printSiteSummaryHeader(out);
|
| 450 |
|
|
out.println("* The following document(s) could not be accessed");
|
| 451 |
|
|
out.println("* at the site. Please check the URL to ensure that it is");
|
| 452 |
|
|
out.println("* accessible at the site.");
|
| 453 |
|
|
for (int i = 0; i < harvestDocumentList.size(); i++) {
|
| 454 |
|
|
harvestDocument = (HarvestDocument) harvestDocumentList.get(i);
|
| 455 |
|
|
if (harvestDocument != null) {
|
| 456 |
|
|
if (harvestDocument.accessError) {
|
| 457 |
|
|
harvestDocument.prettyPrint(out);
|
| 458 |
|
|
}
|
| 459 |
|
|
}
|
| 460 |
|
|
}
|
| 461 |
|
|
printSiteSummaryTrailer(out);
|
| 462 |
|
|
}
|
| 463 |
|
|
|
| 464 |
|
|
if (nUploadError > 0) {
|
| 465 |
|
|
printSiteSummaryHeader(out);
|
| 466 |
|
|
out.println("* The following document(s) could not be uploaded to");
|
| 467 |
|
|
out.println("* Metacat because an error of some kind occurred.");
|
| 468 |
2777
|
costa
|
out.println("* (See log entries below for additional details.) :");
|
| 469 |
2155
|
costa
|
for (int i = 0; i < harvestDocumentList.size(); i++) {
|
| 470 |
|
|
harvestDocument = (HarvestDocument) harvestDocumentList.get(i);
|
| 471 |
|
|
if (harvestDocument != null) {
|
| 472 |
|
|
if (harvestDocument.uploadError) {
|
| 473 |
|
|
harvestDocument.prettyPrint(out);
|
| 474 |
|
|
}
|
| 475 |
|
|
}
|
| 476 |
|
|
}
|
| 477 |
|
|
printSiteSummaryTrailer(out);
|
| 478 |
|
|
}
|
| 479 |
|
|
|
| 480 |
|
|
if (nMetacatHasIt > 0) {
|
| 481 |
|
|
printSiteSummaryHeader(out);
|
| 482 |
2777
|
costa
|
out.println("* The following document(s) were already found in Metacat:");
|
| 483 |
|
|
|
| 484 |
2155
|
costa
|
for (int i = 0; i < harvestDocumentList.size(); i++) {
|
| 485 |
|
|
harvestDocument = (HarvestDocument) harvestDocumentList.get(i);
|
| 486 |
|
|
if (harvestDocument != null) {
|
| 487 |
|
|
if (harvestDocument.metacatHasIt) {
|
| 488 |
|
|
harvestDocument.prettyPrint(out);
|
| 489 |
|
|
}
|
| 490 |
|
|
}
|
| 491 |
|
|
}
|
| 492 |
|
|
printSiteSummaryTrailer(out);
|
| 493 |
|
|
}
|
| 494 |
|
|
|
| 495 |
|
|
}
|
| 496 |
|
|
|
| 497 |
|
|
|
| 498 |
2022
|
costa
|
/**
|
| 499 |
2155
|
costa
|
* Prints the header lines of a site summary entry.
|
| 500 |
|
|
*
|
| 501 |
|
|
* @param out the PrintStream to write to
|
| 502 |
2022
|
costa
|
*/
|
| 503 |
2155
|
costa
|
void printSiteSummaryHeader(PrintStream out) {
|
| 504 |
|
|
final String filler = Harvester.filler;
|
| 505 |
|
|
final String marker = Harvester.marker;
|
| 506 |
|
|
|
| 507 |
|
|
out.println("");
|
| 508 |
|
|
out.println(marker);
|
| 509 |
|
|
out.println(filler);
|
| 510 |
|
|
}
|
| 511 |
|
|
|
| 512 |
|
|
|
| 513 |
|
|
/**
|
| 514 |
|
|
* Prints the trailing lines of a site summary entry.
|
| 515 |
|
|
*
|
| 516 |
|
|
* @param out the PrintStream to write to
|
| 517 |
|
|
*/
|
| 518 |
|
|
void printSiteSummaryTrailer(PrintStream out) {
|
| 519 |
|
|
final String filler = Harvester.filler;
|
| 520 |
|
|
final String marker = Harvester.marker;
|
| 521 |
|
|
|
| 522 |
|
|
out.println(filler);
|
| 523 |
|
|
out.println(marker);
|
| 524 |
|
|
}
|
| 525 |
|
|
|
| 526 |
|
|
|
| 527 |
|
|
/**
|
| 528 |
|
|
* Sends a report to the Site Contact summarizing the results of the harvest
|
| 529 |
|
|
* at that site.
|
| 530 |
|
|
*/
|
| 531 |
|
|
void reportToSiteContact() {
|
| 532 |
2086
|
costa
|
PrintStream body;
|
| 533 |
2105
|
costa
|
String from = harvester.harvesterAdministrator;
|
| 534 |
2330
|
costa
|
String[] fromArray;
|
| 535 |
2765
|
costa
|
String maxCodeLevel = "notice";
|
| 536 |
2086
|
costa
|
MailMessage msg;
|
| 537 |
2105
|
costa
|
int nErrors = 0;
|
| 538 |
2108
|
costa
|
String subject = "Report from Metacat Harvester: " + harvester.timestamp;
|
| 539 |
2086
|
costa
|
String to = contactEmail;
|
| 540 |
2330
|
costa
|
String[] toArray;
|
| 541 |
2086
|
costa
|
|
| 542 |
|
|
if (!to.equals("")) {
|
| 543 |
|
|
System.out.println("Sending report to siteScheduleID=" + siteScheduleID +
|
| 544 |
|
|
" at address: " + contactEmail);
|
| 545 |
|
|
try {
|
| 546 |
2105
|
costa
|
msg = new MailMessage(harvester.smtpServer);
|
| 547 |
2330
|
costa
|
|
| 548 |
|
|
if (from.indexOf(',') > 0) {
|
| 549 |
|
|
fromArray = from.split(",");
|
| 550 |
|
|
|
| 551 |
|
|
for (int i = 0; i < fromArray.length; i++) {
|
| 552 |
|
|
if (i == 0) {
|
| 553 |
|
|
msg.from(fromArray[i]);
|
| 554 |
|
|
}
|
| 555 |
|
|
|
| 556 |
|
|
msg.cc(fromArray[i]);
|
| 557 |
|
|
|
| 558 |
|
|
}
|
| 559 |
|
|
}
|
| 560 |
|
|
else if (from.indexOf(';') > 0) {
|
| 561 |
|
|
fromArray = from.split(";");
|
| 562 |
|
|
|
| 563 |
|
|
for (int i = 0; i < fromArray.length; i++) {
|
| 564 |
|
|
if (i == 0) {
|
| 565 |
|
|
msg.from(fromArray[i]);
|
| 566 |
|
|
}
|
| 567 |
|
|
|
| 568 |
|
|
msg.cc(fromArray[i]);
|
| 569 |
|
|
|
| 570 |
|
|
}
|
| 571 |
|
|
}
|
| 572 |
|
|
else {
|
| 573 |
|
|
msg.from(from);
|
| 574 |
|
|
msg.cc(from);
|
| 575 |
|
|
}
|
| 576 |
|
|
|
| 577 |
|
|
if (to.indexOf(',') > 0) {
|
| 578 |
|
|
toArray = to.split(",");
|
| 579 |
|
|
|
| 580 |
|
|
for (int i = 0; i < toArray.length; i++) {
|
| 581 |
|
|
msg.to(toArray[i]);
|
| 582 |
|
|
}
|
| 583 |
|
|
}
|
| 584 |
|
|
else if (to.indexOf(';') > 0) {
|
| 585 |
|
|
toArray = to.split(";");
|
| 586 |
|
|
|
| 587 |
|
|
for (int i = 0; i < toArray.length; i++) {
|
| 588 |
|
|
msg.to(toArray[i]);
|
| 589 |
|
|
}
|
| 590 |
|
|
}
|
| 591 |
|
|
else {
|
| 592 |
|
|
msg.to(to);
|
| 593 |
|
|
}
|
| 594 |
|
|
|
| 595 |
2086
|
costa
|
msg.setSubject(subject);
|
| 596 |
|
|
body = msg.getPrintStream();
|
| 597 |
2155
|
costa
|
harvester.printHarvestHeader(body, siteScheduleID);
|
| 598 |
|
|
printSiteSummary(body);
|
| 599 |
2105
|
costa
|
harvester.printHarvestLog(body, maxCodeLevel, siteScheduleID);
|
| 600 |
|
|
msg.sendAndClose();
|
| 601 |
2086
|
costa
|
}
|
| 602 |
|
|
catch (IOException e) {
|
| 603 |
|
|
System.out.println("There was a problem sending email to " + to);
|
| 604 |
|
|
System.out.println("IOException: " + e.getMessage());
|
| 605 |
|
|
}
|
| 606 |
|
|
}
|
| 607 |
2022
|
costa
|
}
|
| 608 |
|
|
|
| 609 |
|
|
|
| 610 |
|
|
/**
|
| 611 |
2139
|
costa
|
* Accessor method for setting the value of the schemaLocation field.
|
| 612 |
|
|
*
|
| 613 |
|
|
* @param schemaLocation the new value of the schemaLocation field
|
| 614 |
|
|
*/
|
| 615 |
|
|
public void setSchemaLocation(String schemaLocation) {
|
| 616 |
|
|
this.schemaLocation = schemaLocation;
|
| 617 |
|
|
}
|
| 618 |
|
|
|
| 619 |
|
|
|
| 620 |
|
|
/**
|
| 621 |
2022
|
costa
|
* This inner class extends DefaultHandler. It parses the document list,
|
| 622 |
|
|
* creating a new HarvestDocument object every time it finds a </Document>
|
| 623 |
|
|
* end tag.
|
| 624 |
|
|
*/
|
| 625 |
2031
|
costa
|
class DocumentListHandler extends DefaultHandler implements ErrorHandler {
|
| 626 |
2022
|
costa
|
|
| 627 |
|
|
public String scope;
|
| 628 |
|
|
public int identifier;
|
| 629 |
2036
|
costa
|
public String identifierString;
|
| 630 |
|
|
public String documentType;
|
| 631 |
2022
|
costa
|
public int revision;
|
| 632 |
2036
|
costa
|
public String revisionString;
|
| 633 |
2022
|
costa
|
public String documentURL;
|
| 634 |
|
|
private String currentQname;
|
| 635 |
2031
|
costa
|
public final static String DEFAULT_PARSER =
|
| 636 |
|
|
"org.apache.xerces.parsers.SAXParser";
|
| 637 |
|
|
private boolean schemaValidate = true;
|
| 638 |
2022
|
costa
|
|
| 639 |
|
|
|
| 640 |
2031
|
costa
|
/**
|
| 641 |
|
|
* This method is called for any plain text within an element.
|
| 642 |
|
|
* It parses the value for any of the following elements:
|
| 643 |
|
|
* <scope>, <identifier>, <revision>, <documentType>, <documentURL>
|
| 644 |
|
|
*
|
| 645 |
|
|
* @param ch the character array holding the parsed text
|
| 646 |
|
|
* @param start the start index
|
| 647 |
|
|
* @param length the text length
|
| 648 |
|
|
*
|
| 649 |
2022
|
costa
|
*/
|
| 650 |
2031
|
costa
|
public void characters (char ch[], int start, int length) {
|
| 651 |
|
|
String s = new String(ch, start, length);
|
| 652 |
|
|
|
| 653 |
|
|
if (length > 0) {
|
| 654 |
|
|
if (currentQname.equals("scope")) {
|
| 655 |
2036
|
costa
|
scope += s;
|
| 656 |
2031
|
costa
|
}
|
| 657 |
|
|
else if (currentQname.equals("identifier")) {
|
| 658 |
2036
|
costa
|
identifierString += s;
|
| 659 |
2031
|
costa
|
}
|
| 660 |
|
|
else if (currentQname.equals("revision")) {
|
| 661 |
2036
|
costa
|
revisionString += s;
|
| 662 |
2031
|
costa
|
}
|
| 663 |
|
|
else if (currentQname.equals("documentType")) {
|
| 664 |
2036
|
costa
|
documentType += s;
|
| 665 |
2031
|
costa
|
}
|
| 666 |
|
|
else if (currentQname.equals("documentURL")) {
|
| 667 |
2036
|
costa
|
documentURL += s;
|
| 668 |
2031
|
costa
|
}
|
| 669 |
|
|
}
|
| 670 |
2022
|
costa
|
}
|
| 671 |
|
|
|
| 672 |
|
|
|
| 673 |
|
|
/**
|
| 674 |
|
|
* Handles an end-of-document event.
|
| 675 |
|
|
*/
|
| 676 |
|
|
public void endDocument () {
|
| 677 |
|
|
System.out.println("Finished parsing " + documentListURL);
|
| 678 |
|
|
}
|
| 679 |
|
|
|
| 680 |
|
|
|
| 681 |
|
|
/**
|
| 682 |
|
|
* Handles an end-of-element event. If the end tag is </Document>, then
|
| 683 |
|
|
* creates a new HarvestDocument object and pushes it to the document
|
| 684 |
|
|
* list.
|
| 685 |
|
|
*
|
| 686 |
|
|
* @param uri
|
| 687 |
|
|
* @param localname
|
| 688 |
|
|
* @param qname
|
| 689 |
|
|
*/
|
| 690 |
|
|
public void endElement(String uri,
|
| 691 |
|
|
String localname,
|
| 692 |
|
|
String qname) {
|
| 693 |
|
|
|
| 694 |
|
|
HarvestDocument harvestDocument;
|
| 695 |
|
|
|
| 696 |
2036
|
costa
|
if (qname.equals("identifier")) {
|
| 697 |
|
|
identifier = Integer.parseInt(identifierString);
|
| 698 |
|
|
}
|
| 699 |
|
|
else if (qname.equals("revision")) {
|
| 700 |
|
|
revision = Integer.parseInt(revisionString);
|
| 701 |
|
|
}
|
| 702 |
|
|
else if (qname.equals("document")) {
|
| 703 |
2022
|
costa
|
harvestDocument = new HarvestDocument(
|
| 704 |
|
|
harvester,
|
| 705 |
|
|
HarvestSiteSchedule.this,
|
| 706 |
|
|
scope,
|
| 707 |
|
|
identifier,
|
| 708 |
|
|
revision,
|
| 709 |
|
|
documentType,
|
| 710 |
|
|
documentURL
|
| 711 |
|
|
);
|
| 712 |
2031
|
costa
|
harvestDocumentList.add(harvestDocument);
|
| 713 |
2022
|
costa
|
}
|
| 714 |
2036
|
costa
|
|
| 715 |
|
|
currentQname = "";
|
| 716 |
2022
|
costa
|
}
|
| 717 |
|
|
|
| 718 |
|
|
|
| 719 |
2031
|
costa
|
/**
|
| 720 |
|
|
* Method for handling errors during a parse
|
| 721 |
|
|
*
|
| 722 |
|
|
* @param exception The parsing error
|
| 723 |
|
|
* @exception SAXException Description of Exception
|
| 724 |
2022
|
costa
|
*/
|
| 725 |
2031
|
costa
|
public void error(SAXParseException e) throws SAXParseException {
|
| 726 |
|
|
System.out.println("SAXParseException: " + e.getMessage());
|
| 727 |
|
|
throw e;
|
| 728 |
|
|
}
|
| 729 |
|
|
|
| 730 |
|
|
|
| 731 |
|
|
/**
|
| 732 |
|
|
* Run the validating parser
|
| 733 |
|
|
*
|
| 734 |
|
|
* @param xml the xml stream to be validated
|
| 735 |
|
|
* @schemaLocation relative path the to XML Schema file, e.g. "."
|
| 736 |
|
|
* @exception IOException thrown when test files can't be opened
|
| 737 |
|
|
* @exception ClassNotFoundException thrown when SAX Parser class not found
|
| 738 |
|
|
* @exception SAXException
|
| 739 |
|
|
* @exception SAXParserException
|
| 740 |
|
|
*/
|
| 741 |
|
|
public void runParser(Reader xml, String schemaLocation)
|
| 742 |
|
|
throws IOException, ClassNotFoundException,
|
| 743 |
|
|
SAXException, SAXParseException {
|
| 744 |
|
|
|
| 745 |
|
|
// Get an instance of the parser
|
| 746 |
|
|
XMLReader parser;
|
| 747 |
|
|
|
| 748 |
|
|
parser = XMLReaderFactory.createXMLReader(DEFAULT_PARSER);
|
| 749 |
|
|
// Set Handlers in the parser
|
| 750 |
|
|
parser.setContentHandler((ContentHandler)this);
|
| 751 |
|
|
parser.setErrorHandler((ErrorHandler)this);
|
| 752 |
|
|
parser.setFeature("http://xml.org/sax/features/namespaces", true);
|
| 753 |
|
|
parser.setFeature("http://xml.org/sax/features/namespace-prefixes", true);
|
| 754 |
|
|
parser.setFeature("http://xml.org/sax/features/validation", true);
|
| 755 |
|
|
parser.setProperty(
|
| 756 |
|
|
"http://apache.org/xml/properties/schema/external-schemaLocation",
|
| 757 |
|
|
schemaLocation);
|
| 758 |
|
|
|
| 759 |
|
|
if (schemaValidate) {
|
| 760 |
|
|
parser.setFeature("http://apache.org/xml/features/validation/schema",
|
| 761 |
|
|
true);
|
| 762 |
2022
|
costa
|
}
|
| 763 |
2031
|
costa
|
|
| 764 |
|
|
// Parse the document
|
| 765 |
|
|
parser.parse(new InputSource(xml));
|
| 766 |
2022
|
costa
|
}
|
| 767 |
2031
|
costa
|
/**
|
| 768 |
|
|
* Handles a start-of-document event.
|
| 769 |
|
|
*/
|
| 770 |
|
|
public void startDocument () {
|
| 771 |
|
|
System.out.println("Started parsing " + documentListURL);
|
| 772 |
|
|
}
|
| 773 |
2022
|
costa
|
|
| 774 |
2031
|
costa
|
|
| 775 |
|
|
/**
|
| 776 |
|
|
* Handles a start-of-element event.
|
| 777 |
|
|
*
|
| 778 |
|
|
* @param uri
|
| 779 |
|
|
* @param localname
|
| 780 |
|
|
* @param qname
|
| 781 |
|
|
* @param attributes
|
| 782 |
|
|
*/
|
| 783 |
|
|
public void startElement(String uri,
|
| 784 |
|
|
String localname,
|
| 785 |
|
|
String qname,
|
| 786 |
|
|
Attributes attributes) {
|
| 787 |
|
|
|
| 788 |
|
|
currentQname = qname;
|
| 789 |
2036
|
costa
|
|
| 790 |
|
|
if (qname.equals("scope")) {
|
| 791 |
|
|
scope = "";
|
| 792 |
|
|
}
|
| 793 |
|
|
else if (qname.equals("identifier")) {
|
| 794 |
|
|
identifierString = "";
|
| 795 |
|
|
}
|
| 796 |
|
|
else if (qname.equals("revision")) {
|
| 797 |
|
|
revisionString = "";
|
| 798 |
|
|
}
|
| 799 |
|
|
else if (qname.equals("documentType")) {
|
| 800 |
|
|
documentType = "";
|
| 801 |
|
|
}
|
| 802 |
|
|
else if (qname.equals("documentURL")) {
|
| 803 |
|
|
documentURL = "";
|
| 804 |
|
|
}
|
| 805 |
2031
|
costa
|
}
|
| 806 |
2022
|
costa
|
}
|
| 807 |
|
|
}
|