1
|
/**
|
2
|
* Copyright (C) 2007 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
|
|
10
|
package eu.etaxonomy.cdm.io.faunaEuropaea;
|
11
|
|
12
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.A_AUCT;
|
13
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.P_PARENTHESIS;
|
14
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.Q_NO_RESTRICTION;
|
15
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.R_GENUS;
|
16
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.R_SUBGENUS;
|
17
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.R_SPECIES;
|
18
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.R_SUBSPECIES;
|
19
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.T_STATUS_ACCEPTED;
|
20
|
import static eu.etaxonomy.cdm.io.faunaEuropaea.FaunaEuropaeaTransformer.T_STATUS_NOT_ACCEPTED;
|
21
|
|
22
|
import java.sql.ResultSet;
|
23
|
import java.sql.SQLException;
|
24
|
import java.util.Collection;
|
25
|
import java.util.HashMap;
|
26
|
import java.util.HashSet;
|
27
|
import java.util.Iterator;
|
28
|
import java.util.List;
|
29
|
import java.util.Map;
|
30
|
import java.util.Set;
|
31
|
import java.util.UUID;
|
32
|
|
33
|
import org.apache.log4j.Logger;
|
34
|
import org.hibernate.Session;
|
35
|
import org.springframework.stereotype.Component;
|
36
|
import org.springframework.transaction.TransactionStatus;
|
37
|
|
38
|
import eu.etaxonomy.cdm.api.service.IService;
|
39
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
40
|
import eu.etaxonomy.cdm.io.berlinModel.CdmOneToManyMapper;
|
41
|
import eu.etaxonomy.cdm.io.berlinModel.CdmStringMapper;
|
42
|
import eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportState;
|
43
|
import eu.etaxonomy.cdm.io.common.CdmAttributeMapperBase;
|
44
|
import eu.etaxonomy.cdm.io.common.CdmSingleAttributeMapperBase;
|
45
|
import eu.etaxonomy.cdm.io.common.ICdmIO;
|
46
|
import eu.etaxonomy.cdm.io.common.IImportConfigurator;
|
47
|
import eu.etaxonomy.cdm.io.common.ImportHelper;
|
48
|
import eu.etaxonomy.cdm.io.common.MapWrapper;
|
49
|
import eu.etaxonomy.cdm.io.common.Source;
|
50
|
import eu.etaxonomy.cdm.io.profiler.ProfilerController;
|
51
|
import eu.etaxonomy.cdm.io.tcsxml.in.TcsXmlImportState;
|
52
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
53
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
54
|
import eu.etaxonomy.cdm.model.common.ISourceable;
|
55
|
import eu.etaxonomy.cdm.model.common.IdentifiableEntity;
|
56
|
import eu.etaxonomy.cdm.model.common.OriginalSource;
|
57
|
import eu.etaxonomy.cdm.model.name.Rank;
|
58
|
import eu.etaxonomy.cdm.model.name.TaxonNameBase;
|
59
|
import eu.etaxonomy.cdm.model.name.ZoologicalName;
|
60
|
import eu.etaxonomy.cdm.model.reference.Database;
|
61
|
import eu.etaxonomy.cdm.model.reference.Generic;
|
62
|
import eu.etaxonomy.cdm.model.reference.PublicationBase;
|
63
|
import eu.etaxonomy.cdm.model.reference.Publisher;
|
64
|
import eu.etaxonomy.cdm.model.reference.ReferenceBase;
|
65
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
66
|
import eu.etaxonomy.cdm.model.taxon.SynonymRelationshipType;
|
67
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
68
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
69
|
import eu.etaxonomy.cdm.model.taxon.TaxonomicTree;
|
70
|
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException;
|
71
|
|
72
|
import com.yourkit.api.Controller;
|
73
|
|
74
|
|
75
|
|
76
|
/**
|
77
|
* @author a.babadshanjan
|
78
|
* @created 12.05.2009
|
79
|
* @version 1.0
|
80
|
*/
|
81
|
@Component
|
82
|
public class FaunaEuropaeaRelTaxonIncludeImport extends FaunaEuropaeaImportBase {
|
83
|
|
84
|
public static final String OS_NAMESPACE_TAXON = "Taxon";
|
85
|
private static final Logger logger = Logger.getLogger(FaunaEuropaeaRelTaxonIncludeImport.class);
|
86
|
|
87
|
/* Max number of taxa to retrieve (for test purposes) */
|
88
|
private int maxTaxa = 0;
|
89
|
/* Max number of taxa to be saved in CDM DB with one service call */
|
90
|
private int limit = 5000; // TODO: Make configurable
|
91
|
/* Max number of taxa to be retrieved from CDM DB with one service call */
|
92
|
private int limitRetrieve = 10000; // TODO: Make configurable
|
93
|
/* Interval for progress info message when retrieving taxa */
|
94
|
private int modCount = 10000;
|
95
|
/* Highest taxon index in the FauEu database */
|
96
|
private int highestTaxonIndex = 0;
|
97
|
/* Number of times method buildParentName() has been called for one taxon */
|
98
|
private int callCount = 0;
|
99
|
private Map<Integer, FaunaEuropaeaTaxon> fauEuTaxonMap = new HashMap();
|
100
|
|
101
|
|
102
|
|
103
|
/* (non-Javadoc)
|
104
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#doCheck(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
105
|
*/
|
106
|
@Override
|
107
|
protected boolean doCheck(FaunaEuropaeaImportState state) {
|
108
|
boolean result = true;
|
109
|
FaunaEuropaeaImportConfigurator fauEuConfig = state.getConfig();
|
110
|
logger.warn("Checking for Taxa not yet fully implemented");
|
111
|
result &= checkTaxonStatus(fauEuConfig);
|
112
|
|
113
|
return result;
|
114
|
}
|
115
|
|
116
|
/* (non-Javadoc)
|
117
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
118
|
*/
|
119
|
protected boolean isIgnore(FaunaEuropaeaImportState state) {
|
120
|
return ! state.getConfig().isDoTaxa();
|
121
|
}
|
122
|
|
123
|
private boolean checkTaxonStatus(FaunaEuropaeaImportConfigurator fauEuConfig) {
|
124
|
boolean result = true;
|
125
|
// try {
|
126
|
Source source = fauEuConfig.getSource();
|
127
|
String sqlStr = "";
|
128
|
ResultSet rs = source.getResultSet(sqlStr);
|
129
|
return result;
|
130
|
// } catch (SQLException e) {
|
131
|
// e.printStackTrace();
|
132
|
// return false;
|
133
|
// }
|
134
|
}
|
135
|
|
136
|
/* (non-Javadoc)
|
137
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#doInvoke(eu.etaxonomy.cdm.io.common.IImportConfigurator, eu.etaxonomy.cdm.api.application.CdmApplicationController, java.util.Map)
|
138
|
*/
|
139
|
protected boolean doInvokeAlter(FaunaEuropaeaImportState state) {
|
140
|
|
141
|
boolean success = true;
|
142
|
|
143
|
if(logger.isInfoEnabled()) { logger.info("Start making taxa..."); }
|
144
|
|
145
|
// TransactionStatus txStatus = startTransaction();
|
146
|
|
147
|
success = retrieveTaxa(state, fauEuTaxonMap, Q_NO_RESTRICTION);
|
148
|
success = processTaxaFromDatabase(state, fauEuTaxonMap);
|
149
|
|
150
|
// commitTransaction(txStatus);
|
151
|
|
152
|
logger.info("End making taxa...");
|
153
|
return success;
|
154
|
}
|
155
|
|
156
|
|
157
|
protected boolean doInvoke(FaunaEuropaeaImportState state) {
|
158
|
|
159
|
boolean success = true;
|
160
|
|
161
|
|
162
|
ProfilerController.memorySnapshot();
|
163
|
|
164
|
Map<String, MapWrapper<? extends CdmBase>> stores = state.getStores();
|
165
|
MapWrapper<TaxonBase> taxonStore = (MapWrapper<TaxonBase>)stores.get(ICdmIO.TAXON_STORE);
|
166
|
taxonStore.makeEmpty();
|
167
|
MapWrapper<TeamOrPersonBase> authorStore = (MapWrapper<TeamOrPersonBase>)stores.get(ICdmIO.TEAM_STORE);
|
168
|
authorStore.makeEmpty();
|
169
|
|
170
|
if(logger.isInfoEnabled()) { logger.info("Start making taxonomically included relationships..."); }
|
171
|
|
172
|
// TransactionStatus txStatus = startTransaction();
|
173
|
|
174
|
success = retrieveChildParentUuidMap(state);
|
175
|
ProfilerController.memorySnapshot();
|
176
|
success = createRelationships(state);
|
177
|
|
178
|
// commitTransaction(txStatus);
|
179
|
|
180
|
logger.info("End making taxa...");
|
181
|
ProfilerController.memorySnapshot();
|
182
|
|
183
|
return success;
|
184
|
}
|
185
|
|
186
|
/** Retrieve child-parent uuid map from CDM DB */
|
187
|
private boolean retrieveChildParentUuidMap(FaunaEuropaeaImportState state) {
|
188
|
|
189
|
Map<UUID, UUID> childParentMap = state.getChildParentMap();
|
190
|
FaunaEuropaeaImportConfigurator fauEuConfig = state.getConfig();
|
191
|
Source source = fauEuConfig.getSource();
|
192
|
int i = 0;
|
193
|
boolean success = true;
|
194
|
|
195
|
try {
|
196
|
|
197
|
String strQuery =
|
198
|
" SELECT dbo.Taxon.UUID AS ChildUuid, Parent.UUID AS ParentUuid " +
|
199
|
" FROM dbo.Taxon INNER JOIN dbo.Taxon AS Parent " +
|
200
|
" ON dbo.Taxon.TAX_TAX_IDPARENT = Parent.TAX_ID " +
|
201
|
" WHERE (dbo.Taxon.TAX_VALID <> 0) AND (dbo.Taxon.TAX_AUT_ID <> " + A_AUCT + ")";
|
202
|
|
203
|
if (logger.isInfoEnabled()) {
|
204
|
logger.info("Query: " + strQuery);
|
205
|
}
|
206
|
|
207
|
ResultSet rs = source.getResultSet(strQuery);
|
208
|
|
209
|
while (rs.next()) {
|
210
|
|
211
|
if ((i++ % modCount) == 0 && i != 1 ) {
|
212
|
if(logger.isInfoEnabled()) {
|
213
|
logger.info("Parent-child mappings retrieved: " + (i-1));
|
214
|
}
|
215
|
}
|
216
|
|
217
|
String childUuidStr = rs.getString("ChildUuid");
|
218
|
String parentUuidStr = rs.getString("ParentUuid");
|
219
|
UUID childUuid = UUID.fromString(childUuidStr);
|
220
|
UUID parentUuid = UUID.fromString(parentUuidStr);
|
221
|
|
222
|
if (!childParentMap.containsKey(childUuid)) {
|
223
|
|
224
|
childParentMap.put(childUuid, parentUuid);
|
225
|
|
226
|
} else {
|
227
|
if(logger.isDebugEnabled()) {
|
228
|
logger.debug("Duplicated child UUID (" + childUuid + ")");
|
229
|
}
|
230
|
}
|
231
|
}
|
232
|
|
233
|
} catch (SQLException e) {
|
234
|
logger.error("SQLException:" + e);
|
235
|
success = false;
|
236
|
}
|
237
|
return success;
|
238
|
}
|
239
|
|
240
|
|
241
|
/** Retrieve taxa from FauEu DB and build FauEuTaxonMap only */
|
242
|
private boolean retrieveTaxa(FaunaEuropaeaImportState state,
|
243
|
Map<Integer, FaunaEuropaeaTaxon> fauEuTaxonMap, int valid) {
|
244
|
|
245
|
Map<String, MapWrapper<? extends CdmBase>> stores = state.getStores();
|
246
|
MapWrapper<TaxonBase> taxonStore = (MapWrapper<TaxonBase>)stores.get(ICdmIO.TAXON_STORE);
|
247
|
FaunaEuropaeaImportConfigurator fauEuConfig = state.getConfig();
|
248
|
ReferenceBase<?> sourceRef = fauEuConfig.getSourceReference();
|
249
|
MapWrapper<TeamOrPersonBase> authorStore = (MapWrapper<TeamOrPersonBase>)stores.get(ICdmIO.TEAM_STORE);
|
250
|
|
251
|
Source source = fauEuConfig.getSource();
|
252
|
// String namespace = "Taxon";
|
253
|
int i = 0;
|
254
|
boolean success = true;
|
255
|
|
256
|
try {
|
257
|
|
258
|
String strQuery =
|
259
|
" SELECT MAX(TAX_ID) AS TAX_ID FROM dbo.Taxon ";
|
260
|
|
261
|
ResultSet rs = source.getResultSet(strQuery);
|
262
|
while (rs.next()) {
|
263
|
int maxTaxonId = rs.getInt("TAX_ID");
|
264
|
highestTaxonIndex = maxTaxonId;
|
265
|
}
|
266
|
|
267
|
String top = "";
|
268
|
if (maxTaxa > 0) {
|
269
|
top = "TOP " + maxTaxa;
|
270
|
}
|
271
|
|
272
|
String validClause = "";
|
273
|
if (valid == T_STATUS_ACCEPTED || valid == T_STATUS_NOT_ACCEPTED) {
|
274
|
validClause = " AND " + " TAX_VALID = " + valid;
|
275
|
}
|
276
|
|
277
|
strQuery =
|
278
|
" SELECT " + top + " Taxon.*, rank.*, author.* " +
|
279
|
" FROM dbo.Taxon " +
|
280
|
" LEFT OUTER JOIN dbo.author ON dbo.Taxon.TAX_AUT_ID = dbo.author.aut_id " +
|
281
|
" LEFT OUTER JOIN dbo.rank ON dbo.Taxon.TAX_RNK_ID = dbo.rank.rnk_id " +
|
282
|
" WHERE (1=1)" +
|
283
|
validClause;
|
284
|
|
285
|
if (logger.isDebugEnabled()) {
|
286
|
logger.debug("Query: " + strQuery);
|
287
|
}
|
288
|
rs = source.getResultSet(strQuery);
|
289
|
|
290
|
while (rs.next()) {
|
291
|
|
292
|
if ((i++ % modCount) == 0 && i != 1 ) {
|
293
|
if(logger.isInfoEnabled()) {
|
294
|
logger.info("Taxa retrieved: " + (i-1));
|
295
|
}
|
296
|
}
|
297
|
|
298
|
int taxonId = rs.getInt("TAX_ID");
|
299
|
String localName = rs.getString("TAX_NAME");
|
300
|
int rankId = rs.getInt("TAX_RNK_ID");
|
301
|
int parentId = rs.getInt("TAX_TAX_IDPARENT");
|
302
|
int familyId = rs.getInt("TAX_TAX_IDFAMILY");
|
303
|
int originalGenusId = rs.getInt("TAX_TAX_IDGENUS");
|
304
|
int autId = rs.getInt("TAX_AUT_ID");
|
305
|
int status = rs.getInt("TAX_VALID");
|
306
|
int year = rs.getInt("TAX_YEAR");
|
307
|
int parenthesis = rs.getInt("TAX_PARENTHESIS");
|
308
|
String autName = rs.getString("aut_name");
|
309
|
Rank rank = null;
|
310
|
// UUID taxonBaseUuid = UUID.randomUUID();
|
311
|
|
312
|
FaunaEuropaeaTaxon fauEuTaxon = new FaunaEuropaeaTaxon();
|
313
|
// fauEuTaxon.setUuid(taxonBaseUuid);
|
314
|
fauEuTaxon.setLocalName(localName);
|
315
|
fauEuTaxon.setParentId(parentId);
|
316
|
fauEuTaxon.setOriginalGenusId(originalGenusId);
|
317
|
fauEuTaxon.setId(taxonId);
|
318
|
fauEuTaxon.setRankId(rankId);
|
319
|
fauEuTaxon.setYear(year);
|
320
|
fauEuTaxon.setAuthor(autName);
|
321
|
if (parenthesis == P_PARENTHESIS) {
|
322
|
fauEuTaxon.setParenthesis(true);
|
323
|
} else {
|
324
|
fauEuTaxon.setParenthesis(false);
|
325
|
}
|
326
|
if (status == T_STATUS_ACCEPTED) {
|
327
|
fauEuTaxon.setValid(true);
|
328
|
} else {
|
329
|
fauEuTaxon.setValid(false);
|
330
|
}
|
331
|
|
332
|
try {
|
333
|
rank = FaunaEuropaeaTransformer.rankId2Rank(rs, false);
|
334
|
} catch (UnknownCdmTypeException e) {
|
335
|
logger.warn("Taxon (" + taxonId + ") has unknown rank (" + rankId + ") and could not be saved.");
|
336
|
continue;
|
337
|
} catch (NullPointerException e) {
|
338
|
logger.warn("Taxon (" + taxonId + ") has rank null and can not be saved.");
|
339
|
continue;
|
340
|
}
|
341
|
|
342
|
try {
|
343
|
|
344
|
|
345
|
if (!fauEuTaxonMap.containsKey(taxonId)) {
|
346
|
if (fauEuTaxon == null) {
|
347
|
if (logger.isDebugEnabled()) {
|
348
|
logger.debug("Taxon base is null. Taxon (" + taxonId + ") ignored.");
|
349
|
}
|
350
|
continue;
|
351
|
}
|
352
|
|
353
|
|
354
|
// taxonStore.put(taxonId, taxonBase);
|
355
|
|
356
|
fauEuTaxonMap.put(taxonId, fauEuTaxon);
|
357
|
|
358
|
// if (logger.isDebugEnabled()) {
|
359
|
// logger.debug("Stored taxon base (" + taxonId + ") " + localName);
|
360
|
// }
|
361
|
} else {
|
362
|
logger.warn("Not imported taxon base with duplicated TAX_ID (" + taxonId +
|
363
|
") " + localName);
|
364
|
}
|
365
|
} catch (Exception e) {
|
366
|
logger.warn("An exception occurred when creating taxon base with id " + taxonId +
|
367
|
". Taxon base could not be saved.");
|
368
|
}
|
369
|
}
|
370
|
} catch (SQLException e) {
|
371
|
logger.error("SQLException:" + e);
|
372
|
success = false;
|
373
|
}
|
374
|
|
375
|
return success;
|
376
|
}
|
377
|
|
378
|
|
379
|
public Map<UUID, UUID> partMap(int border, Map<UUID, UUID> map) {
|
380
|
|
381
|
if (logger.isInfoEnabled()) {
|
382
|
logger.info("Map size: " + map.size());
|
383
|
}
|
384
|
Set<Map.Entry<UUID, UUID>> entries = map.entrySet();
|
385
|
Iterator<Map.Entry<UUID, UUID>> entryIter = entries.iterator();
|
386
|
Map<UUID, UUID> partMap = new HashMap<UUID, UUID>();
|
387
|
|
388
|
for (int i = 0; i < border; i++) {
|
389
|
//while (entryIter.hasNext()) {
|
390
|
|
391
|
Map.Entry<UUID, UUID> mapEntry = (Map.Entry<UUID, UUID>)entryIter.next();
|
392
|
partMap.put(mapEntry.getKey(), mapEntry.getValue());
|
393
|
entryIter.remove();
|
394
|
}
|
395
|
|
396
|
if (logger.isDebugEnabled()) {
|
397
|
logger.debug("Map size: " + map.size());
|
398
|
}
|
399
|
return partMap;
|
400
|
}
|
401
|
|
402
|
// public Map<UUID, UUID> childParentMap partMap(int start, int limit, Map<UUID, UUID> childParentMap) {
|
403
|
//
|
404
|
// int index = 0;
|
405
|
//
|
406
|
// for (int i = 0; i < limit; i++) {
|
407
|
//
|
408
|
// int j = start + i;
|
409
|
//
|
410
|
// Object object = childParentMap.get(j);
|
411
|
// if(object != null) {
|
412
|
// childParentMap.put(index, childParentMap.get(j));
|
413
|
// index++;
|
414
|
// } else {
|
415
|
// if (logger.isDebugEnabled()) { logger.debug("Object (" + j + ") is null"); }
|
416
|
// }
|
417
|
// }
|
418
|
// return (Map<UUID, UUID> childParentMap)internalPartMap.values();
|
419
|
// }
|
420
|
|
421
|
|
422
|
/** Creates parent-child relationships.
|
423
|
* Single Parent-child pairs are retrieved via findByUUID(UUID) from CDM DB
|
424
|
* This takes inacceptable long time. */
|
425
|
private boolean createRelationships_(FaunaEuropaeaImportState state) {
|
426
|
|
427
|
Map<String, MapWrapper<? extends CdmBase>> stores = state.getStores();
|
428
|
MapWrapper<TaxonBase> taxonStore = (MapWrapper<TaxonBase>)stores.get(ICdmIO.TAXON_STORE);
|
429
|
taxonStore.makeEmpty();
|
430
|
Map<UUID, UUID> childParentMap = state.getChildParentMap();
|
431
|
ReferenceBase<?> sourceRef = state.getConfig().getSourceReference();
|
432
|
|
433
|
int upperBorder = childParentMap.size();
|
434
|
int nbrOfBlocks = 0;
|
435
|
|
436
|
boolean success = true;
|
437
|
|
438
|
if (upperBorder < limit) { // TODO: test with critical values
|
439
|
limit = upperBorder;
|
440
|
} else {
|
441
|
nbrOfBlocks = upperBorder / limit;
|
442
|
}
|
443
|
|
444
|
if(logger.isInfoEnabled()) {
|
445
|
logger.info("number of child-parent pairs = " + upperBorder
|
446
|
+ ", limit = " + limit
|
447
|
+ ", number of blocks = " + nbrOfBlocks);
|
448
|
}
|
449
|
|
450
|
for (int j = 1; j <= nbrOfBlocks + 1; j++) {
|
451
|
int offset = j - 1;
|
452
|
int start = offset * limit;
|
453
|
|
454
|
if(logger.isInfoEnabled()) { logger.info("Processing child-parent pairs: " + start + " - " + (start + limit - 1)); }
|
455
|
|
456
|
if(logger.isInfoEnabled()) {
|
457
|
logger.info("index = " + j
|
458
|
+ ", offset = " + offset
|
459
|
+ ", start = " + start);
|
460
|
}
|
461
|
|
462
|
if (j == nbrOfBlocks + 1) {
|
463
|
limit = upperBorder - nbrOfBlocks * limit;
|
464
|
if(logger.isInfoEnabled()) { logger.info("number of blocks = " + nbrOfBlocks + " limit = " + limit); }
|
465
|
}
|
466
|
|
467
|
TransactionStatus txStatus = startTransaction();
|
468
|
|
469
|
|
470
|
// for (int k = 1; k <= start + offset; k++) { // TODO: test borders
|
471
|
// int k = 0;
|
472
|
|
473
|
Map<UUID, UUID> childParentPartMap = partMap(limit, childParentMap);
|
474
|
Set<TaxonBase> childSet = new HashSet<TaxonBase>(limit);
|
475
|
|
476
|
if (logger.isInfoEnabled()) {
|
477
|
logger.info("Partmap size: " + childParentPartMap.size());
|
478
|
}
|
479
|
|
480
|
for (UUID childUuid : childParentPartMap.keySet()) {
|
481
|
// for (UUID childUuid : childParentMap.keySet()) {
|
482
|
|
483
|
UUID parentUuid = childParentPartMap.get(childUuid);
|
484
|
|
485
|
try {
|
486
|
TaxonBase<?> parent = getTaxonService().findByUuid(parentUuid);
|
487
|
if (logger.isTraceEnabled()) {
|
488
|
logger.trace("Parent find called (" + parentUuid + ")");
|
489
|
}
|
490
|
TaxonBase<?> child = getTaxonService().findByUuid(childUuid);
|
491
|
if (logger.isTraceEnabled()) {
|
492
|
logger.trace("Child find called (" + childUuid + ")");
|
493
|
}
|
494
|
Taxon parentTaxon = parent.deproxy(parent, Taxon.class);
|
495
|
Taxon childTaxon = child.deproxy(child, Taxon.class);
|
496
|
|
497
|
if (childTaxon != null && parentTaxon != null) {
|
498
|
|
499
|
// makeTaxonomicallyIncluded(state, parentTaxon, childTaxon, sourceRef, null);
|
500
|
|
501
|
if (logger.isDebugEnabled()) {
|
502
|
logger.debug("Parent-child (" + parentUuid + "-" + childUuid +
|
503
|
") relationship created");
|
504
|
}
|
505
|
if (!childSet.contains(childTaxon)) {
|
506
|
|
507
|
childSet.add(childTaxon);
|
508
|
|
509
|
if (logger.isTraceEnabled()) {
|
510
|
logger.trace("Child taxon (" + childUuid + ") added to Set");
|
511
|
}
|
512
|
|
513
|
} else {
|
514
|
if (logger.isDebugEnabled()) {
|
515
|
logger.debug("Duplicated child taxon (" + childUuid + ")");
|
516
|
}
|
517
|
}
|
518
|
} else {
|
519
|
if (logger.isDebugEnabled()) {
|
520
|
logger.debug("Parent(" + parentUuid + ") or child (" + childUuid + " is null");
|
521
|
}
|
522
|
}
|
523
|
|
524
|
// if (childTaxon != null && !childSet.contains(childTaxon)) {
|
525
|
// childSet.add(childTaxon);
|
526
|
// if (logger.isDebugEnabled()) {
|
527
|
// logger.debug("Child taxon (" + childUuid + ") added to Set");
|
528
|
// }
|
529
|
// } else {
|
530
|
// if (logger.isDebugEnabled()) {
|
531
|
// logger.debug("Duplicated child taxon (" + childUuid + ")");
|
532
|
// }
|
533
|
// }
|
534
|
|
535
|
} catch (Exception e) {
|
536
|
logger.error("Error creating taxonomically included relationship parent-child (" +
|
537
|
parentUuid + "-" + childUuid + ")");
|
538
|
}
|
539
|
|
540
|
}
|
541
|
getTaxonService().saveTaxonAll(childSet);
|
542
|
commitTransaction(txStatus);
|
543
|
}
|
544
|
return success;
|
545
|
}
|
546
|
|
547
|
|
548
|
/* Creates parent-child relationships.
|
549
|
* Parent-child pairs are retrieved in blocks via findByUUID(Set<UUID>) from CDM DB.
|
550
|
* It takes about 5min to save a block of 5000 taxa.*/
|
551
|
private boolean createRelationships(FaunaEuropaeaImportState state) {
|
552
|
|
553
|
Map<UUID, UUID> childParentUuidMap = state.getChildParentMap();
|
554
|
ReferenceBase<?> sourceRef = state.getConfig().getSourceReference();
|
555
|
// UUID treeUuid = state.getTree(sourceRef).getUuid();
|
556
|
// TaxonomicTree tree = getTaxonService().getTaxonomicTreeByUuid(treeUuid);
|
557
|
// TaxonomicTree tree = state.getTree(sourceRef);
|
558
|
|
559
|
int upperBorder = childParentUuidMap.size();
|
560
|
int nbrOfBlocks = 0;
|
561
|
|
562
|
boolean success = true;
|
563
|
|
564
|
if (upperBorder < limit) { // TODO: test with critical values
|
565
|
limit = upperBorder;
|
566
|
} else {
|
567
|
nbrOfBlocks = upperBorder / limit;
|
568
|
}
|
569
|
|
570
|
if(logger.isInfoEnabled()) {
|
571
|
logger.info("number of child-parent pairs = " + upperBorder
|
572
|
+ ", limit = " + limit
|
573
|
+ ", number of blocks = " + nbrOfBlocks);
|
574
|
}
|
575
|
|
576
|
for (int j = 1; j <= nbrOfBlocks + 1; j++) {
|
577
|
int offset = j - 1;
|
578
|
int start = offset * limit;
|
579
|
|
580
|
if(logger.isInfoEnabled()) { logger.info("Processing child-parent pairs: " + start + " - " + (start + limit - 1)); }
|
581
|
|
582
|
if(logger.isInfoEnabled()) {
|
583
|
logger.info("index = " + j
|
584
|
+ ", offset = " + offset
|
585
|
+ ", start = " + start);
|
586
|
}
|
587
|
|
588
|
if (j == nbrOfBlocks + 1) {
|
589
|
limit = upperBorder - nbrOfBlocks * limit;
|
590
|
if(logger.isInfoEnabled()) { logger.info("number of blocks = " + nbrOfBlocks + " limit = " + limit); }
|
591
|
}
|
592
|
|
593
|
TransactionStatus txStatus = startTransaction();
|
594
|
//add tree to new session
|
595
|
TaxonomicTree tree = state.getTree(sourceRef);
|
596
|
if (tree == null){
|
597
|
tree = makeTree(state, sourceRef);
|
598
|
}
|
599
|
getTaxonService().saveTaxonomicTree(tree);
|
600
|
|
601
|
Map<UUID, UUID> childParentPartUuidMap = partMap(limit, childParentUuidMap);
|
602
|
Set<TaxonBase> childSet = new HashSet<TaxonBase>(limit);
|
603
|
|
604
|
Set<UUID> childKeysSet = childParentPartUuidMap.keySet();
|
605
|
Set<UUID> parentValuesSet = new HashSet<UUID>(childParentPartUuidMap.values());
|
606
|
|
607
|
if (logger.isInfoEnabled()) {
|
608
|
logger.info("Start reading children and parents");
|
609
|
}
|
610
|
List<TaxonBase> children = getTaxonService().findByUuid(childKeysSet);
|
611
|
List<TaxonBase> parents = getTaxonService().findByUuid(parentValuesSet);
|
612
|
Map<UUID, TaxonBase> parentsMap = new HashMap<UUID, TaxonBase>();
|
613
|
for (TaxonBase taxonBase : parents){
|
614
|
parentsMap.put(taxonBase.getUuid(), taxonBase);
|
615
|
}
|
616
|
|
617
|
|
618
|
if (logger.isInfoEnabled()) {
|
619
|
logger.info("End reading children and parents");
|
620
|
}
|
621
|
|
622
|
|
623
|
if (logger.isTraceEnabled()) {
|
624
|
for (UUID uuid : childKeysSet) {
|
625
|
logger.trace("child uuid query: " + uuid);
|
626
|
}
|
627
|
}
|
628
|
if (logger.isTraceEnabled()) {
|
629
|
for (UUID uuid : parentValuesSet) {
|
630
|
logger.trace("parent uuid query: " + uuid);
|
631
|
}
|
632
|
}
|
633
|
if (logger.isTraceEnabled()) {
|
634
|
for (TaxonBase tb : children) {
|
635
|
logger.trace("child uuid result: " + tb.getUuid());
|
636
|
}
|
637
|
}
|
638
|
if (logger.isTraceEnabled()) {
|
639
|
for (TaxonBase tb : parents) {
|
640
|
logger.trace("parent uuid result: " + tb.getUuid());
|
641
|
}
|
642
|
}
|
643
|
|
644
|
UUID mappedParentUuid = null;
|
645
|
UUID parentUuid = null;
|
646
|
UUID childUuid = null;
|
647
|
|
648
|
for (TaxonBase child : children) {
|
649
|
|
650
|
try {
|
651
|
Taxon childTaxon = child.deproxy(child, Taxon.class);
|
652
|
childUuid = childTaxon.getUuid();
|
653
|
mappedParentUuid = childParentPartUuidMap.get(childUuid);
|
654
|
TaxonBase parent = null;
|
655
|
|
656
|
TaxonBase potentialParent = parentsMap.get(mappedParentUuid);
|
657
|
// for (TaxonBase potentialParent : parents ) {
|
658
|
// parentUuid = potentialParent.getUuid();
|
659
|
// if(parentUuid.equals(mappedParentUuid)) {
|
660
|
parent = potentialParent;
|
661
|
if (logger.isDebugEnabled()) {
|
662
|
logger.debug("Parent (" + parentUuid + ") found for child (" + childUuid + ")");
|
663
|
}
|
664
|
// break;
|
665
|
// }
|
666
|
// }
|
667
|
|
668
|
Taxon parentTaxon = parent.deproxy(parent, Taxon.class);
|
669
|
|
670
|
if (childTaxon != null && parentTaxon != null) {
|
671
|
|
672
|
// makeTaxonomicallyIncluded(state, parentTaxon, childTaxon, sourceRef, null, tree);
|
673
|
makeTaxonomicallyIncluded(state, parentTaxon, childTaxon, sourceRef, null);
|
674
|
|
675
|
if (logger.isDebugEnabled()) {
|
676
|
logger.debug("Parent-child (" + parentUuid + "-" + childUuid +
|
677
|
") relationship created");
|
678
|
}
|
679
|
if (!childSet.contains(childTaxon)) {
|
680
|
|
681
|
childSet.add(childTaxon);
|
682
|
|
683
|
if (logger.isTraceEnabled()) {
|
684
|
logger.trace("Child taxon (" + childUuid + ") added to Set");
|
685
|
}
|
686
|
|
687
|
} else {
|
688
|
if (logger.isDebugEnabled()) {
|
689
|
logger.debug("Duplicated child taxon (" + childUuid + ")");
|
690
|
}
|
691
|
}
|
692
|
} else {
|
693
|
if (logger.isDebugEnabled()) {
|
694
|
logger.debug("Parent(" + parentUuid + ") or child (" + childUuid + " is null");
|
695
|
}
|
696
|
}
|
697
|
|
698
|
if (childTaxon != null && !childSet.contains(childTaxon)) {
|
699
|
childSet.add(childTaxon);
|
700
|
if (logger.isDebugEnabled()) {
|
701
|
logger.debug("Child taxon (" + childUuid + ") added to Set");
|
702
|
}
|
703
|
} else {
|
704
|
if (logger.isDebugEnabled()) {
|
705
|
logger.debug("Duplicated child taxon (" + childUuid + ")");
|
706
|
}
|
707
|
}
|
708
|
|
709
|
} catch (Exception e) {
|
710
|
logger.error("Error creating taxonomically included relationship parent-child (" +
|
711
|
parentUuid + "-" + childUuid + ")");
|
712
|
}
|
713
|
|
714
|
}
|
715
|
if (logger.isInfoEnabled()) {
|
716
|
logger.info("Start saving childSet");
|
717
|
}
|
718
|
getTaxonService().saveTaxonAll(childSet);
|
719
|
if (logger.isInfoEnabled()) {
|
720
|
logger.info("End saving childSet");
|
721
|
}
|
722
|
// getTaxonService().clear();
|
723
|
// if (logger.isInfoEnabled()) {
|
724
|
// logger.info("End clearing session");
|
725
|
// }
|
726
|
commitTransaction(txStatus);
|
727
|
if (logger.isInfoEnabled()) {
|
728
|
logger.info("End commit transaction");
|
729
|
}
|
730
|
parentValuesSet = null;
|
731
|
childSet = null;
|
732
|
childParentPartUuidMap = null;
|
733
|
children = null;
|
734
|
parents = null;
|
735
|
}
|
736
|
return success;
|
737
|
}
|
738
|
|
739
|
|
740
|
/* Creates parent-child relationships.
|
741
|
* Taxon bases are retrieved in blocks from CDM DB.
|
742
|
* Parent is retrieved from CDM DB via original source id if not found in current block.
|
743
|
* In case of blocksize = 20.000 this takes ca. 1-2 hours per block.
|
744
|
* */
|
745
|
private boolean createRelationships_old(FaunaEuropaeaTaxon fauEuTaxon,
|
746
|
TaxonBase<?> taxonBase, TaxonNameBase<?,?> taxonName, List<Taxon> taxa,
|
747
|
Map<Integer, FaunaEuropaeaTaxon> fauEuTaxonMap, FaunaEuropaeaImportState state) {
|
748
|
|
749
|
int parentId = fauEuTaxon.getParentId();
|
750
|
int taxonId = fauEuTaxon.getId();
|
751
|
FaunaEuropaeaTaxon parentFauEuTaxon = fauEuTaxonMap.get(parentId);
|
752
|
if (parentFauEuTaxon == null) {
|
753
|
if (logger.isInfoEnabled()) {
|
754
|
logger.info("Parent taxon is null (" + parentId + ")");
|
755
|
}
|
756
|
return false;
|
757
|
}
|
758
|
// UUID parentUuid = parentFauEuTaxon.getUuid();
|
759
|
Map<String, MapWrapper<? extends CdmBase>> stores = state.getStores();
|
760
|
MapWrapper<TaxonBase> parentTaxonStore = (MapWrapper<TaxonBase>)stores.get(ICdmIO.TAXON_STORE);
|
761
|
ReferenceBase<?> sourceRef = state.getConfig().getSourceReference();
|
762
|
|
763
|
TaxonBase<?> parentTaxonBase = null;
|
764
|
|
765
|
// for (TaxonBase<?> potentialParentTaxon : taxonBases) {
|
766
|
// if(potentialParentTaxon.getUuid().equals(parentUuid)) {
|
767
|
// parentTaxonBase = potentialParentTaxon;
|
768
|
// break;
|
769
|
// }
|
770
|
// }
|
771
|
// if (parentTaxonBase == null) {
|
772
|
// parentTaxonBase = getTaxonService().getTaxonByUuid(parentUuid);
|
773
|
// }
|
774
|
|
775
|
// TODO: Copy parents from taxonBases to parentTaxonStore
|
776
|
|
777
|
if (parentTaxonStore.containsId(parentId)) {
|
778
|
parentTaxonBase = parentTaxonStore.get(parentId);
|
779
|
if (logger.isDebugEnabled()) {
|
780
|
logger.debug("Parent (" + parentId + ") found in parent taxon store");
|
781
|
}
|
782
|
// } else {
|
783
|
// for (TaxonBase<?> potentialParentTaxon : taxonBases) {
|
784
|
// if(potentialParentTaxon.getId() == parentId) {
|
785
|
// parentTaxonBase = potentialParentTaxon;
|
786
|
// if (logger.isInfoEnabled()) {
|
787
|
// logger.info("Parent (" + parentId + ") found in taxon base list");
|
788
|
// }
|
789
|
// break;
|
790
|
// }
|
791
|
// }
|
792
|
}
|
793
|
if (parentTaxonBase == null) {
|
794
|
ISourceable sourceable =
|
795
|
getCommonService().getSourcedObjectByIdInSource(TaxonBase.class, Integer.toString(parentId), OS_NAMESPACE_TAXON);
|
796
|
parentTaxonBase = ((IdentifiableEntity)sourceable).deproxy(sourceable, TaxonBase.class);
|
797
|
if (logger.isDebugEnabled()) {
|
798
|
logger.debug("Parent (" + parentId + ") retrieved from DB via original source id");
|
799
|
}
|
800
|
}
|
801
|
|
802
|
if (!parentTaxonStore.containsId(parentId)) {
|
803
|
parentTaxonStore.put(parentId, parentTaxonBase);
|
804
|
}
|
805
|
|
806
|
|
807
|
|
808
|
Taxon parentTaxon = parentTaxonBase.deproxy(parentTaxonBase, Taxon.class);
|
809
|
|
810
|
boolean success = true;
|
811
|
|
812
|
// if (!fauEuTaxon.isValid()) { // FauEu Synonym
|
813
|
|
814
|
// } else if (fauEuTaxon.isValid()) { // FauEu Taxon
|
815
|
|
816
|
Taxon taxon = taxonBase.deproxy(taxonBase, Taxon.class);
|
817
|
|
818
|
try {
|
819
|
// add this taxon as child to parent
|
820
|
if (parentTaxon != null) {
|
821
|
// makeTaxonomicallyIncluded(state, parentTaxon, taxon, sourceRef, null);
|
822
|
if (logger.isDebugEnabled()) {
|
823
|
logger.debug("Parent-child (" + parentId + "-" + taxonId +
|
824
|
") relationship created");
|
825
|
}
|
826
|
}
|
827
|
|
828
|
} catch (Exception e) {
|
829
|
logger.error("Error creating taxonomically included relationship Parent-child (" +
|
830
|
parentId + "-" + taxonId + ")");
|
831
|
}
|
832
|
|
833
|
|
834
|
// }
|
835
|
|
836
|
return success;
|
837
|
}
|
838
|
|
839
|
|
840
|
private boolean makeTaxonomicallyIncluded(FaunaEuropaeaImportState state, Taxon toTaxon, Taxon fromTaxon,
|
841
|
ReferenceBase citation, String microCitation){
|
842
|
boolean success = true;
|
843
|
ReferenceBase sec = toTaxon.getSec();
|
844
|
sec = CdmBase.deproxy(sec, ReferenceBase.class);
|
845
|
sec = citation;
|
846
|
TaxonomicTree tree = state.getTree(sec);
|
847
|
|
848
|
|
849
|
|
850
|
// Session session = getTaxonService().getSession();
|
851
|
|
852
|
// if (session.contains(sec)) {
|
853
|
// logger.debug("Sec contained in session. Id = " + sec.getId());
|
854
|
// } else {
|
855
|
// logger.info("Sec not contained in session. Id = " + sec.getId());
|
856
|
// getReferenceService().merge(sec);
|
857
|
// }
|
858
|
|
859
|
if (tree == null){
|
860
|
tree = makeTree(state, sec);
|
861
|
}
|
862
|
|
863
|
// if (session.contains(tree)) {
|
864
|
// logger.debug("Taxonomic tree contained in session. Id = " + tree.getId());
|
865
|
// } else {
|
866
|
// logger.info("Taxonomic tree not contained in session. Id = " + tree.getId());
|
867
|
// UUID treeUuid = state.getTree(sec).getUuid();
|
868
|
// tree = getTaxonService().getTaxonomicTreeByUuid(treeUuid);
|
869
|
// logger.info("Tree retrieved");
|
870
|
// }
|
871
|
|
872
|
success = tree.addParentChild(toTaxon, fromTaxon, citation, microCitation);
|
873
|
return success;
|
874
|
}
|
875
|
|
876
|
|
877
|
// public int calculateBlockSize(int limit, int upperBorder) {
|
878
|
//
|
879
|
// int blockSize = 0;
|
880
|
//
|
881
|
// if (upperBorder < limit) {
|
882
|
// limit = upperBorder;
|
883
|
// } else {
|
884
|
// blockSize = upperBorder / limit;
|
885
|
// }
|
886
|
// }
|
887
|
|
888
|
|
889
|
private boolean processTaxaFromDatabase(FaunaEuropaeaImportState state,
|
890
|
Map<Integer, FaunaEuropaeaTaxon> fauEuTaxonMap) {
|
891
|
|
892
|
if(logger.isInfoEnabled()) { logger.info("Processing taxa second pass..."); }
|
893
|
|
894
|
MapWrapper<TaxonBase> taxonBaseMap = new MapWrapper<TaxonBase>(null);
|
895
|
|
896
|
int nbrOfTaxa = getTaxonService().count(Taxon.class);
|
897
|
int n = 0;
|
898
|
|
899
|
boolean success = true;
|
900
|
|
901
|
if (nbrOfTaxa < limit) { // TODO: test with critical values
|
902
|
limit = nbrOfTaxa;
|
903
|
} else {
|
904
|
n = nbrOfTaxa / limit;
|
905
|
}
|
906
|
|
907
|
if(logger.isInfoEnabled()) {
|
908
|
logger.info("number of taxa = " + nbrOfTaxa
|
909
|
+ ", limit = " + limit
|
910
|
+ ", n = " + n);
|
911
|
}
|
912
|
|
913
|
// process taxa in chunks of <=limit
|
914
|
|
915
|
for (int j = 1; j <= n + 1; j++)
|
916
|
{
|
917
|
int offset = j - 1;
|
918
|
int start = offset * limit;
|
919
|
|
920
|
if(logger.isInfoEnabled()) { logger.info("Processing taxa: " + start + " - " + (start + limit - 1)); }
|
921
|
|
922
|
if(logger.isInfoEnabled()) {
|
923
|
logger.info("index = " + j
|
924
|
+ ", offset = " + offset
|
925
|
+ ", start = " + start);
|
926
|
}
|
927
|
|
928
|
if (j == n + 1) {
|
929
|
limit = nbrOfTaxa - n * limit;
|
930
|
if(logger.isInfoEnabled()) { logger.info("n = " + n + " limit = " + limit); }
|
931
|
}
|
932
|
|
933
|
TransactionStatus txStatus = startTransaction();
|
934
|
|
935
|
List<Taxon> taxa = getTaxonService().getAllTaxa(limit, start);
|
936
|
if(logger.isInfoEnabled()) {
|
937
|
logger.info(taxa.size() + " taxa retrieved from CDM DB");
|
938
|
}
|
939
|
|
940
|
for (TaxonBase taxonBase : taxa) {
|
941
|
|
942
|
TaxonNameBase<?,?> taxonName = taxonBase.getName();
|
943
|
|
944
|
FaunaEuropaeaTaxon fauEuTaxon = findFauEuTaxonByOriginalSourceId(taxonBase, fauEuTaxonMap);
|
945
|
|
946
|
|
947
|
if (logger.isDebugEnabled()) {
|
948
|
logger.debug("Taxon # " + fauEuTaxon.getId());
|
949
|
}
|
950
|
//createRelationships(fauEuTaxon, taxonBase, taxonName, taxa, fauEuTaxonMap, state);
|
951
|
}
|
952
|
|
953
|
getTaxonService().saveTaxonAll(taxa);
|
954
|
taxa = null;
|
955
|
|
956
|
commitTransaction(txStatus);
|
957
|
|
958
|
// empty parent taxon store
|
959
|
// Map<String, MapWrapper<? extends CdmBase>> stores = state.getStores();
|
960
|
// MapWrapper<TaxonBase> parentTaxonStore = (MapWrapper<TaxonBase>)stores.get(ICdmIO.TAXON_STORE);
|
961
|
// parentTaxonStore.makeEmpty();
|
962
|
}
|
963
|
return success;
|
964
|
}
|
965
|
|
966
|
|
967
|
private FaunaEuropaeaTaxon findFauEuTaxonByOriginalSourceId(TaxonBase<?> taxonBase,
|
968
|
Map<Integer, FaunaEuropaeaTaxon> fauEuTaxonMap) {
|
969
|
|
970
|
Set set = taxonBase.getSources();
|
971
|
Object[] array = set.toArray();
|
972
|
if (array.length == 0) { return null; }
|
973
|
OriginalSource os = (OriginalSource) taxonBase.getSources().toArray()[0];
|
974
|
String taxonBaseIdStr = os.getIdInSource();
|
975
|
int taxonBaseId = Integer.parseInt(taxonBaseIdStr);
|
976
|
FaunaEuropaeaTaxon fauEuTaxon = fauEuTaxonMap.get(taxonBaseId);
|
977
|
|
978
|
return fauEuTaxon;
|
979
|
}
|
980
|
|
981
|
|
982
|
}
|