1 |
6705
|
leinfelder
|
/**
|
2 |
|
|
* '$RCSfile$'
|
3 |
|
|
* Purpose: A Class for upgrading the database to version 1.5
|
4 |
|
|
* Copyright: 2000 Regents of the University of California and the
|
5 |
|
|
* National Center for Ecological Analysis and Synthesis
|
6 |
|
|
* Authors: Saurabh Garg
|
7 |
|
|
*
|
8 |
|
|
* '$Author$'
|
9 |
|
|
* '$Date$'
|
10 |
|
|
* '$Revision$'
|
11 |
|
|
*
|
12 |
|
|
* This program is free software; you can redistribute it and/or modify
|
13 |
|
|
* it under the terms of the GNU General Public License as published by
|
14 |
|
|
* the Free Software Foundation; either version 2 of the License, or
|
15 |
|
|
* (at your option) any later version.
|
16 |
|
|
*
|
17 |
|
|
* This program is distributed in the hope that it will be useful,
|
18 |
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
19 |
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
20 |
|
|
* GNU General Public License for more details.
|
21 |
|
|
*
|
22 |
|
|
* You should have received a copy of the GNU General Public License
|
23 |
|
|
* along with this program; if not, write to the Free Software
|
24 |
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
25 |
|
|
*/
|
26 |
|
|
package edu.ucsb.nceas.metacat.dataone;
|
27 |
|
|
|
28 |
|
|
import java.io.IOException;
|
29 |
|
|
import java.io.InputStream;
|
30 |
|
|
import java.math.BigInteger;
|
31 |
6852
|
leinfelder
|
import java.net.URL;
|
32 |
6873
|
leinfelder
|
import java.net.URLConnection;
|
33 |
6705
|
leinfelder
|
import java.security.NoSuchAlgorithmException;
|
34 |
|
|
import java.sql.SQLException;
|
35 |
6712
|
leinfelder
|
import java.util.ArrayList;
|
36 |
6727
|
leinfelder
|
import java.util.Collections;
|
37 |
6705
|
leinfelder
|
import java.util.Date;
|
38 |
6712
|
leinfelder
|
import java.util.HashMap;
|
39 |
6705
|
leinfelder
|
import java.util.Hashtable;
|
40 |
6712
|
leinfelder
|
import java.util.List;
|
41 |
|
|
import java.util.Map;
|
42 |
6709
|
leinfelder
|
import java.util.Vector;
|
43 |
6705
|
leinfelder
|
|
44 |
|
|
import javax.xml.parsers.ParserConfigurationException;
|
45 |
|
|
import javax.xml.xpath.XPathExpressionException;
|
46 |
|
|
|
47 |
6712
|
leinfelder
|
import org.apache.commons.beanutils.BeanUtils;
|
48 |
|
|
import org.apache.commons.io.IOUtils;
|
49 |
6705
|
leinfelder
|
import org.apache.log4j.Logger;
|
50 |
6721
|
leinfelder
|
import org.apache.wicket.protocol.http.MockHttpServletRequest;
|
51 |
6705
|
leinfelder
|
import org.dataone.client.ObjectFormatCache;
|
52 |
6712
|
leinfelder
|
import org.dataone.ore.ResourceMapFactory;
|
53 |
6705
|
leinfelder
|
import org.dataone.service.exceptions.BaseException;
|
54 |
|
|
import org.dataone.service.exceptions.NotFound;
|
55 |
6721
|
leinfelder
|
import org.dataone.service.types.v1.AccessPolicy;
|
56 |
6705
|
leinfelder
|
import org.dataone.service.types.v1.Checksum;
|
57 |
|
|
import org.dataone.service.types.v1.Identifier;
|
58 |
|
|
import org.dataone.service.types.v1.NodeReference;
|
59 |
|
|
import org.dataone.service.types.v1.ObjectFormatIdentifier;
|
60 |
6721
|
leinfelder
|
import org.dataone.service.types.v1.Session;
|
61 |
6705
|
leinfelder
|
import org.dataone.service.types.v1.Subject;
|
62 |
|
|
import org.dataone.service.types.v1.SystemMetadata;
|
63 |
|
|
import org.dataone.service.types.v1.util.ChecksumUtil;
|
64 |
6709
|
leinfelder
|
import org.dataone.service.util.DateTimeMarshaller;
|
65 |
6712
|
leinfelder
|
import org.dspace.foresite.ResourceMap;
|
66 |
6705
|
leinfelder
|
import org.ecoinformatics.datamanager.DataManager;
|
67 |
|
|
import org.ecoinformatics.datamanager.database.DatabaseConnectionPoolInterface;
|
68 |
|
|
import org.ecoinformatics.datamanager.parser.DataPackage;
|
69 |
|
|
import org.jibx.runtime.JiBXException;
|
70 |
|
|
import org.xml.sax.SAXException;
|
71 |
|
|
|
72 |
|
|
import edu.ucsb.nceas.metacat.AccessionNumber;
|
73 |
|
|
import edu.ucsb.nceas.metacat.AccessionNumberException;
|
74 |
6709
|
leinfelder
|
import edu.ucsb.nceas.metacat.DBUtil;
|
75 |
6705
|
leinfelder
|
import edu.ucsb.nceas.metacat.IdentifierManager;
|
76 |
|
|
import edu.ucsb.nceas.metacat.McdbDocNotFoundException;
|
77 |
|
|
import edu.ucsb.nceas.metacat.McdbException;
|
78 |
6712
|
leinfelder
|
import edu.ucsb.nceas.metacat.MetaCatServlet;
|
79 |
6705
|
leinfelder
|
import edu.ucsb.nceas.metacat.MetacatHandler;
|
80 |
6708
|
leinfelder
|
import edu.ucsb.nceas.metacat.accesscontrol.AccessControlException;
|
81 |
6705
|
leinfelder
|
import edu.ucsb.nceas.metacat.client.InsufficientKarmaException;
|
82 |
|
|
import edu.ucsb.nceas.metacat.dataone.hazelcast.HazelcastService;
|
83 |
|
|
import edu.ucsb.nceas.metacat.dataquery.MetacatDatabaseConnectionPoolFactory;
|
84 |
|
|
import edu.ucsb.nceas.metacat.properties.PropertyService;
|
85 |
6708
|
leinfelder
|
import edu.ucsb.nceas.metacat.replication.ReplicationService;
|
86 |
6721
|
leinfelder
|
import edu.ucsb.nceas.metacat.shared.AccessException;
|
87 |
6708
|
leinfelder
|
import edu.ucsb.nceas.metacat.shared.HandlerException;
|
88 |
6705
|
leinfelder
|
import edu.ucsb.nceas.metacat.util.DocumentUtil;
|
89 |
|
|
import edu.ucsb.nceas.utilities.ParseLSIDException;
|
90 |
|
|
import edu.ucsb.nceas.utilities.PropertyNotFoundException;
|
91 |
|
|
|
92 |
|
|
public class SystemMetadataFactory {
|
93 |
6706
|
leinfelder
|
|
94 |
6800
|
leinfelder
|
private static final String resourceMapPrefix = "resourceMap_";
|
95 |
6707
|
leinfelder
|
private static Logger logMetacat = Logger.getLogger(SystemMetadataFactory.class);
|
96 |
6712
|
leinfelder
|
|
97 |
6705
|
leinfelder
|
/**
|
98 |
6706
|
leinfelder
|
* Creates a system metadata object for insertion into metacat
|
99 |
|
|
*
|
100 |
|
|
* @param localId
|
101 |
|
|
* The local document identifier
|
102 |
|
|
* @param user
|
103 |
|
|
* The user submitting the system metadata document
|
104 |
|
|
* @param groups
|
105 |
|
|
* The groups the user belongs to
|
106 |
|
|
*
|
107 |
|
|
* @return sysMeta The system metadata object created
|
108 |
6708
|
leinfelder
|
* @throws SAXException
|
109 |
|
|
* @throws HandlerException
|
110 |
|
|
* @throws AccessControlException
|
111 |
6721
|
leinfelder
|
* @throws AccessException
|
112 |
6706
|
leinfelder
|
*/
|
113 |
6852
|
leinfelder
|
public static SystemMetadata createSystemMetadata(String localId, boolean includeORE, boolean downloadData)
|
114 |
6706
|
leinfelder
|
throws McdbException, McdbDocNotFoundException, SQLException,
|
115 |
|
|
IOException, AccessionNumberException, ClassNotFoundException,
|
116 |
|
|
InsufficientKarmaException, ParseLSIDException,
|
117 |
|
|
PropertyNotFoundException, BaseException, NoSuchAlgorithmException,
|
118 |
6721
|
leinfelder
|
JiBXException, AccessControlException, HandlerException, SAXException, AccessException {
|
119 |
6707
|
leinfelder
|
|
120 |
6706
|
leinfelder
|
logMetacat.debug("MetacatHandler.createSystemMetadata() called.");
|
121 |
|
|
logMetacat.debug("provided localId: " + localId);
|
122 |
6705
|
leinfelder
|
|
123 |
6706
|
leinfelder
|
// create system metadata for the document
|
124 |
|
|
SystemMetadata sysMeta = new SystemMetadata();
|
125 |
|
|
sysMeta.setSerialVersion(BigInteger.valueOf(1));
|
126 |
|
|
AccessionNumber accNum = new AccessionNumber(localId, "NONE");
|
127 |
6808
|
leinfelder
|
int rev = Integer.valueOf(accNum.getRev());
|
128 |
|
|
|
129 |
6706
|
leinfelder
|
String guid = null;
|
130 |
|
|
try {
|
131 |
|
|
// get the guid if it exists
|
132 |
6707
|
leinfelder
|
guid = IdentifierManager.getInstance().getGUID(accNum.getDocid(), rev);
|
133 |
6706
|
leinfelder
|
} catch (McdbDocNotFoundException dnfe) {
|
134 |
|
|
// otherwise create the mapping
|
135 |
6707
|
leinfelder
|
logMetacat.debug("There was a problem getting the guid from "
|
136 |
6706
|
leinfelder
|
+ "the given localId (docid and revision). The error message was: "
|
137 |
|
|
+ dnfe.getMessage());
|
138 |
6707
|
leinfelder
|
logMetacat.debug("No guid in the identifier table. adding it for " + localId);
|
139 |
6706
|
leinfelder
|
IdentifierManager.getInstance().createMapping(localId, localId);
|
140 |
|
|
logMetacat.debug("Mapping created for " + localId);
|
141 |
|
|
logMetacat.debug("accessionNumber: " + accNum);
|
142 |
6707
|
leinfelder
|
guid = IdentifierManager.getInstance().getGUID(accNum.getDocid(), rev);
|
143 |
6706
|
leinfelder
|
}
|
144 |
|
|
Identifier identifier = new Identifier();
|
145 |
|
|
identifier.setValue(guid);
|
146 |
6705
|
leinfelder
|
|
147 |
6706
|
leinfelder
|
// set the id
|
148 |
|
|
sysMeta.setIdentifier(identifier);
|
149 |
6705
|
leinfelder
|
|
150 |
6706
|
leinfelder
|
// get the data or metadata object
|
151 |
|
|
InputStream inputStream;
|
152 |
|
|
try {
|
153 |
|
|
inputStream = MetacatHandler.read(localId);
|
154 |
|
|
} catch (ParseLSIDException ple) {
|
155 |
|
|
logMetacat.debug("There was a problem parsing the LSID from "
|
156 |
|
|
+ localId + ". The error message was: " + ple.getMessage());
|
157 |
|
|
throw ple;
|
158 |
6705
|
leinfelder
|
|
159 |
6706
|
leinfelder
|
} catch (PropertyNotFoundException pnfe) {
|
160 |
|
|
logMetacat.debug("There was a problem finding a property. "
|
161 |
|
|
+ "The error message was: " + pnfe.getMessage());
|
162 |
|
|
throw pnfe;
|
163 |
|
|
|
164 |
|
|
} catch (McdbException me) {
|
165 |
|
|
logMetacat.debug("There was a Metacat problem. "
|
166 |
|
|
+ "The error message was: " + me.getMessage());
|
167 |
|
|
throw me;
|
168 |
|
|
|
169 |
|
|
} catch (SQLException sqle) {
|
170 |
|
|
logMetacat.debug("There was a SQL problem. "
|
171 |
|
|
+ "The error message was: " + sqle.getMessage());
|
172 |
|
|
throw sqle;
|
173 |
|
|
|
174 |
|
|
} catch (ClassNotFoundException cnfe) {
|
175 |
|
|
logMetacat.debug("There was a problem finding a class. "
|
176 |
|
|
+ "The error message was: " + cnfe.getMessage());
|
177 |
|
|
throw cnfe;
|
178 |
|
|
|
179 |
|
|
} catch (IOException ioe) {
|
180 |
|
|
logMetacat.debug("There was an I/O exception. "
|
181 |
|
|
+ "The error message was: " + ioe.getMessage());
|
182 |
|
|
throw ioe;
|
183 |
|
|
|
184 |
|
|
} // end try()
|
185 |
|
|
|
186 |
|
|
// get additional docinfo
|
187 |
6708
|
leinfelder
|
Hashtable<String, String> docInfo = ReplicationService.getDocumentInfoMap(localId);
|
188 |
6706
|
leinfelder
|
// set the default object format
|
189 |
6708
|
leinfelder
|
String doctype = docInfo.get("doctype");
|
190 |
6706
|
leinfelder
|
ObjectFormatIdentifier fmtid = null;
|
191 |
|
|
|
192 |
|
|
// set the object format, fall back to defaults
|
193 |
|
|
try {
|
194 |
6707
|
leinfelder
|
fmtid = ObjectFormatCache.getInstance().getFormat(doctype).getFormatId();
|
195 |
6706
|
leinfelder
|
} catch (NotFound nfe) {
|
196 |
|
|
|
197 |
|
|
try {
|
198 |
|
|
// format is not registered, use default
|
199 |
|
|
if (doctype.trim().equals("BIN")) {
|
200 |
|
|
fmtid = ObjectFormatCache.getInstance().getFormat(
|
201 |
|
|
"application/octet-stream").getFormatId();
|
202 |
|
|
|
203 |
|
|
} else {
|
204 |
|
|
fmtid = ObjectFormatCache.getInstance().getFormat(
|
205 |
|
|
"text/plain").getFormatId();
|
206 |
|
|
}
|
207 |
|
|
|
208 |
|
|
} catch (NotFound nf) {
|
209 |
6707
|
leinfelder
|
logMetacat.error("There was a problem getting the default format "
|
210 |
6706
|
leinfelder
|
+ "from the ObjectFormatCache: "
|
211 |
|
|
+ nf.getMessage());
|
212 |
|
|
throw nf;
|
213 |
|
|
}
|
214 |
|
|
|
215 |
|
|
}
|
216 |
|
|
|
217 |
|
|
sysMeta.setFormatId(fmtid);
|
218 |
6707
|
leinfelder
|
logMetacat.debug("The ObjectFormat for " + localId + " is " + fmtid.getValue());
|
219 |
6706
|
leinfelder
|
|
220 |
6721
|
leinfelder
|
// create the checksum
|
221 |
|
|
inputStream = MetacatHandler.read(localId);
|
222 |
|
|
String algorithm = "MD5";
|
223 |
|
|
Checksum checksum = ChecksumUtil.checksum(inputStream, algorithm);
|
224 |
|
|
sysMeta.setChecksum(checksum);
|
225 |
|
|
|
226 |
|
|
// set the size
|
227 |
|
|
inputStream = MetacatHandler.read(localId);
|
228 |
|
|
String sizeStr = new Long(sizeOfStream(inputStream)).toString();
|
229 |
|
|
sysMeta.setSize(new BigInteger(sizeStr));
|
230 |
|
|
|
231 |
|
|
// submitter
|
232 |
|
|
Subject submitter = new Subject();
|
233 |
|
|
submitter.setValue(docInfo.get("user_updated"));
|
234 |
|
|
sysMeta.setSubmitter(submitter);
|
235 |
|
|
|
236 |
|
|
// rights holder
|
237 |
|
|
Subject owner = new Subject();
|
238 |
|
|
owner.setValue(docInfo.get("user_owner"));
|
239 |
|
|
sysMeta.setRightsHolder(owner);
|
240 |
|
|
|
241 |
|
|
// dates
|
242 |
|
|
String createdDateString = docInfo.get("date_created");
|
243 |
|
|
String updatedDateString = docInfo.get("date_updated");
|
244 |
|
|
Date createdDate = DateTimeMarshaller.deserializeDateToUTC(createdDateString);
|
245 |
|
|
Date updatedDate = DateTimeMarshaller.deserializeDateToUTC(updatedDateString);
|
246 |
|
|
sysMeta.setDateUploaded(createdDate);
|
247 |
|
|
sysMeta.setDateSysMetadataModified(updatedDate);
|
248 |
|
|
|
249 |
|
|
// set the revision history
|
250 |
|
|
String docidWithoutRev = accNum.getDocid();
|
251 |
|
|
Identifier obsoletedBy = null;
|
252 |
|
|
Identifier obsoletes = null;
|
253 |
|
|
Vector<Integer> revisions = DBUtil.getRevListFromRevisionTable(docidWithoutRev);
|
254 |
6727
|
leinfelder
|
// ensure this ordering since processing depends on it
|
255 |
|
|
Collections.sort(revisions);
|
256 |
6721
|
leinfelder
|
for (int existingRev: revisions) {
|
257 |
|
|
// use the docid+rev as the guid
|
258 |
|
|
String existingPid = docidWithoutRev + "." + existingRev;
|
259 |
|
|
if (existingRev < rev) {
|
260 |
|
|
// it's the old docid, until it's not
|
261 |
|
|
obsoletes = new Identifier();
|
262 |
|
|
obsoletes.setValue(existingPid);
|
263 |
|
|
}
|
264 |
|
|
if (existingRev > rev) {
|
265 |
|
|
// it's the newer docid
|
266 |
|
|
obsoletedBy = new Identifier();
|
267 |
|
|
obsoletedBy.setValue(existingPid);
|
268 |
|
|
// only want the version just after it
|
269 |
|
|
break;
|
270 |
|
|
}
|
271 |
|
|
}
|
272 |
6725
|
leinfelder
|
// set them on our object
|
273 |
6721
|
leinfelder
|
sysMeta.setObsoletedBy(obsoletedBy);
|
274 |
|
|
sysMeta.setObsoletes(obsoletes);
|
275 |
|
|
|
276 |
6725
|
leinfelder
|
// update the system metadata for the object[s] we are revising
|
277 |
|
|
if (obsoletedBy != null) {
|
278 |
6747
|
leinfelder
|
//SystemMetadata obsoletedBySysMeta = HazelcastService.getInstance().getSystemMetadataMap().get(obsoletedBy);
|
279 |
|
|
SystemMetadata obsoletedBySysMeta = IdentifierManager.getInstance().getSystemMetadata(obsoletedBy.getValue());
|
280 |
6725
|
leinfelder
|
if (obsoletedBySysMeta != null) {
|
281 |
|
|
obsoletedBySysMeta.setObsoletes(identifier);
|
282 |
|
|
HazelcastService.getInstance().getSystemMetadataMap().put(obsoletedBy, obsoletedBySysMeta);
|
283 |
|
|
}
|
284 |
|
|
}
|
285 |
|
|
if (obsoletes != null) {
|
286 |
6747
|
leinfelder
|
//SystemMetadata obsoletesSysMeta = HazelcastService.getInstance().getSystemMetadataMap().get(obsoletes);
|
287 |
|
|
SystemMetadata obsoletesSysMeta = IdentifierManager.getInstance().getSystemMetadata(obsoletes.getValue());
|
288 |
6725
|
leinfelder
|
if (obsoletesSysMeta != null) {
|
289 |
|
|
obsoletesSysMeta.setObsoletedBy(identifier);
|
290 |
|
|
HazelcastService.getInstance().getSystemMetadataMap().put(obsoletes, obsoletesSysMeta);
|
291 |
|
|
}
|
292 |
|
|
}
|
293 |
|
|
|
294 |
6744
|
leinfelder
|
// look up the access control policy we have in metacat
|
295 |
|
|
AccessPolicy accessPolicy = IdentifierManager.getInstance().getAccessPolicy(guid);
|
296 |
6721
|
leinfelder
|
sysMeta.setAccessPolicy(accessPolicy);
|
297 |
|
|
|
298 |
|
|
// authoritative node
|
299 |
|
|
NodeReference nr = new NodeReference();
|
300 |
|
|
nr.setValue(PropertyService.getProperty("dataone.memberNodeId"));
|
301 |
|
|
sysMeta.setOriginMemberNode(nr);
|
302 |
|
|
sysMeta.setAuthoritativeMemberNode(nr);
|
303 |
|
|
|
304 |
6706
|
leinfelder
|
// further parse EML documents to get data object format,
|
305 |
|
|
// describes and describedBy information
|
306 |
|
|
if (fmtid == ObjectFormatCache.getInstance().getFormat(
|
307 |
|
|
"eml://ecoinformatics.org/eml-2.0.0").getFormatId()
|
308 |
|
|
|| fmtid == ObjectFormatCache.getInstance().getFormat(
|
309 |
|
|
"eml://ecoinformatics.org/eml-2.0.1").getFormatId()
|
310 |
|
|
|| fmtid == ObjectFormatCache.getInstance().getFormat(
|
311 |
|
|
"eml://ecoinformatics.org/eml-2.1.0").getFormatId()
|
312 |
|
|
|| fmtid == ObjectFormatCache.getInstance().getFormat(
|
313 |
|
|
"eml://ecoinformatics.org/eml-2.1.1").getFormatId()) {
|
314 |
|
|
|
315 |
|
|
try {
|
316 |
6721
|
leinfelder
|
inputStream = MetacatHandler.read(localId);
|
317 |
6707
|
leinfelder
|
DatabaseConnectionPoolInterface connectionPool =
|
318 |
|
|
MetacatDatabaseConnectionPoolFactory.getDatabaseConnectionPoolInterface();
|
319 |
|
|
DataManager dataManager =
|
320 |
|
|
DataManager.getInstance(connectionPool, connectionPool.getDBAdapterName());
|
321 |
|
|
DataPackage dataPackage = dataManager.parseMetadata(inputStream);
|
322 |
6706
|
leinfelder
|
|
323 |
6721
|
leinfelder
|
// iterate through the data objects in the EML doc and add sysmeta
|
324 |
6707
|
leinfelder
|
logMetacat.debug("In createSystemMetadata() the number of data "
|
325 |
6706
|
leinfelder
|
+ "entities is: "
|
326 |
|
|
+ dataPackage.getEntityNumber());
|
327 |
|
|
|
328 |
6712
|
leinfelder
|
// for generating the ORE map
|
329 |
|
|
Map<Identifier, List<Identifier>> idMap = new HashMap<Identifier, List<Identifier>>();
|
330 |
|
|
List<Identifier> dataIds = new ArrayList<Identifier>();
|
331 |
|
|
|
332 |
6706
|
leinfelder
|
// iterate through data objects described by the EML
|
333 |
6744
|
leinfelder
|
if (dataPackage.getEntityList() != null) {
|
334 |
|
|
for (int j = 0; j < dataPackage.getEntityList().length; j++) {
|
335 |
|
|
|
336 |
|
|
String dataDocUrl = dataPackage.getEntityList()[j].getURL();
|
337 |
|
|
String dataDocMimeType = dataPackage.getEntityList()[j].getDataFormat();
|
338 |
|
|
// default to binary
|
339 |
|
|
if (dataDocMimeType == null) {
|
340 |
|
|
dataDocMimeType = ObjectFormatCache.getInstance()
|
341 |
|
|
.getFormat("application/octet-stream")
|
342 |
|
|
.getFormatId().getValue();
|
343 |
6721
|
leinfelder
|
}
|
344 |
6852
|
leinfelder
|
|
345 |
|
|
// process the data
|
346 |
6855
|
leinfelder
|
boolean remoteData = false;
|
347 |
6852
|
leinfelder
|
String dataDocLocalId = null;
|
348 |
|
|
Identifier dataGuid = new Identifier();
|
349 |
|
|
|
350 |
|
|
// handle ecogrid, or downloadable data
|
351 |
|
|
String ecogridPrefix = "ecogrid://knb/";
|
352 |
|
|
if (dataDocUrl.trim().startsWith(ecogridPrefix)) {
|
353 |
|
|
dataDocLocalId = dataDocUrl.substring(dataDocUrl.indexOf(ecogridPrefix) + ecogridPrefix.length());
|
354 |
|
|
} else {
|
355 |
|
|
// should we try downloading the remote data?
|
356 |
|
|
if (downloadData) {
|
357 |
|
|
InputStream dataObject = null;
|
358 |
|
|
try {
|
359 |
|
|
// download the data from the URL
|
360 |
|
|
URL dataURL = new URL(dataDocUrl);
|
361 |
6873
|
leinfelder
|
URLConnection dataConnection = dataURL.openConnection();
|
362 |
|
|
|
363 |
|
|
// default is to download the data
|
364 |
|
|
dataObject = dataConnection.getInputStream();
|
365 |
|
|
|
366 |
|
|
String detectedContentType = dataConnection.getContentType();
|
367 |
|
|
logMetacat.info("Detected content type: " + detectedContentType);
|
368 |
|
|
|
369 |
|
|
if (detectedContentType != null) {
|
370 |
|
|
// seems to be HTML from the remote location
|
371 |
|
|
if (detectedContentType.contains("html")) {
|
372 |
|
|
// if we are not expecting it, we skip it
|
373 |
|
|
if (!dataDocMimeType.contains("html")) {
|
374 |
|
|
// set to null so we don't download it
|
375 |
|
|
dataObject = null;
|
376 |
|
|
logMetacat.warn("Skipping remote resource, unexpected HTML content type at: " + dataDocUrl);
|
377 |
|
|
}
|
378 |
|
|
}
|
379 |
|
|
|
380 |
|
|
// TODO: any other special processing (csv, images, etc)?
|
381 |
|
|
} else {
|
382 |
|
|
// if we don't know what it is, should we skip it?
|
383 |
|
|
dataObject = null;
|
384 |
|
|
logMetacat.warn("Skipping remote resource, unknown content type at: " + dataDocUrl);
|
385 |
|
|
}
|
386 |
|
|
|
387 |
6852
|
leinfelder
|
} catch (Exception e) {
|
388 |
|
|
// error with the download
|
389 |
|
|
logMetacat.warn("Error downloading remote data. " + e.getMessage());
|
390 |
|
|
}
|
391 |
|
|
|
392 |
|
|
if (dataObject != null) {
|
393 |
|
|
// create the local version of it
|
394 |
|
|
dataDocLocalId = DocumentUtil.generateDocumentId(1);
|
395 |
|
|
IdentifierManager.getInstance().createMapping(dataDocLocalId, dataDocLocalId);
|
396 |
|
|
dataGuid.setValue(dataDocLocalId);
|
397 |
|
|
|
398 |
|
|
// save it locally
|
399 |
|
|
Session session = new Session();
|
400 |
|
|
session.setSubject(submitter);
|
401 |
|
|
MockHttpServletRequest request = new MockHttpServletRequest(null, null, null);
|
402 |
|
|
MNodeService.getInstance(request).insertDataObject(dataObject, dataGuid, session);
|
403 |
6855
|
leinfelder
|
|
404 |
|
|
remoteData = true;
|
405 |
6852
|
leinfelder
|
}
|
406 |
|
|
}
|
407 |
|
|
|
408 |
|
|
}
|
409 |
|
|
|
410 |
6744
|
leinfelder
|
logMetacat.debug("Data local ID: " + dataDocLocalId);
|
411 |
|
|
logMetacat.debug("Data URL : " + dataDocUrl);
|
412 |
|
|
logMetacat.debug("Data mime : " + dataDocMimeType);
|
413 |
6852
|
leinfelder
|
|
414 |
|
|
// now we have a local id for the data
|
415 |
|
|
if (dataDocLocalId != null) {
|
416 |
|
|
|
417 |
6744
|
leinfelder
|
// look up the guid for the data
|
418 |
|
|
String dataDocid = DocumentUtil.getSmartDocId(dataDocLocalId);
|
419 |
|
|
int dataRev = DocumentUtil.getRevisionFromAccessionNumber(dataDocLocalId);
|
420 |
|
|
|
421 |
|
|
// check if data system metadata exists already
|
422 |
|
|
SystemMetadata dataSysMeta = null;
|
423 |
|
|
String dataGuidString = null;
|
424 |
6706
|
leinfelder
|
try {
|
425 |
6744
|
leinfelder
|
// look for the identifier
|
426 |
6707
|
leinfelder
|
dataGuidString = IdentifierManager.getInstance().getGUID(dataDocid, dataRev);
|
427 |
6744
|
leinfelder
|
// set it
|
428 |
6706
|
leinfelder
|
dataGuid.setValue(dataGuidString);
|
429 |
6744
|
leinfelder
|
// look up the system metadata
|
430 |
6706
|
leinfelder
|
try {
|
431 |
6744
|
leinfelder
|
dataSysMeta = HazelcastService.getInstance().getSystemMetadataMap().get(dataGuid);
|
432 |
|
|
} catch (Exception e) {
|
433 |
|
|
// probably not in the system
|
434 |
|
|
dataSysMeta = null;
|
435 |
6706
|
leinfelder
|
}
|
436 |
6744
|
leinfelder
|
//dataSysMeta = IdentifierManager.getInstance().getSystemMetadata(dataGuidString);
|
437 |
|
|
} catch (McdbDocNotFoundException nf) {
|
438 |
|
|
// we didn't find it
|
439 |
|
|
dataSysMeta = null;
|
440 |
|
|
}
|
441 |
6712
|
leinfelder
|
|
442 |
6744
|
leinfelder
|
// we'll have to generate it
|
443 |
|
|
if (dataSysMeta == null) {
|
444 |
|
|
// System metadata for data doesn't exist yet, so create it
|
445 |
|
|
logMetacat.debug("There was not an existing system metadata document for " + dataDocLocalId);
|
446 |
|
|
try {
|
447 |
|
|
logMetacat.debug("Creating a system metadata " + "document for " + dataDocLocalId);
|
448 |
6852
|
leinfelder
|
dataSysMeta = createSystemMetadata(dataDocLocalId, includeORE, false);
|
449 |
6744
|
leinfelder
|
|
450 |
|
|
// now look it up again
|
451 |
|
|
dataGuidString = IdentifierManager.getInstance().getGUID(dataDocid, dataRev);
|
452 |
|
|
|
453 |
|
|
// set the guid
|
454 |
|
|
dataGuid.setValue(dataGuidString);
|
455 |
|
|
|
456 |
|
|
// set object format
|
457 |
|
|
logMetacat.debug("Updating system metadata for "
|
458 |
|
|
+ dataGuid.getValue() + " to "
|
459 |
|
|
+ dataDocMimeType);
|
460 |
|
|
try {
|
461 |
|
|
ObjectFormatIdentifier fmt =
|
462 |
|
|
ObjectFormatCache.getInstance().getFormat(dataDocMimeType).getFormatId();
|
463 |
|
|
dataSysMeta.setFormatId(fmt);
|
464 |
|
|
} catch (NotFound nfe) {
|
465 |
|
|
logMetacat.debug("Couldn't find format identifier for: "
|
466 |
|
|
+ dataDocMimeType
|
467 |
|
|
+ ". Setting it to application/octet-stream.");
|
468 |
|
|
ObjectFormatIdentifier newFmtid = new ObjectFormatIdentifier();
|
469 |
|
|
newFmtid.setValue("application/octet-stream");
|
470 |
|
|
}
|
471 |
6855
|
leinfelder
|
|
472 |
|
|
// inherit access rules from metadata, if we don't have our own
|
473 |
|
|
if (remoteData) {
|
474 |
|
|
dataSysMeta.setAccessPolicy(sysMeta.getAccessPolicy());
|
475 |
|
|
// TODO: use access rules defined in EML, per data file
|
476 |
|
|
}
|
477 |
|
|
|
478 |
6744
|
leinfelder
|
// update the values
|
479 |
|
|
HazelcastService.getInstance().getSystemMetadataMap().put(dataSysMeta.getIdentifier(), dataSysMeta);
|
480 |
|
|
|
481 |
|
|
|
482 |
|
|
} catch (McdbDocNotFoundException mdnf) {
|
483 |
|
|
mdnf.printStackTrace();
|
484 |
|
|
throw mdnf;
|
485 |
|
|
} catch (NumberFormatException nfe) {
|
486 |
|
|
nfe.printStackTrace();
|
487 |
|
|
throw nfe;
|
488 |
|
|
} catch (AccessionNumberException ane) {
|
489 |
|
|
ane.printStackTrace();
|
490 |
|
|
throw ane;
|
491 |
|
|
} catch (SQLException sqle) {
|
492 |
|
|
sqle.printStackTrace();
|
493 |
|
|
throw sqle;
|
494 |
|
|
} catch (NoSuchAlgorithmException nsae) {
|
495 |
|
|
nsae.printStackTrace();
|
496 |
|
|
throw nsae;
|
497 |
|
|
} catch (IOException ioe) {
|
498 |
|
|
ioe.printStackTrace();
|
499 |
|
|
throw ioe;
|
500 |
|
|
} catch (PropertyNotFoundException pnfe) {
|
501 |
|
|
pnfe.printStackTrace();
|
502 |
|
|
throw pnfe;
|
503 |
|
|
} catch (BaseException be) {
|
504 |
|
|
be.printStackTrace();
|
505 |
|
|
throw be;
|
506 |
|
|
}
|
507 |
|
|
|
508 |
|
|
}
|
509 |
6721
|
leinfelder
|
|
510 |
6744
|
leinfelder
|
// part of the ORE package
|
511 |
|
|
dataIds.add(dataGuid);
|
512 |
|
|
|
513 |
|
|
} // end if (EML package)
|
514 |
|
|
|
515 |
|
|
} // end for (data entities)
|
516 |
|
|
|
517 |
|
|
} // data entities not null
|
518 |
|
|
|
519 |
6712
|
leinfelder
|
// ORE map
|
520 |
6713
|
leinfelder
|
if (includeORE) {
|
521 |
6800
|
leinfelder
|
// can we generate them?
|
522 |
6713
|
leinfelder
|
if (!dataIds.isEmpty()) {
|
523 |
6800
|
leinfelder
|
// it doesn't exist in the system?
|
524 |
|
|
if (!oreExistsFor(sysMeta.getIdentifier())) {
|
525 |
|
|
|
526 |
|
|
// generate the ORE map for this datapackage
|
527 |
|
|
Identifier resourceMapId = new Identifier();
|
528 |
|
|
// want to be able to run this over and over again for now
|
529 |
|
|
resourceMapId.setValue(resourceMapPrefix + sysMeta.getIdentifier().getValue());
|
530 |
|
|
idMap.put(sysMeta.getIdentifier(), dataIds);
|
531 |
|
|
ResourceMap rm = ResourceMapFactory.getInstance().createResourceMap(resourceMapId, idMap);
|
532 |
|
|
String resourceMapXML = ResourceMapFactory.getInstance().serializeResourceMap(rm);
|
533 |
|
|
// copy most of the same system metadata as the packaging metadata
|
534 |
|
|
SystemMetadata resourceMapSysMeta = new SystemMetadata();
|
535 |
|
|
BeanUtils.copyProperties(resourceMapSysMeta, sysMeta);
|
536 |
|
|
resourceMapSysMeta.setIdentifier(resourceMapId);
|
537 |
|
|
Checksum oreChecksum = ChecksumUtil.checksum(IOUtils.toInputStream(resourceMapXML, MetaCatServlet.DEFAULT_ENCODING), "MD5");
|
538 |
|
|
resourceMapSysMeta.setChecksum(oreChecksum);
|
539 |
|
|
ObjectFormatIdentifier formatId = ObjectFormatCache.getInstance().getFormat("http://www.openarchives.org/ore/terms").getFormatId();
|
540 |
|
|
resourceMapSysMeta.setFormatId(formatId);
|
541 |
|
|
resourceMapSysMeta.setSize(BigInteger.valueOf(sizeOfStream(IOUtils.toInputStream(resourceMapXML, MetaCatServlet.DEFAULT_ENCODING))));
|
542 |
|
|
|
543 |
|
|
// set the revision graph
|
544 |
|
|
resourceMapSysMeta.setObsoletes(null);
|
545 |
|
|
resourceMapSysMeta.setObsoletedBy(null);
|
546 |
|
|
// look up the resource map that this one obsoletes
|
547 |
|
|
if (sysMeta.getObsoletes() != null) {
|
548 |
|
|
Identifier resourceMapObsoletes = new Identifier();
|
549 |
|
|
resourceMapObsoletes.setValue(resourceMapPrefix + sysMeta.getObsoletes().getValue());
|
550 |
|
|
resourceMapSysMeta.setObsoletes(resourceMapObsoletes);
|
551 |
|
|
SystemMetadata resourceMapObsoletesSystemMetadata = HazelcastService.getInstance().getSystemMetadataMap().get(resourceMapObsoletes);
|
552 |
|
|
if (resourceMapObsoletesSystemMetadata != null) {
|
553 |
|
|
resourceMapObsoletesSystemMetadata.setObsoletedBy(resourceMapId);
|
554 |
|
|
HazelcastService.getInstance().getSystemMetadataMap().put(resourceMapObsoletes, resourceMapObsoletesSystemMetadata);
|
555 |
|
|
}
|
556 |
|
|
}
|
557 |
|
|
// look up the resource map that this one is obsoletedBy
|
558 |
|
|
if (sysMeta.getObsoletedBy() != null) {
|
559 |
|
|
Identifier resourceMapObsoletedBy = new Identifier();
|
560 |
|
|
resourceMapObsoletedBy.setValue(resourceMapPrefix + sysMeta.getObsoletedBy().getValue());
|
561 |
|
|
resourceMapSysMeta.setObsoletedBy(resourceMapObsoletedBy);
|
562 |
|
|
SystemMetadata resourceMapObsoletedBySystemMetadata = HazelcastService.getInstance().getSystemMetadataMap().get(resourceMapObsoletedBy);
|
563 |
|
|
if (resourceMapObsoletedBySystemMetadata != null) {
|
564 |
|
|
resourceMapObsoletedBySystemMetadata.setObsoletes(resourceMapId);
|
565 |
|
|
HazelcastService.getInstance().getSystemMetadataMap().put(resourceMapObsoletedBy, resourceMapObsoletedBySystemMetadata);
|
566 |
|
|
}
|
567 |
|
|
}
|
568 |
|
|
|
569 |
|
|
// save it locally
|
570 |
|
|
Session session = new Session();
|
571 |
|
|
session.setSubject(submitter);
|
572 |
|
|
MockHttpServletRequest request = new MockHttpServletRequest(null, null, null);
|
573 |
6808
|
leinfelder
|
MNodeService.getInstance(request).insertDataObject(IOUtils.toInputStream(resourceMapXML, MetaCatServlet.DEFAULT_ENCODING), resourceMapId, session);
|
574 |
|
|
MNodeService.getInstance(request).insertSystemMetadata(resourceMapSysMeta);
|
575 |
6800
|
leinfelder
|
}
|
576 |
6713
|
leinfelder
|
}
|
577 |
|
|
}
|
578 |
6706
|
leinfelder
|
|
579 |
|
|
} catch (ParserConfigurationException pce) {
|
580 |
6707
|
leinfelder
|
logMetacat.debug("There was a problem parsing the EML document. "
|
581 |
6706
|
leinfelder
|
+ "The error message was: " + pce.getMessage());
|
582 |
|
|
|
583 |
|
|
} catch (SAXException saxe) {
|
584 |
6707
|
leinfelder
|
logMetacat.debug("There was a problem traversing the EML document. "
|
585 |
6706
|
leinfelder
|
+ "The error message was: " + saxe.getMessage());
|
586 |
|
|
|
587 |
|
|
} catch (XPathExpressionException xpee) {
|
588 |
6707
|
leinfelder
|
logMetacat.debug("There was a problem searching the EML document. "
|
589 |
6706
|
leinfelder
|
+ "The error message was: " + xpee.getMessage());
|
590 |
|
|
} catch (Exception e) {
|
591 |
6707
|
leinfelder
|
logMetacat.debug("There was a problem creating System Metadata. "
|
592 |
6706
|
leinfelder
|
+ "The error message was: " + e.getMessage());
|
593 |
6721
|
leinfelder
|
e.printStackTrace();
|
594 |
6706
|
leinfelder
|
} // end try()
|
595 |
|
|
|
596 |
|
|
} // end if()
|
597 |
|
|
|
598 |
|
|
return sysMeta;
|
599 |
|
|
}
|
600 |
6800
|
leinfelder
|
|
601 |
|
|
/**
|
602 |
|
|
* Determines if we already have registered an ORE map for this package
|
603 |
|
|
* @param guid of the EML/packaging object
|
604 |
|
|
* @return true if there is an ORE map for the given package
|
605 |
|
|
*/
|
606 |
|
|
private static boolean oreExistsFor(Identifier guid) {
|
607 |
|
|
// TODO: implement call to CN.search()
|
608 |
|
|
return false;
|
609 |
|
|
}
|
610 |
6706
|
leinfelder
|
|
611 |
6707
|
leinfelder
|
/**
|
612 |
6706
|
leinfelder
|
* Find the size (in bytes) of a stream. Note: This needs to refactored out
|
613 |
|
|
* of MetacatHandler and into a utility when stream i/o in Metacat is
|
614 |
|
|
* evaluated.
|
615 |
|
|
*
|
616 |
|
|
* @param is The InputStream of bytes
|
617 |
|
|
*
|
618 |
|
|
* @return size The size in bytes of the input stream as a long
|
619 |
|
|
*
|
620 |
|
|
* @throws IOException
|
621 |
|
|
*/
|
622 |
|
|
private static long sizeOfStream(InputStream is) throws IOException {
|
623 |
|
|
|
624 |
|
|
long size = 0;
|
625 |
|
|
byte[] b = new byte[1024];
|
626 |
|
|
int numread = is.read(b, 0, 1024);
|
627 |
|
|
while (numread != -1) {
|
628 |
|
|
size += numread;
|
629 |
|
|
numread = is.read(b, 0, 1024);
|
630 |
|
|
}
|
631 |
|
|
return size;
|
632 |
|
|
|
633 |
|
|
}
|
634 |
6705
|
leinfelder
|
}
|