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