|
1
|
/**
|
|
2
|
* Copyright (C) 2009 EDIT
|
|
3
|
* European Distributed Institute of Taxonomy
|
|
4
|
* http://www.e-taxonomy.eu
|
|
5
|
*
|
|
6
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
|
7
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
|
8
|
*/
|
|
9
|
package eu.etaxonomy.cdm.io.pesi.out;
|
|
10
|
|
|
11
|
import java.sql.SQLException;
|
|
12
|
import java.util.ArrayList;
|
|
13
|
import java.util.HashSet;
|
|
14
|
import java.util.List;
|
|
15
|
import java.util.Set;
|
|
16
|
|
|
17
|
import org.apache.commons.lang3.StringUtils;
|
|
18
|
import org.apache.log4j.Logger;
|
|
19
|
import org.springframework.stereotype.Component;
|
|
20
|
import org.springframework.transaction.TransactionStatus;
|
|
21
|
|
|
22
|
import eu.etaxonomy.cdm.io.common.IExportConfigurator.DO_REFERENCES;
|
|
23
|
import eu.etaxonomy.cdm.io.common.Source;
|
|
24
|
import eu.etaxonomy.cdm.io.common.mapping.out.DbAnnotationMapper;
|
|
25
|
import eu.etaxonomy.cdm.io.common.mapping.out.DbDoiMapper;
|
|
26
|
import eu.etaxonomy.cdm.io.common.mapping.out.DbExtensionMapper;
|
|
27
|
import eu.etaxonomy.cdm.io.common.mapping.out.DbStringMapper;
|
|
28
|
import eu.etaxonomy.cdm.io.common.mapping.out.DbTimePeriodMapper;
|
|
29
|
import eu.etaxonomy.cdm.io.common.mapping.out.IdMapper;
|
|
30
|
import eu.etaxonomy.cdm.io.common.mapping.out.MethodMapper;
|
|
31
|
import eu.etaxonomy.cdm.io.pesi.erms.ErmsTransformer;
|
|
32
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
|
33
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
|
34
|
import eu.etaxonomy.cdm.model.common.Extension;
|
|
35
|
import eu.etaxonomy.cdm.model.common.ExtensionType;
|
|
36
|
import eu.etaxonomy.cdm.model.common.IdentifiableSource;
|
|
37
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
|
38
|
import eu.etaxonomy.cdm.model.reference.ReferenceType;
|
|
39
|
|
|
40
|
/**
|
|
41
|
* The export class for {@link eu.etaxonomy.cdm.model.reference.Reference References}.<p>
|
|
42
|
* Inserts into DataWarehouse database table <code>Source</code>.
|
|
43
|
*
|
|
44
|
* @author e.-m.lee
|
|
45
|
* @since 11.02.2010
|
|
46
|
*/
|
|
47
|
@Component
|
|
48
|
public class PesiSourceExport extends PesiExportBase {
|
|
49
|
|
|
50
|
private static final long serialVersionUID = -3084883718722120651L;
|
|
51
|
private static final Logger logger = Logger.getLogger(PesiSourceExport.class);
|
|
52
|
|
|
53
|
private static final Class<? extends CdmBase> standardMethodParameter = Reference.class;
|
|
54
|
|
|
55
|
private static int modCount = 1000;
|
|
56
|
public static final String dbTableName = "Source";
|
|
57
|
private static final String pluralString = "Sources";
|
|
58
|
List<Integer> storedSourceIds = new ArrayList<>();
|
|
59
|
|
|
60
|
public PesiSourceExport() {
|
|
61
|
super();
|
|
62
|
}
|
|
63
|
|
|
64
|
@Override
|
|
65
|
public Class<? extends CdmBase> getStandardMethodParameter() {
|
|
66
|
return standardMethodParameter;
|
|
67
|
}
|
|
68
|
|
|
69
|
/**
|
|
70
|
* Checks whether a sourceId was stored already.
|
|
71
|
* @param sourceId
|
|
72
|
* @return
|
|
73
|
*/
|
|
74
|
protected boolean isStoredSourceId(Integer sourceId) {
|
|
75
|
if (storedSourceIds.contains(sourceId)) {
|
|
76
|
return true;
|
|
77
|
} else {
|
|
78
|
return false;
|
|
79
|
}
|
|
80
|
}
|
|
81
|
|
|
82
|
/**
|
|
83
|
* Adds a sourceId to the list of storedSourceIds.
|
|
84
|
* @param sourceId
|
|
85
|
*/
|
|
86
|
protected void addToStoredSourceIds(Integer sourceId) {
|
|
87
|
if (! storedSourceIds.contains(sourceId)) {
|
|
88
|
this.storedSourceIds.add(sourceId);
|
|
89
|
}
|
|
90
|
}
|
|
91
|
|
|
92
|
@Override
|
|
93
|
protected void doInvoke(PesiExportState state) {
|
|
94
|
try{
|
|
95
|
logger.info("*** Started Making " + pluralString + " ...");
|
|
96
|
|
|
97
|
PesiExportConfigurator pesiExportConfigurator = state.getConfig();
|
|
98
|
|
|
99
|
// Get the limit for objects to save within a single transaction.
|
|
100
|
int limit = pesiExportConfigurator.getLimitSave();
|
|
101
|
|
|
102
|
// Stores whether this invoke was successful or not.
|
|
103
|
boolean success = true ;
|
|
104
|
|
|
105
|
// PESI: Clear the database table Source.
|
|
106
|
//doDelete(state); -> done by stored procedure
|
|
107
|
|
|
108
|
// Get specific mappings: (CDM) Reference -> (PESI) Source
|
|
109
|
PesiExportMapping mapping = getMapping();
|
|
110
|
|
|
111
|
// Initialize the db mapper
|
|
112
|
mapping.initialize(state);
|
|
113
|
|
|
114
|
// Create the Sources
|
|
115
|
int count = 0;
|
|
116
|
int pastCount = 0;
|
|
117
|
TransactionStatus txStatus = null;
|
|
118
|
List<Reference> list = null;
|
|
119
|
|
|
120
|
// logger.error("PHASE 1...");
|
|
121
|
// Start transaction
|
|
122
|
txStatus = startTransaction(true);
|
|
123
|
logger.info("Started new transaction. Fetching some " + pluralString + " (max: " + limit + ") ...");
|
|
124
|
while ((list = getReferenceService().list(null, limit, count, null, null)).size() > 0) {
|
|
125
|
|
|
126
|
logger.debug("Fetched " + list.size() + " " + pluralString + ". Exporting...");
|
|
127
|
for (Reference reference : list) {
|
|
128
|
doCount(count++, modCount, pluralString);
|
|
129
|
success &= mapping.invoke(reference);
|
|
130
|
}
|
|
131
|
|
|
132
|
// Commit transaction
|
|
133
|
commitTransaction(txStatus);
|
|
134
|
logger.debug("Committed transaction.");
|
|
135
|
logger.info("Exported " + (count - pastCount) + " " + pluralString + ". Total: " + count);
|
|
136
|
pastCount = count;
|
|
137
|
|
|
138
|
// Start transaction
|
|
139
|
txStatus = startTransaction(true);
|
|
140
|
if (logger.isDebugEnabled()) {
|
|
141
|
logger.debug("Started new transaction. Fetching some " + pluralString + " (max: " + limit + ") ...");
|
|
142
|
}
|
|
143
|
}
|
|
144
|
if (list.size() == 0) {
|
|
145
|
logger.info("No " + pluralString + " left to fetch.");
|
|
146
|
}
|
|
147
|
// Commit transaction
|
|
148
|
commitTransaction(txStatus);
|
|
149
|
logger.info("Committed transaction.");
|
|
150
|
|
|
151
|
logger.info("*** Finished Making " + pluralString + " ..." + getSuccessString(success));
|
|
152
|
|
|
153
|
if (!success){
|
|
154
|
state.getResult().addError("An error occurred in PesiSourceExport.invoke");
|
|
155
|
}
|
|
156
|
return;
|
|
157
|
} catch (SQLException e) {
|
|
158
|
e.printStackTrace();
|
|
159
|
logger.error(e.getMessage());
|
|
160
|
state.getResult().addException(e);
|
|
161
|
return;
|
|
162
|
}
|
|
163
|
}
|
|
164
|
|
|
165
|
/**
|
|
166
|
* Deletes all entries of database tables related to <code>Source</code>.
|
|
167
|
* @param state The {@link PesiExportState PesiExportState}.
|
|
168
|
* @return Whether the delete operation was successful or not.
|
|
169
|
*/
|
|
170
|
protected boolean doDelete(PesiExportState state) {
|
|
171
|
PesiExportConfigurator pesiConfig = state.getConfig();
|
|
172
|
|
|
173
|
String sql;
|
|
174
|
Source destination = pesiConfig.getDestination();
|
|
175
|
|
|
176
|
// Clear Occurrences
|
|
177
|
sql = "DELETE FROM Occurrence";
|
|
178
|
destination.setQuery(sql);
|
|
179
|
destination.update(sql);
|
|
180
|
|
|
181
|
// Clear Taxa
|
|
182
|
sql = "DELETE FROM Taxon";
|
|
183
|
destination.setQuery(sql);
|
|
184
|
destination.update(sql);
|
|
185
|
|
|
186
|
// Clear Sources
|
|
187
|
sql = "DELETE FROM " + dbTableName;
|
|
188
|
destination.setQuery(sql);
|
|
189
|
destination.update(sql);
|
|
190
|
|
|
191
|
return true;
|
|
192
|
}
|
|
193
|
|
|
194
|
/**
|
|
195
|
* Returns the <code>IMIS_Id</code> attribute.
|
|
196
|
* @param reference The {@link Reference Reference}.
|
|
197
|
* @return The <code>IMIS_Id</code> attribute.
|
|
198
|
* @see MethodMapper
|
|
199
|
*/
|
|
200
|
@SuppressWarnings("unused")
|
|
201
|
private static Integer getIMIS_Id(Reference reference) {
|
|
202
|
return null;
|
|
203
|
}
|
|
204
|
|
|
205
|
/**
|
|
206
|
* Returns the <code>SourceCategoryFK</code> attribute.
|
|
207
|
* @param reference The {@link Reference Reference}.
|
|
208
|
* @return The <code>SourceCategoryFK</code> attribute.
|
|
209
|
* @see MethodMapper
|
|
210
|
*/
|
|
211
|
@SuppressWarnings("unused")
|
|
212
|
private static Integer getSourceCategoryFK(Reference reference) {
|
|
213
|
Integer result = null;
|
|
214
|
try {
|
|
215
|
result = PesiTransformer.reference2SourceCategoryFK(reference);
|
|
216
|
} catch (Exception e) {
|
|
217
|
e.printStackTrace();
|
|
218
|
}
|
|
219
|
return result;
|
|
220
|
}
|
|
221
|
|
|
222
|
/**
|
|
223
|
* Returns the <code>SourceCategoryCache</code> attribute.
|
|
224
|
* @param reference The {@link Reference Reference}.
|
|
225
|
* @return The <code>SourceCategoryCache</code> attribute.
|
|
226
|
* @see MethodMapper
|
|
227
|
*/
|
|
228
|
@SuppressWarnings("unused")
|
|
229
|
private static String getSourceCategoryCache(Reference reference, PesiExportState state) {
|
|
230
|
return state.getTransformer().getCacheByReference(reference);
|
|
231
|
}
|
|
232
|
|
|
233
|
/**
|
|
234
|
* Returns the <code>Name</code> attribute. The corresponding CDM attribute is <code>title</code>.
|
|
235
|
* @param reference The {@link Reference Reference}.
|
|
236
|
* @return The <code>Name</code> attribute.
|
|
237
|
* @see MethodMapper
|
|
238
|
*/
|
|
239
|
@SuppressWarnings("unused")
|
|
240
|
private static String getName(Reference reference) {
|
|
241
|
if (reference != null) {
|
|
242
|
//TODO performance, maybe getExtensions only if titleCache ends with ... or if it has a minimum size, e.g. > 250 or >795
|
|
243
|
String extensionStr = getExtension(reference, ErmsTransformer.uuidExtTruncatedCache);
|
|
244
|
if (StringUtils.isNotBlank(extensionStr)){
|
|
245
|
return extensionStr;
|
|
246
|
}else{
|
|
247
|
return reference.getTitleCache(); // was getTitle()
|
|
248
|
}
|
|
249
|
} else {
|
|
250
|
return null;
|
|
251
|
}
|
|
252
|
}
|
|
253
|
|
|
254
|
|
|
255
|
/**
|
|
256
|
* Returns the <code>AuthorString</code> attribute. The corresponding CDM attribute is the <code>titleCache</code> of an <code>authorTeam</code>.
|
|
257
|
* @param reference The {@link Reference Reference}.
|
|
258
|
* @return The <code>AuthorString</code> attribute.
|
|
259
|
* @see MethodMapper
|
|
260
|
*/
|
|
261
|
@SuppressWarnings("unused")
|
|
262
|
private static String getAuthorString(Reference reference) {
|
|
263
|
String result = null;
|
|
264
|
Set<Extension> extensions;
|
|
265
|
try {
|
|
266
|
if (reference != null) {
|
|
267
|
TeamOrPersonBase<?> team = reference.getAuthorship();
|
|
268
|
if (team != null) {
|
|
269
|
result = team.getTitleCache();
|
|
270
|
}else if (!reference.getExtensions(ErmsTransformer.uuidExtAuthor).isEmpty()){
|
|
271
|
result = reference.getExtensions(ErmsTransformer.uuidExtAuthor).iterator().next();
|
|
272
|
}
|
|
273
|
}
|
|
274
|
} catch (Exception e) {
|
|
275
|
e.printStackTrace();
|
|
276
|
}
|
|
277
|
|
|
278
|
return result;
|
|
279
|
}
|
|
280
|
|
|
281
|
/**
|
|
282
|
* Returns the <code>NomRefCache</code> attribute. The corresponding CDM attribute is <code>titleCache</code>.
|
|
283
|
* @param reference The {@link Reference Reference}.
|
|
284
|
* @return The <code>NomRefCache</code> attribute.
|
|
285
|
* @see MethodMapper
|
|
286
|
*/
|
|
287
|
@SuppressWarnings("unused")
|
|
288
|
private static String getNomRefCache(Reference reference) {
|
|
289
|
return null;
|
|
290
|
// if (reference != null) {
|
|
291
|
// return reference.getTitleCache();
|
|
292
|
// } else {
|
|
293
|
// return null;
|
|
294
|
// }
|
|
295
|
}
|
|
296
|
|
|
297
|
/**
|
|
298
|
* Returns the <code>RefIdInSource</code> attribute.
|
|
299
|
* @param reference The {@link Reference Reference}.
|
|
300
|
* @return The <code>RefIdInSource</code> attribute.
|
|
301
|
* @see MethodMapper
|
|
302
|
*/
|
|
303
|
@SuppressWarnings("unused")
|
|
304
|
private static String getRefIdInSource(Reference reference) {
|
|
305
|
String result = null;
|
|
306
|
|
|
307
|
try {
|
|
308
|
if (reference != null) {
|
|
309
|
Set<IdentifiableSource> sourceAll = reference.getSources();
|
|
310
|
Set<IdentifiableSource> sourceCandidates = filterOriginalPesiDbSources(sourceAll);
|
|
311
|
|
|
312
|
if (sourceCandidates.size() == 1) {
|
|
313
|
result = sourceCandidates.iterator().next().getIdInSource();
|
|
314
|
} else if (sourceCandidates.size() > 1) {
|
|
315
|
logger.warn("Reference for RefIdInSource has multiple IdentifiableSources which are candidates for a PESI originalDbSource. RefIdInSource can't be determined correctly and will be left out: " + reference.getUuid() + " (" + reference.getTitleCache() + ")");
|
|
316
|
int count = 1;
|
|
317
|
// for (IdentifiableSource source : sources) {
|
|
318
|
// result += source.getIdInSource();
|
|
319
|
// if (count < sources.size()) {
|
|
320
|
// result += "; ";
|
|
321
|
// }
|
|
322
|
// count++;
|
|
323
|
// }
|
|
324
|
}
|
|
325
|
}
|
|
326
|
} catch (Exception e) {
|
|
327
|
e.printStackTrace();
|
|
328
|
}
|
|
329
|
|
|
330
|
return result;
|
|
331
|
}
|
|
332
|
|
|
333
|
private static Set<IdentifiableSource> filterOriginalPesiDbSources(
|
|
334
|
Set<IdentifiableSource> sourceAll) {
|
|
335
|
Set<IdentifiableSource> sourceCandidates = new HashSet<IdentifiableSource>();
|
|
336
|
for (IdentifiableSource source : sourceAll){
|
|
337
|
if (isOriginalPesiDbSource(source)){
|
|
338
|
sourceCandidates.add(source);
|
|
339
|
}
|
|
340
|
}
|
|
341
|
return sourceCandidates;
|
|
342
|
}
|
|
343
|
|
|
344
|
private static boolean isOriginalPesiDbSource(IdentifiableSource source) {
|
|
345
|
return (source.getCitation() != null) &&
|
|
346
|
source.getCitation().getType().equals(ReferenceType.Database);
|
|
347
|
}
|
|
348
|
|
|
349
|
/**
|
|
350
|
* Returns the <code>OriginalDB</code> attribute. The corresponding CDM attribute is the <code>titleCache</code> of a <code>citation</code>.
|
|
351
|
* @param reference The {@link Reference Reference}.
|
|
352
|
* @return The <code>OriginalDB</code> attribute.
|
|
353
|
* @see MethodMapper
|
|
354
|
*/
|
|
355
|
@SuppressWarnings("unused")
|
|
356
|
private static String getOriginalDB(Reference reference) {
|
|
357
|
String result = "";
|
|
358
|
|
|
359
|
try {
|
|
360
|
if (reference != null) {
|
|
361
|
Set<IdentifiableSource> sourcesAll = reference.getSources();
|
|
362
|
Set<IdentifiableSource> sourceCandidates = filterOriginalPesiDbSources(sourcesAll);
|
|
363
|
|
|
364
|
if (sourceCandidates.size() == 1) {
|
|
365
|
Reference citation = sourceCandidates.iterator().next().getCitation();
|
|
366
|
if (citation != null) {
|
|
367
|
result = PesiTransformer.databaseString2Abbreviation(citation.getTitleCache()); //or just title
|
|
368
|
} else {
|
|
369
|
logger.warn("OriginalDB can not be determined because the citation of this source is NULL: " + sourceCandidates.iterator().next().getUuid());
|
|
370
|
}
|
|
371
|
} else if (sourceCandidates.size() > 1) {
|
|
372
|
logger.warn("Taxon has multiple IdentifiableSources: " + reference.getUuid() + " (" + reference.getTitleCache() + ")");
|
|
373
|
int count = 1;
|
|
374
|
for (IdentifiableSource source : sourceCandidates) {
|
|
375
|
Reference citation = source.getCitation();
|
|
376
|
if (citation != null) {
|
|
377
|
result += PesiTransformer.databaseString2Abbreviation(citation.getTitleCache());
|
|
378
|
if (count < sourceCandidates.size()) {
|
|
379
|
result += "; ";
|
|
380
|
}
|
|
381
|
count++;
|
|
382
|
}
|
|
383
|
}
|
|
384
|
} else {
|
|
385
|
result = null;
|
|
386
|
}
|
|
387
|
}
|
|
388
|
} catch (Exception e) {
|
|
389
|
e.printStackTrace();
|
|
390
|
}
|
|
391
|
|
|
392
|
return result;
|
|
393
|
}
|
|
394
|
|
|
395
|
@Override
|
|
396
|
protected boolean doCheck(PesiExportState state) {
|
|
397
|
return true;
|
|
398
|
}
|
|
399
|
|
|
400
|
@Override
|
|
401
|
protected boolean isIgnore(PesiExportState state) {
|
|
402
|
return ! state.getConfig().getDoReferences().equals(DO_REFERENCES.ALL);
|
|
403
|
}
|
|
404
|
|
|
405
|
/**
|
|
406
|
* Returns the CDM to PESI specific export mappings.
|
|
407
|
* @return The {@link PesiExportMapping PesiExportMapping}.
|
|
408
|
*/
|
|
409
|
private PesiExportMapping getMapping() {
|
|
410
|
PesiExportMapping mapping = new PesiExportMapping(dbTableName);
|
|
411
|
|
|
412
|
mapping.addMapper(IdMapper.NewInstance("SourceId"));
|
|
413
|
|
|
414
|
// IMIS_Id
|
|
415
|
ExtensionType imisExtensionType = (ExtensionType)getTermService().find(ErmsTransformer.uuidExtImis);
|
|
416
|
if (imisExtensionType != null) {
|
|
417
|
mapping.addMapper(DbExtensionMapper.NewInstance(imisExtensionType, "IMIS_Id"));
|
|
418
|
} else {
|
|
419
|
mapping.addMapper(MethodMapper.NewInstance("IMIS_Id", this));
|
|
420
|
}
|
|
421
|
|
|
422
|
mapping.addMapper(MethodMapper.NewInstance("SourceCategoryFK", this));
|
|
423
|
mapping.addMapper(MethodMapper.NewInstance("SourceCategoryCache", this, Reference.class, PesiExportState.class));
|
|
424
|
mapping.addMapper(MethodMapper.NewInstance("Name", this));
|
|
425
|
mapping.addMapper(DbStringMapper.NewInstance("referenceAbstract", "Abstract"));
|
|
426
|
mapping.addMapper(DbStringMapper.NewInstance("title", "Title"));
|
|
427
|
mapping.addMapper(MethodMapper.NewInstance("AuthorString", this));
|
|
428
|
mapping.addMapper(DbTimePeriodMapper.NewInstance("datePublished", "RefYear"));
|
|
429
|
mapping.addMapper(DbDoiMapper.NewInstance("doi", "Doi"));
|
|
430
|
|
|
431
|
mapping.addMapper(MethodMapper.NewInstance("NomRefCache", this));
|
|
432
|
mapping.addMapper(DbExtensionMapper.NewInstance(ExtensionType.URL(), "Link"));
|
|
433
|
mapping.addMapper(DbAnnotationMapper.NewInstance((String)null, "Notes"));
|
|
434
|
mapping.addMapper(MethodMapper.NewInstance("RefIdInSource", this));
|
|
435
|
mapping.addMapper(MethodMapper.NewInstance("OriginalDB", this));
|
|
436
|
|
|
437
|
return mapping;
|
|
438
|
}
|
|
439
|
}
|