Revision 6706
Added by ben leinfelder about 13 years ago
src/edu/ucsb/nceas/metacat/dataone/SystemMetadataFactory.java | ||
---|---|---|
71 | 71 |
import edu.ucsb.nceas.utilities.PropertyNotFoundException; |
72 | 72 |
|
73 | 73 |
public class SystemMetadataFactory { |
74 |
|
|
75 |
private static Logger logMetacat = Logger.getLogger(SystemMetadataFactory.class); |
|
76 |
|
|
74 |
|
|
75 |
private static Logger logMetacat = Logger |
|
76 |
.getLogger(SystemMetadataFactory.class); |
|
77 |
|
|
77 | 78 |
/** |
78 |
* Creates a system metadata object for insertion into metacat |
|
79 |
* |
|
80 |
* @param localId The local document identifier |
|
81 |
* @param user The user submitting the system metadata document |
|
82 |
* @param groups The groups the user belongs to |
|
83 |
* |
|
84 |
* @return sysMeta The system metadata object created |
|
85 |
*/ |
|
86 |
public static SystemMetadata createSystemMetadata(String localId) |
|
87 |
throws McdbException, McdbDocNotFoundException, SQLException, |
|
88 |
IOException, AccessionNumberException, ClassNotFoundException, |
|
89 |
InsufficientKarmaException, ParseLSIDException, PropertyNotFoundException, |
|
90 |
BaseException, NoSuchAlgorithmException, JiBXException |
|
91 |
{ |
|
92 |
logMetacat.debug("MetacatHandler.createSystemMetadata() called."); |
|
93 |
logMetacat.debug("provided localId: " + localId); |
|
94 |
|
|
95 |
// create system metadata for the document |
|
96 |
SystemMetadata sysMeta = new SystemMetadata(); |
|
97 |
sysMeta.setSerialVersion(BigInteger.valueOf(1)); |
|
98 |
int rev = IdentifierManager.getInstance().getLatestRevForLocalId(localId); |
|
99 |
AccessionNumber accNum = new AccessionNumber(localId, "NONE"); |
|
100 |
String guid = null; |
|
101 |
try { |
|
102 |
// get the guid if it exists |
|
103 |
guid = IdentifierManager.getInstance().getGUID(accNum.getDocid(), rev); |
|
104 |
} catch ( McdbDocNotFoundException dnfe ) { |
|
105 |
// otherwise create the mapping |
|
106 |
logMetacat.debug("There was a problem getting the guid from " + |
|
107 |
"the given localId (docid and revision). The error message was: " + dnfe.getMessage()); |
|
108 |
logMetacat.debug("No guid in the identifier table. adding it for " + localId); |
|
109 |
IdentifierManager.getInstance().createMapping(localId, localId); |
|
110 |
logMetacat.debug("Mapping created for " + localId); |
|
111 |
logMetacat.debug("accessionNumber: " + accNum); |
|
112 |
guid = IdentifierManager.getInstance().getGUID(accNum.getDocid(), rev); |
|
113 |
} |
|
114 |
Identifier identifier = new Identifier(); |
|
115 |
identifier.setValue(guid); |
|
116 |
|
|
117 |
//set the id |
|
118 |
sysMeta.setIdentifier(identifier); |
|
119 |
|
|
120 |
// get the data or metadata object |
|
121 |
InputStream inputStream; |
|
122 |
try { |
|
123 |
inputStream = MetacatHandler.read(localId); |
|
124 |
} catch ( ParseLSIDException ple ) { |
|
125 |
logMetacat.debug("There was a problem parsing the LSID from " + |
|
126 |
localId + ". The error message was: " + ple.getMessage()); |
|
127 |
throw ple; |
|
128 |
|
|
129 |
} catch ( PropertyNotFoundException pnfe ) { |
|
130 |
logMetacat.debug("There was a problem finding a property. " + |
|
131 |
"The error message was: " + pnfe.getMessage()); |
|
132 |
throw pnfe; |
|
133 |
|
|
134 |
} catch ( McdbException me ) { |
|
135 |
logMetacat.debug("There was a Metacat problem. " + |
|
136 |
"The error message was: " + me.getMessage()); |
|
137 |
throw me; |
|
138 |
|
|
139 |
} catch ( SQLException sqle ) { |
|
140 |
logMetacat.debug("There was a SQL problem. " + |
|
141 |
"The error message was: " + sqle.getMessage()); |
|
142 |
throw sqle; |
|
143 |
|
|
144 |
} catch ( ClassNotFoundException cnfe ) { |
|
145 |
logMetacat.debug("There was a problem finding a class. " + |
|
146 |
"The error message was: " + cnfe.getMessage()); |
|
147 |
throw cnfe; |
|
148 |
|
|
149 |
} catch ( IOException ioe ) { |
|
150 |
logMetacat.debug("There was an I/O exception. " + |
|
151 |
"The error message was: " + ioe.getMessage()); |
|
152 |
throw ioe; |
|
153 |
|
|
154 |
} // end try() |
|
155 |
|
|
156 |
// get additional docinfo |
|
157 |
Hashtable<String, Object> docInfo = IdentifierManager.getInstance().getDocumentInfo(localId); |
|
158 |
//set the default object format |
|
159 |
String doctype = (String) docInfo.get("doctype"); |
|
160 |
ObjectFormatIdentifier fmtid = null; |
|
161 |
|
|
162 |
// set the object format, fall back to defaults |
|
163 |
try { |
|
164 |
fmtid = ObjectFormatCache.getInstance().getFormat(doctype).getFormatId(); |
|
79 |
* Creates a system metadata object for insertion into metacat |
|
80 |
* |
|
81 |
* @param localId |
|
82 |
* The local document identifier |
|
83 |
* @param user |
|
84 |
* The user submitting the system metadata document |
|
85 |
* @param groups |
|
86 |
* The groups the user belongs to |
|
87 |
* |
|
88 |
* @return sysMeta The system metadata object created |
|
89 |
*/ |
|
90 |
public static SystemMetadata createSystemMetadata(String localId) |
|
91 |
throws McdbException, McdbDocNotFoundException, SQLException, |
|
92 |
IOException, AccessionNumberException, ClassNotFoundException, |
|
93 |
InsufficientKarmaException, ParseLSIDException, |
|
94 |
PropertyNotFoundException, BaseException, NoSuchAlgorithmException, |
|
95 |
JiBXException { |
|
96 |
logMetacat.debug("MetacatHandler.createSystemMetadata() called."); |
|
97 |
logMetacat.debug("provided localId: " + localId); |
|
165 | 98 |
|
166 |
} catch (NotFound nfe) { |
|
167 |
|
|
168 |
try { |
|
169 |
// format is not registered, use default |
|
170 |
if (doctype.trim().equals("BIN")) { |
|
171 |
fmtid = ObjectFormatCache.getInstance().getFormat( |
|
172 |
"application/octet-stream").getFormatId(); |
|
173 |
|
|
174 |
} else { |
|
175 |
fmtid = ObjectFormatCache.getInstance().getFormat("text/plain").getFormatId(); |
|
176 |
} |
|
177 |
|
|
178 |
} catch (NotFound nf) { |
|
179 |
logMetacat.error("There was a problem getting the default format " + |
|
180 |
"from the ObjectFormatCache: " + nf.getMessage()); |
|
181 |
throw nf; |
|
182 |
} |
|
99 |
// create system metadata for the document |
|
100 |
SystemMetadata sysMeta = new SystemMetadata(); |
|
101 |
sysMeta.setSerialVersion(BigInteger.valueOf(1)); |
|
102 |
int rev = IdentifierManager.getInstance().getLatestRevForLocalId( |
|
103 |
localId); |
|
104 |
AccessionNumber accNum = new AccessionNumber(localId, "NONE"); |
|
105 |
String guid = null; |
|
106 |
try { |
|
107 |
// get the guid if it exists |
|
108 |
guid = IdentifierManager.getInstance().getGUID(accNum.getDocid(), |
|
109 |
rev); |
|
110 |
} catch (McdbDocNotFoundException dnfe) { |
|
111 |
// otherwise create the mapping |
|
112 |
logMetacat |
|
113 |
.debug("There was a problem getting the guid from " |
|
114 |
+ "the given localId (docid and revision). The error message was: " |
|
115 |
+ dnfe.getMessage()); |
|
116 |
logMetacat.debug("No guid in the identifier table. adding it for " |
|
117 |
+ localId); |
|
118 |
IdentifierManager.getInstance().createMapping(localId, localId); |
|
119 |
logMetacat.debug("Mapping created for " + localId); |
|
120 |
logMetacat.debug("accessionNumber: " + accNum); |
|
121 |
guid = IdentifierManager.getInstance().getGUID(accNum.getDocid(), |
|
122 |
rev); |
|
123 |
} |
|
124 |
Identifier identifier = new Identifier(); |
|
125 |
identifier.setValue(guid); |
|
183 | 126 |
|
184 |
} |
|
185 |
|
|
186 |
sysMeta.setFormatId(fmtid); |
|
187 |
logMetacat.debug("The ObjectFormat for " + localId + " is " + fmtid.getValue()); |
|
188 |
|
|
189 |
// further parse EML documents to get data object format, |
|
190 |
// describes and describedBy information |
|
191 |
if ( fmtid == ObjectFormatCache.getInstance().getFormat("eml://ecoinformatics.org/eml-2.0.0").getFormatId() || |
|
192 |
fmtid == ObjectFormatCache.getInstance().getFormat("eml://ecoinformatics.org/eml-2.0.1").getFormatId() || |
|
193 |
fmtid == ObjectFormatCache.getInstance().getFormat("eml://ecoinformatics.org/eml-2.1.0").getFormatId() || |
|
194 |
fmtid == ObjectFormatCache.getInstance().getFormat("eml://ecoinformatics.org/eml-2.1.1").getFormatId() ) { |
|
195 |
|
|
196 |
try { |
|
197 |
|
|
198 |
DatabaseConnectionPoolInterface connectionPool = MetacatDatabaseConnectionPoolFactory.getDatabaseConnectionPoolInterface(); |
|
199 |
DataManager dataManager = DataManager.getInstance(connectionPool, connectionPool.getDBAdapterName()); |
|
200 |
DataPackage dataPackage = dataManager.parseMetadata(inputStream); |
|
201 |
|
|
202 |
// iterate through the data objects in the EML doc and add sysmeta |
|
203 |
logMetacat.debug("In createSystemMetadata() the number of data " + |
|
204 |
"entities is: " + dataPackage.getEntityNumber()); |
|
205 |
|
|
206 |
// iterate through data objects described by the EML |
|
207 |
for (int j = 0; j < dataPackage.getEntityList().length; j++) { |
|
208 |
|
|
209 |
|
|
210 |
String dataDocUrl = dataPackage.getEntityList()[j].getURL(); |
|
211 |
String dataDocMimeType = dataPackage.getEntityList()[j].getDataFormat(); |
|
212 |
// default to binary |
|
213 |
if (dataDocMimeType == null) { |
|
214 |
dataDocMimeType = |
|
215 |
ObjectFormatCache.getInstance().getFormat("application/octet-stream").getFormatId().getValue(); |
|
216 |
} |
|
217 |
String dataDocLocalId = ""; |
|
218 |
logMetacat.debug("Data local ID: " + dataDocLocalId); |
|
219 |
logMetacat.debug("Data URL : " + dataDocUrl); |
|
220 |
logMetacat.debug("Data mime : " + dataDocMimeType); |
|
221 |
|
|
222 |
//we only handle ecogrid urls right now |
|
223 |
String ecogridPrefix = "ecogrid://knb/"; |
|
224 |
if ( dataDocUrl.trim().startsWith(ecogridPrefix) ) { |
|
225 |
dataDocLocalId = |
|
226 |
dataDocUrl.substring(dataDocUrl.indexOf(ecogridPrefix) + ecogridPrefix.length()); |
|
227 |
|
|
228 |
// look up the guid for the data |
|
229 |
String dataDocid = DocumentUtil.getSmartDocId(dataDocLocalId); |
|
230 |
int dataRev = DocumentUtil.getRevisionFromAccessionNumber(dataDocLocalId); |
|
231 |
|
|
232 |
SystemMetadata dataSysMeta = null; |
|
233 |
// check if data system metadata exists |
|
234 |
String dataGuidString = null; |
|
235 |
try { |
|
236 |
dataGuidString = IdentifierManager.getInstance().getGUID(dataDocid, dataRev); |
|
237 |
dataSysMeta = IdentifierManager.getInstance().getSystemMetadata(dataGuidString); |
|
238 |
} catch ( McdbDocNotFoundException nf ) { |
|
239 |
// System metadata for data doesn't exist yet, so create it |
|
240 |
logMetacat.debug("There was not an existing system metadata " + "document for " + dataDocLocalId); |
|
241 |
try { |
|
242 |
logMetacat.debug("Creating a system metadata " + "document for " + dataDocLocalId); |
|
243 |
dataSysMeta = createSystemMetadata(dataDocLocalId); |
|
244 |
|
|
245 |
// now look it up again |
|
246 |
dataGuidString = IdentifierManager.getInstance().getGUID(dataDocid, dataRev); |
|
127 |
// set the id |
|
128 |
sysMeta.setIdentifier(identifier); |
|
247 | 129 |
|
248 |
//set the guid |
|
249 |
Identifier dataGuid = new Identifier(); |
|
250 |
dataGuid.setValue(dataGuidString); |
|
251 |
|
|
252 |
// set object format |
|
253 |
logMetacat.debug("Updating system metadata for " + dataGuid.getValue() + " to " + dataDocMimeType); |
|
254 |
try { |
|
255 |
ObjectFormatIdentifier fmt = |
|
256 |
ObjectFormatCache.getInstance().getFormat(dataDocMimeType).getFormatId(); |
|
257 |
dataSysMeta.setFormatId(fmt); |
|
258 |
|
|
259 |
} catch (NotFound nfe) { |
|
260 |
logMetacat.debug("Couldn't find format identifier for: " + |
|
261 |
dataDocMimeType + ". Setting it to application/octet-stream."); |
|
262 |
ObjectFormatIdentifier newFmtid = new ObjectFormatIdentifier(); |
|
263 |
newFmtid.setValue("application/octet-stream"); |
|
264 |
|
|
265 |
} |
|
266 |
|
|
267 |
// update the values |
|
268 |
HazelcastService.getInstance().getSystemMetadataMap().put(dataSysMeta.getIdentifier(), dataSysMeta); |
|
269 |
|
|
270 |
} catch ( McdbDocNotFoundException mdnf) { |
|
271 |
mdnf.printStackTrace(); |
|
272 |
throw mdnf; |
|
273 |
} catch ( NumberFormatException nfe) { |
|
274 |
nfe.printStackTrace(); |
|
275 |
throw nfe; |
|
276 |
} catch ( AccessionNumberException ane) { |
|
277 |
ane.printStackTrace(); |
|
278 |
throw ane; |
|
279 |
} catch ( SQLException sqle) { |
|
280 |
sqle.printStackTrace(); |
|
281 |
throw sqle; |
|
282 |
} catch ( NoSuchAlgorithmException nsae) { |
|
283 |
nsae.printStackTrace(); |
|
284 |
throw nsae; |
|
285 |
} catch ( IOException ioe) { |
|
286 |
ioe.printStackTrace(); |
|
287 |
throw ioe; |
|
288 |
} catch ( PropertyNotFoundException pnfe) { |
|
289 |
pnfe.printStackTrace(); |
|
290 |
throw pnfe; |
|
291 |
} catch ( BaseException be) { |
|
292 |
be.printStackTrace(); |
|
293 |
throw be; |
|
294 |
} |
|
295 |
} |
|
296 |
|
|
297 |
} // end if() |
|
298 |
|
|
299 |
} // end for() |
|
300 |
|
|
301 |
} catch ( ParserConfigurationException pce ) { |
|
302 |
logMetacat.debug("There was a problem parsing the EML document. " + |
|
303 |
"The error message was: " + pce.getMessage()); |
|
304 |
|
|
305 |
} catch ( SAXException saxe ) { |
|
306 |
logMetacat.debug("There was a problem traversing the EML document. " + |
|
307 |
"The error message was: " + saxe.getMessage()); |
|
308 |
|
|
309 |
} catch ( XPathExpressionException xpee ) { |
|
310 |
logMetacat.debug("There was a problem searching the EML document. " + |
|
311 |
"The error message was: " + xpee.getMessage()); |
|
312 |
} catch ( Exception e ) { |
|
313 |
logMetacat.debug("There was a problem creating System Metadata. " + |
|
314 |
"The error message was: " + e.getMessage()); |
|
315 |
} // end try() |
|
316 |
|
|
317 |
} // end if() |
|
318 |
|
|
319 |
|
|
320 |
//create the checksum |
|
321 |
inputStream = MetacatHandler.read(localId); |
|
322 |
String algorithm = "MD5"; |
|
323 |
Checksum checksum = ChecksumUtil.checksum(inputStream, algorithm); |
|
324 |
sysMeta.setChecksum(checksum); |
|
325 |
|
|
326 |
//set the size |
|
327 |
inputStream = MetacatHandler.read(localId); |
|
328 |
String sizeStr = new Long(sizeOfStream(inputStream)).toString(); |
|
329 |
sysMeta.setSize(new BigInteger(sizeStr)); |
|
330 |
|
|
331 |
//submitter |
|
332 |
Subject subject = new Subject(); |
|
333 |
subject.setValue((String) docInfo.get("user_owner")); |
|
334 |
sysMeta.setSubmitter(subject); |
|
335 |
sysMeta.setRightsHolder(subject); |
|
130 |
// get the data or metadata object |
|
131 |
InputStream inputStream; |
|
132 |
try { |
|
133 |
inputStream = MetacatHandler.read(localId); |
|
134 |
} catch (ParseLSIDException ple) { |
|
135 |
logMetacat.debug("There was a problem parsing the LSID from " |
|
136 |
+ localId + ". The error message was: " + ple.getMessage()); |
|
137 |
throw ple; |
|
336 | 138 |
|
337 |
try { |
|
338 |
SimpleDateFormat dateFormat = new SimpleDateFormat("yyyy-MM-DD"); |
|
339 |
dateFormat.setTimeZone(TimeZone.getDefault()); |
|
340 |
|
|
341 |
Date dateCreated = dateFormat.parse((String) docInfo.get("date_created")); |
|
342 |
sysMeta.setDateUploaded(dateCreated); |
|
343 |
Date dateUpdated = dateFormat.parse((String) docInfo.get("date_updated")); |
|
344 |
sysMeta.setDateSysMetadataModified(dateUpdated); |
|
345 |
|
|
346 |
} catch (ParseException pe) { |
|
347 |
logMetacat.debug("There was a problem parsing a Metacat date. The " + |
|
348 |
"error message was: " + pe.getMessage()); |
|
349 |
Date dateCreated = new Date(); |
|
350 |
sysMeta.setDateUploaded(dateCreated); |
|
351 |
Date dateUpdated = new Date(); |
|
352 |
sysMeta.setDateSysMetadataModified(dateUpdated); |
|
353 |
|
|
354 |
} |
|
355 |
NodeReference nr = new NodeReference(); |
|
356 |
nr.setValue(PropertyService.getProperty("dataone.memberNodeId")); |
|
357 |
sysMeta.setOriginMemberNode(nr); |
|
358 |
sysMeta.setAuthoritativeMemberNode(nr); |
|
359 |
|
|
360 |
return sysMeta; |
|
361 |
} |
|
362 |
|
|
363 |
/* |
|
364 |
* Find the size (in bytes) of a stream. Note: This needs to refactored out |
|
365 |
* of MetacatHandler and into a utility when stream i/o in Metacat is |
|
366 |
* evaluated. |
|
367 |
* |
|
368 |
* @param is The InputStream of bytes |
|
369 |
* @return size The size in bytes of the input stream as a long |
|
370 |
* @throws IOException |
|
371 |
*/ |
|
372 |
private static long sizeOfStream(InputStream is) |
|
373 |
throws IOException { |
|
374 |
|
|
375 |
long size = 0; |
|
376 |
byte[] b = new byte[1024]; |
|
377 |
int numread = is.read(b, 0, 1024); |
|
378 |
while(numread != -1) |
|
379 |
{ |
|
380 |
size += numread; |
|
381 |
numread = is.read(b, 0, 1024); |
|
382 |
} |
|
383 |
return size; |
|
384 |
|
|
385 |
} |
|
139 |
} catch (PropertyNotFoundException pnfe) { |
|
140 |
logMetacat.debug("There was a problem finding a property. " |
|
141 |
+ "The error message was: " + pnfe.getMessage()); |
|
142 |
throw pnfe; |
|
143 |
|
|
144 |
} catch (McdbException me) { |
|
145 |
logMetacat.debug("There was a Metacat problem. " |
|
146 |
+ "The error message was: " + me.getMessage()); |
|
147 |
throw me; |
|
148 |
|
|
149 |
} catch (SQLException sqle) { |
|
150 |
logMetacat.debug("There was a SQL problem. " |
|
151 |
+ "The error message was: " + sqle.getMessage()); |
|
152 |
throw sqle; |
|
153 |
|
|
154 |
} catch (ClassNotFoundException cnfe) { |
|
155 |
logMetacat.debug("There was a problem finding a class. " |
|
156 |
+ "The error message was: " + cnfe.getMessage()); |
|
157 |
throw cnfe; |
|
158 |
|
|
159 |
} catch (IOException ioe) { |
|
160 |
logMetacat.debug("There was an I/O exception. " |
|
161 |
+ "The error message was: " + ioe.getMessage()); |
|
162 |
throw ioe; |
|
163 |
|
|
164 |
} // end try() |
|
165 |
|
|
166 |
// get additional docinfo |
|
167 |
Hashtable<String, Object> docInfo = IdentifierManager.getInstance() |
|
168 |
.getDocumentInfo(localId); |
|
169 |
// set the default object format |
|
170 |
String doctype = (String) docInfo.get("doctype"); |
|
171 |
ObjectFormatIdentifier fmtid = null; |
|
172 |
|
|
173 |
// set the object format, fall back to defaults |
|
174 |
try { |
|
175 |
fmtid = ObjectFormatCache.getInstance().getFormat(doctype) |
|
176 |
.getFormatId(); |
|
177 |
|
|
178 |
} catch (NotFound nfe) { |
|
179 |
|
|
180 |
try { |
|
181 |
// format is not registered, use default |
|
182 |
if (doctype.trim().equals("BIN")) { |
|
183 |
fmtid = ObjectFormatCache.getInstance().getFormat( |
|
184 |
"application/octet-stream").getFormatId(); |
|
185 |
|
|
186 |
} else { |
|
187 |
fmtid = ObjectFormatCache.getInstance().getFormat( |
|
188 |
"text/plain").getFormatId(); |
|
189 |
} |
|
190 |
|
|
191 |
} catch (NotFound nf) { |
|
192 |
logMetacat |
|
193 |
.error("There was a problem getting the default format " |
|
194 |
+ "from the ObjectFormatCache: " |
|
195 |
+ nf.getMessage()); |
|
196 |
throw nf; |
|
197 |
} |
|
198 |
|
|
199 |
} |
|
200 |
|
|
201 |
sysMeta.setFormatId(fmtid); |
|
202 |
logMetacat.debug("The ObjectFormat for " + localId + " is " |
|
203 |
+ fmtid.getValue()); |
|
204 |
|
|
205 |
// further parse EML documents to get data object format, |
|
206 |
// describes and describedBy information |
|
207 |
if (fmtid == ObjectFormatCache.getInstance().getFormat( |
|
208 |
"eml://ecoinformatics.org/eml-2.0.0").getFormatId() |
|
209 |
|| fmtid == ObjectFormatCache.getInstance().getFormat( |
|
210 |
"eml://ecoinformatics.org/eml-2.0.1").getFormatId() |
|
211 |
|| fmtid == ObjectFormatCache.getInstance().getFormat( |
|
212 |
"eml://ecoinformatics.org/eml-2.1.0").getFormatId() |
|
213 |
|| fmtid == ObjectFormatCache.getInstance().getFormat( |
|
214 |
"eml://ecoinformatics.org/eml-2.1.1").getFormatId()) { |
|
215 |
|
|
216 |
try { |
|
217 |
|
|
218 |
DatabaseConnectionPoolInterface connectionPool = MetacatDatabaseConnectionPoolFactory |
|
219 |
.getDatabaseConnectionPoolInterface(); |
|
220 |
DataManager dataManager = DataManager.getInstance( |
|
221 |
connectionPool, connectionPool.getDBAdapterName()); |
|
222 |
DataPackage dataPackage = dataManager |
|
223 |
.parseMetadata(inputStream); |
|
224 |
|
|
225 |
// iterate through the data objects in the EML doc and add |
|
226 |
// sysmeta |
|
227 |
logMetacat |
|
228 |
.debug("In createSystemMetadata() the number of data " |
|
229 |
+ "entities is: " |
|
230 |
+ dataPackage.getEntityNumber()); |
|
231 |
|
|
232 |
// iterate through data objects described by the EML |
|
233 |
for (int j = 0; j < dataPackage.getEntityList().length; j++) { |
|
234 |
|
|
235 |
String dataDocUrl = dataPackage.getEntityList()[j].getURL(); |
|
236 |
String dataDocMimeType = dataPackage.getEntityList()[j] |
|
237 |
.getDataFormat(); |
|
238 |
// default to binary |
|
239 |
if (dataDocMimeType == null) { |
|
240 |
dataDocMimeType = ObjectFormatCache.getInstance() |
|
241 |
.getFormat("application/octet-stream") |
|
242 |
.getFormatId().getValue(); |
|
243 |
} |
|
244 |
String dataDocLocalId = ""; |
|
245 |
logMetacat.debug("Data local ID: " + dataDocLocalId); |
|
246 |
logMetacat.debug("Data URL : " + dataDocUrl); |
|
247 |
logMetacat.debug("Data mime : " + dataDocMimeType); |
|
248 |
|
|
249 |
// we only handle ecogrid urls right now |
|
250 |
String ecogridPrefix = "ecogrid://knb/"; |
|
251 |
if (dataDocUrl.trim().startsWith(ecogridPrefix)) { |
|
252 |
dataDocLocalId = dataDocUrl.substring(dataDocUrl |
|
253 |
.indexOf(ecogridPrefix) |
|
254 |
+ ecogridPrefix.length()); |
|
255 |
|
|
256 |
// look up the guid for the data |
|
257 |
String dataDocid = DocumentUtil |
|
258 |
.getSmartDocId(dataDocLocalId); |
|
259 |
int dataRev = DocumentUtil |
|
260 |
.getRevisionFromAccessionNumber(dataDocLocalId); |
|
261 |
|
|
262 |
SystemMetadata dataSysMeta = null; |
|
263 |
// check if data system metadata exists |
|
264 |
String dataGuidString = null; |
|
265 |
try { |
|
266 |
dataGuidString = IdentifierManager.getInstance() |
|
267 |
.getGUID(dataDocid, dataRev); |
|
268 |
dataSysMeta = IdentifierManager.getInstance() |
|
269 |
.getSystemMetadata(dataGuidString); |
|
270 |
} catch (McdbDocNotFoundException nf) { |
|
271 |
// System metadata for data doesn't exist yet, so |
|
272 |
// create it |
|
273 |
logMetacat |
|
274 |
.debug("There was not an existing system metadata " |
|
275 |
+ "document for " + dataDocLocalId); |
|
276 |
try { |
|
277 |
logMetacat.debug("Creating a system metadata " |
|
278 |
+ "document for " + dataDocLocalId); |
|
279 |
dataSysMeta = createSystemMetadata(dataDocLocalId); |
|
280 |
|
|
281 |
// now look it up again |
|
282 |
dataGuidString = IdentifierManager |
|
283 |
.getInstance().getGUID(dataDocid, |
|
284 |
dataRev); |
|
285 |
|
|
286 |
// set the guid |
|
287 |
Identifier dataGuid = new Identifier(); |
|
288 |
dataGuid.setValue(dataGuidString); |
|
289 |
|
|
290 |
// set object format |
|
291 |
logMetacat |
|
292 |
.debug("Updating system metadata for " |
|
293 |
+ dataGuid.getValue() + " to " |
|
294 |
+ dataDocMimeType); |
|
295 |
try { |
|
296 |
ObjectFormatIdentifier fmt = ObjectFormatCache |
|
297 |
.getInstance().getFormat( |
|
298 |
dataDocMimeType) |
|
299 |
.getFormatId(); |
|
300 |
dataSysMeta.setFormatId(fmt); |
|
301 |
|
|
302 |
} catch (NotFound nfe) { |
|
303 |
logMetacat |
|
304 |
.debug("Couldn't find format identifier for: " |
|
305 |
+ dataDocMimeType |
|
306 |
+ ". Setting it to application/octet-stream."); |
|
307 |
ObjectFormatIdentifier newFmtid = new ObjectFormatIdentifier(); |
|
308 |
newFmtid |
|
309 |
.setValue("application/octet-stream"); |
|
310 |
|
|
311 |
} |
|
312 |
|
|
313 |
// update the values |
|
314 |
HazelcastService.getInstance() |
|
315 |
.getSystemMetadataMap().put( |
|
316 |
dataSysMeta.getIdentifier(), |
|
317 |
dataSysMeta); |
|
318 |
|
|
319 |
} catch (McdbDocNotFoundException mdnf) { |
|
320 |
mdnf.printStackTrace(); |
|
321 |
throw mdnf; |
|
322 |
} catch (NumberFormatException nfe) { |
|
323 |
nfe.printStackTrace(); |
|
324 |
throw nfe; |
|
325 |
} catch (AccessionNumberException ane) { |
|
326 |
ane.printStackTrace(); |
|
327 |
throw ane; |
|
328 |
} catch (SQLException sqle) { |
|
329 |
sqle.printStackTrace(); |
|
330 |
throw sqle; |
|
331 |
} catch (NoSuchAlgorithmException nsae) { |
|
332 |
nsae.printStackTrace(); |
|
333 |
throw nsae; |
|
334 |
} catch (IOException ioe) { |
|
335 |
ioe.printStackTrace(); |
|
336 |
throw ioe; |
|
337 |
} catch (PropertyNotFoundException pnfe) { |
|
338 |
pnfe.printStackTrace(); |
|
339 |
throw pnfe; |
|
340 |
} catch (BaseException be) { |
|
341 |
be.printStackTrace(); |
|
342 |
throw be; |
|
343 |
} |
|
344 |
} |
|
345 |
|
|
346 |
} // end if() |
|
347 |
|
|
348 |
} // end for() |
|
349 |
|
|
350 |
} catch (ParserConfigurationException pce) { |
|
351 |
logMetacat |
|
352 |
.debug("There was a problem parsing the EML document. " |
|
353 |
+ "The error message was: " + pce.getMessage()); |
|
354 |
|
|
355 |
} catch (SAXException saxe) { |
|
356 |
logMetacat |
|
357 |
.debug("There was a problem traversing the EML document. " |
|
358 |
+ "The error message was: " + saxe.getMessage()); |
|
359 |
|
|
360 |
} catch (XPathExpressionException xpee) { |
|
361 |
logMetacat |
|
362 |
.debug("There was a problem searching the EML document. " |
|
363 |
+ "The error message was: " + xpee.getMessage()); |
|
364 |
} catch (Exception e) { |
|
365 |
logMetacat |
|
366 |
.debug("There was a problem creating System Metadata. " |
|
367 |
+ "The error message was: " + e.getMessage()); |
|
368 |
} // end try() |
|
369 |
|
|
370 |
} // end if() |
|
371 |
|
|
372 |
// create the checksum |
|
373 |
inputStream = MetacatHandler.read(localId); |
|
374 |
String algorithm = "MD5"; |
|
375 |
Checksum checksum = ChecksumUtil.checksum(inputStream, algorithm); |
|
376 |
sysMeta.setChecksum(checksum); |
|
377 |
|
|
378 |
// set the size |
|
379 |
inputStream = MetacatHandler.read(localId); |
|
380 |
String sizeStr = new Long(sizeOfStream(inputStream)).toString(); |
|
381 |
sysMeta.setSize(new BigInteger(sizeStr)); |
|
382 |
|
|
383 |
// submitter |
|
384 |
Subject subject = new Subject(); |
|
385 |
subject.setValue((String) docInfo.get("user_owner")); |
|
386 |
sysMeta.setSubmitter(subject); |
|
387 |
sysMeta.setRightsHolder(subject); |
|
388 |
|
|
389 |
try { |
|
390 |
SimpleDateFormat dateFormat = new SimpleDateFormat("yyyy-MM-DD"); |
|
391 |
dateFormat.setTimeZone(TimeZone.getDefault()); |
|
392 |
|
|
393 |
Date dateCreated = dateFormat.parse((String) docInfo |
|
394 |
.get("date_created")); |
|
395 |
sysMeta.setDateUploaded(dateCreated); |
|
396 |
Date dateUpdated = dateFormat.parse((String) docInfo |
|
397 |
.get("date_updated")); |
|
398 |
sysMeta.setDateSysMetadataModified(dateUpdated); |
|
399 |
|
|
400 |
} catch (ParseException pe) { |
|
401 |
logMetacat.debug("There was a problem parsing a Metacat date. The " |
|
402 |
+ "error message was: " + pe.getMessage()); |
|
403 |
Date dateCreated = new Date(); |
|
404 |
sysMeta.setDateUploaded(dateCreated); |
|
405 |
Date dateUpdated = new Date(); |
|
406 |
sysMeta.setDateSysMetadataModified(dateUpdated); |
|
407 |
|
|
408 |
} |
|
409 |
NodeReference nr = new NodeReference(); |
|
410 |
nr.setValue(PropertyService.getProperty("dataone.memberNodeId")); |
|
411 |
sysMeta.setOriginMemberNode(nr); |
|
412 |
sysMeta.setAuthoritativeMemberNode(nr); |
|
413 |
|
|
414 |
return sysMeta; |
|
415 |
} |
|
416 |
|
|
417 |
/* |
|
418 |
* Find the size (in bytes) of a stream. Note: This needs to refactored out |
|
419 |
* of MetacatHandler and into a utility when stream i/o in Metacat is |
|
420 |
* evaluated. |
|
421 |
* |
|
422 |
* @param is The InputStream of bytes |
|
423 |
* |
|
424 |
* @return size The size in bytes of the input stream as a long |
|
425 |
* |
|
426 |
* @throws IOException |
|
427 |
*/ |
|
428 |
private static long sizeOfStream(InputStream is) throws IOException { |
|
429 |
|
|
430 |
long size = 0; |
|
431 |
byte[] b = new byte[1024]; |
|
432 |
int numread = is.read(b, 0, 1024); |
|
433 |
while (numread != -1) { |
|
434 |
size += numread; |
|
435 |
numread = is.read(b, 0, 1024); |
|
436 |
} |
|
437 |
return size; |
|
438 |
|
|
439 |
} |
|
386 | 440 |
} |
Also available in: Unified diff
reformat code -- no changes