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.erms;
|
11
|
|
12
|
import java.sql.ResultSet;
|
13
|
import java.sql.SQLException;
|
14
|
import java.util.HashMap;
|
15
|
import java.util.HashSet;
|
16
|
import java.util.Map;
|
17
|
import java.util.Set;
|
18
|
import java.util.UUID;
|
19
|
|
20
|
import org.apache.log4j.Logger;
|
21
|
import org.springframework.stereotype.Component;
|
22
|
|
23
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
24
|
import eu.etaxonomy.cdm.io.common.IOValidator;
|
25
|
import eu.etaxonomy.cdm.io.common.mapping.DbIgnoreMapper;
|
26
|
import eu.etaxonomy.cdm.io.common.mapping.DbImportExtensionMapper;
|
27
|
import eu.etaxonomy.cdm.io.common.mapping.DbImportMapping;
|
28
|
import eu.etaxonomy.cdm.io.common.mapping.DbImportObjectCreationMapper;
|
29
|
import eu.etaxonomy.cdm.io.common.mapping.DbImportStringMapper;
|
30
|
import eu.etaxonomy.cdm.io.common.mapping.DbNotYetImplementedMapper;
|
31
|
import eu.etaxonomy.cdm.io.common.mapping.IMappingImport;
|
32
|
import eu.etaxonomy.cdm.io.erms.validation.ErmsTaxonImportValidator;
|
33
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
34
|
import eu.etaxonomy.cdm.model.common.ExtensionType;
|
35
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
36
|
import eu.etaxonomy.cdm.model.name.NonViralName;
|
37
|
import eu.etaxonomy.cdm.model.name.Rank;
|
38
|
import eu.etaxonomy.cdm.model.reference.ReferenceBase;
|
39
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
40
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
41
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
42
|
|
43
|
|
44
|
/**
|
45
|
* @author a.mueller
|
46
|
* @created 20.02.2010
|
47
|
* @version 1.0
|
48
|
*/
|
49
|
@Component
|
50
|
public class ErmsTaxonImport extends ErmsImportBase<TaxonBase> implements IMappingImport<TaxonBase, ErmsImportState>{
|
51
|
private static final Logger logger = Logger.getLogger(ErmsTaxonImport.class);
|
52
|
|
53
|
public static final UUID TNS_EXT_UUID = UUID.fromString("41cb0450-ac84-4d73-905e-9c7773c23b05");
|
54
|
|
55
|
private DbImportMapping mapping;
|
56
|
|
57
|
//second path is not used anymore, there is now an ErmsTaxonRelationImport class instead
|
58
|
private boolean isSecondPath = false;
|
59
|
|
60
|
private int modCount = 10000;
|
61
|
private static final String pluralString = "taxa";
|
62
|
private static final String dbTableName = "tu";
|
63
|
private static final Class cdmTargetClass = TaxonBase.class;
|
64
|
|
65
|
public ErmsTaxonImport(){
|
66
|
super(pluralString, dbTableName, cdmTargetClass);
|
67
|
}
|
68
|
|
69
|
|
70
|
|
71
|
// /* (non-Javadoc)
|
72
|
// * @see eu.etaxonomy.cdm.io.erms.ErmsImportBase#getIdQuery()
|
73
|
// */
|
74
|
// @Override
|
75
|
// protected String getIdQuery() {
|
76
|
// String strQuery = " SELECT id FROM tu WHERE id < 300000 " ;
|
77
|
// return strQuery;
|
78
|
// }
|
79
|
|
80
|
|
81
|
/* (non-Javadoc)
|
82
|
* @see eu.etaxonomy.cdm.io.erms.ErmsImportBase#getMapping()
|
83
|
*/
|
84
|
protected DbImportMapping getMapping() {
|
85
|
if (mapping == null){
|
86
|
mapping = new DbImportMapping();
|
87
|
|
88
|
mapping.addMapper(DbImportObjectCreationMapper.NewInstance(this, "id", TAXON_NAMESPACE)); //id + tu_status
|
89
|
UUID tsnUuid = ErmsTransformer.uuidTsn;
|
90
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tsn", tsnUuid, "TSN", "TSN", "TSN"));
|
91
|
// mapping.addMapper(DbImportStringMapper.NewInstance("tu_name", "(NonViralName)name.nameCache"));
|
92
|
|
93
|
UUID displayNameUuid = ErmsTransformer.uuidDisplayName;
|
94
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tu_displayname", displayNameUuid, "display name", "display name", "display name"));
|
95
|
UUID fuzzyNameUuid = ErmsTransformer.uuidFuzzyName;
|
96
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tu_fuzzyname", fuzzyNameUuid, "fuzzy name", "fuzzy name", "fuzzy name"));
|
97
|
mapping.addMapper(DbImportStringMapper.NewInstance("tu_authority", "(NonViralName)name.authorshipCache"));
|
98
|
|
99
|
UUID fossilStatusUuid = ErmsTransformer.uuidFossilStatus;
|
100
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("fossil_name", fossilStatusUuid, "fossil status", "fossil status", "fos. stat."));
|
101
|
// mapping.addMapper(DbImportExtensionTypeCreationMapper.NewInstance("fossil_name", EXTENSION_TYPE_NAMESPACE, "fossil_name", "fossil_name", "fossil_name"));
|
102
|
|
103
|
UUID credibilityUuid = ErmsTransformer.uuidCredibility;
|
104
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tu_credibility", credibilityUuid, "credibility", "credibility", "credibility")); //Werte: null, unknown, marked for deletion
|
105
|
|
106
|
UUID completenessUuid = ErmsTransformer.uuidCompleteness;
|
107
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tu_completeness", completenessUuid, "completeness", "completeness", "completeness")); //null, unknown, tmpflag, tmp2, tmp3, complete
|
108
|
|
109
|
UUID unacceptUuid = ErmsTransformer.uuidUnacceptReason;
|
110
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tu_unacceptreason", unacceptUuid, "unaccept reason", "unaccept reason", "reason"));
|
111
|
|
112
|
UUID qualityUuid = ErmsTransformer.uuidQualityStatus;
|
113
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("qualitystatus_name", qualityUuid, "quality status", "quality status", "quality status")); //checked by Tax Editor ERMS1.1, Added by db management team (2x), checked by Tax Editor
|
114
|
|
115
|
// UUID hiddenUuid = ErmsTransformer.uuidHidden;
|
116
|
// mapping.addMapper(DbImportMarkerCreationMapper.Mapper.NewInstance("qualitystatus_name", qualityUuid, "quality status", "quality status", "quality status")); //checked by Tax Editor ERMS1.1, Added by db management team (2x), checked by Tax Editor
|
117
|
|
118
|
|
119
|
//ignore
|
120
|
mapping.addMapper(DbIgnoreMapper.NewInstance("tu_marine", "marine flag not implemented in PESI"));
|
121
|
mapping.addMapper(DbIgnoreMapper.NewInstance("tu_brackish", "brackish flag not implemented in PESI"));
|
122
|
mapping.addMapper(DbIgnoreMapper.NewInstance("tu_fresh", "freshwater flag not implemented in PESI"));
|
123
|
mapping.addMapper(DbIgnoreMapper.NewInstance("tu_terrestrial", "terrestrial flag not implemented in PESI"));
|
124
|
mapping.addMapper(DbIgnoreMapper.NewInstance("tu_fossil", "tu_fossil implemented as foreign key"));
|
125
|
mapping.addMapper(DbIgnoreMapper.NewInstance("cache_citation", "citation cache not needed in PESI"));
|
126
|
mapping.addMapper(DbNotYetImplementedMapper.NewInstance("tu_sp", "included in rank/object creation"));
|
127
|
|
128
|
|
129
|
//not yet implemented or ignore
|
130
|
mapping.addMapper(DbNotYetImplementedMapper.NewInstance("tu_hidden", "Needs DbImportMarkerMapper implemented"));
|
131
|
|
132
|
// //second path / implemented in ErmsTaxonRelationImport
|
133
|
// DbImportMapping secondPathMapping = new DbImportMapping();
|
134
|
// secondPathMapping.addMapper(DbImportTaxIncludedInMapper.NewInstance("id", "tu_parent", TAXON_NAMESPACE, null)); //there is only one tree
|
135
|
// secondPathMapping.addMapper(DbImportSynonymMapper.NewInstance("id", "tu_acctaxon", TAXON_NAMESPACE, null));
|
136
|
// secondPathMapping.addMapper(DbImportNameTypeDesignationMapper.NewInstance("id", "tu_typetaxon", NAME_NAMESPACE, "tu_typedesignationstatus"));
|
137
|
// secondPathMapping.addMapper(DbNotYetImplementedMapper.NewInstance("tu_acctaxon"));
|
138
|
// mapping.setSecondPathMapping(secondPathMapping);
|
139
|
|
140
|
}
|
141
|
return mapping;
|
142
|
}
|
143
|
|
144
|
/* (non-Javadoc)
|
145
|
* @see eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportBase#getRecordQuery(eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportConfigurator)
|
146
|
*/
|
147
|
@Override
|
148
|
protected String getRecordQuery(ErmsImportConfigurator config) {
|
149
|
String strSelect = " SELECT tu.*, parent1.tu_name AS parent1name, parent2.tu_name AS parent2name, parent3.tu_name AS parent3name, "
|
150
|
+ " parent1.tu_rank AS parent1rank, parent2.tu_rank AS parent2rank, parent3.tu_rank AS parent3rank, " +
|
151
|
" status.status_id as status_id, fossil.fossil_name, qualitystatus.qualitystatus_name";
|
152
|
String strFrom = " FROM tu LEFT OUTER JOIN tu AS parent1 ON parent1.id = tu.tu_parent " +
|
153
|
" LEFT OUTER JOIN tu AS parent2 ON parent2.id = parent1.tu_parent " +
|
154
|
" LEFT OUTER JOIN tu AS parent3 ON parent2.tu_parent = parent3.id " +
|
155
|
" LEFT OUTER JOIN status ON tu.tu_status = status.status_id " +
|
156
|
" LEFT OUTER JOIN fossil ON tu.tu_fossil = fossil.fossil_id " +
|
157
|
" LEFT OUTER JOIN qualitystatus ON tu.tu_qualitystatus = qualitystatus.id ";
|
158
|
String strWhere = " WHERE ( tu.id IN (" + ID_LIST_TOKEN + ") )";
|
159
|
String strRecordQuery = strSelect + strFrom + strWhere;
|
160
|
return strRecordQuery;
|
161
|
}
|
162
|
|
163
|
|
164
|
// /**
|
165
|
// * @param config
|
166
|
// * @return
|
167
|
// */
|
168
|
// private String getSecondPathRecordQuery(ErmsImportConfigurator config) {
|
169
|
// //TODO get automatic by second path mappers
|
170
|
// String selectAttributes = "id, tu_parent, tu_typetaxon, tu_typetaxon, tu_typedesignation, tu_acctaxon, tu_status";
|
171
|
// String strRecordQuery =
|
172
|
// " SELECT " + selectAttributes +
|
173
|
// " FROM tu " +
|
174
|
// " WHERE ( tu.id IN (" + ID_LIST_TOKEN + ") )";
|
175
|
// return strRecordQuery;
|
176
|
// }
|
177
|
|
178
|
|
179
|
// private String getSecondPathIdQuery(){
|
180
|
// return getIdQuery();
|
181
|
// }
|
182
|
|
183
|
/* (non-Javadoc)
|
184
|
* @see eu.etaxonomy.cdm.io.erms.ErmsImportBase#doInvoke(eu.etaxonomy.cdm.io.erms.ErmsImportState)
|
185
|
*/
|
186
|
@Override
|
187
|
protected boolean doInvoke(ErmsImportState state) {
|
188
|
//first path
|
189
|
boolean success = super.doInvoke(state);
|
190
|
|
191
|
// //second path
|
192
|
// isSecondPath = true;
|
193
|
// ErmsImportConfigurator config = state.getConfig();
|
194
|
// Source source = config.getSource();
|
195
|
//
|
196
|
// String strIdQuery = getSecondPathIdQuery();
|
197
|
// String strRecordQuery = getSecondPathRecordQuery(config);
|
198
|
//
|
199
|
// int recordsPerTransaction = config.getRecordsPerTransaction();
|
200
|
// try{
|
201
|
// ResultSetPartitioner partitioner = ResultSetPartitioner.NewInstance(source, strIdQuery, strRecordQuery, recordsPerTransaction);
|
202
|
// while (partitioner.nextPartition()){
|
203
|
// partitioner.doPartition(this, state);
|
204
|
// }
|
205
|
// } catch (SQLException e) {
|
206
|
// logger.error("SQLException:" + e);
|
207
|
// return false;
|
208
|
// }
|
209
|
//
|
210
|
// isSecondPath = false;
|
211
|
//
|
212
|
// logger.info("end make " + getPluralString() + " ... " + getSuccessString(success));
|
213
|
return success;
|
214
|
|
215
|
}
|
216
|
|
217
|
|
218
|
|
219
|
/* (non-Javadoc)
|
220
|
* @see eu.etaxonomy.cdm.io.berlinModel.in.IPartitionedIO#getRelatedObjectsForPartition(java.sql.ResultSet)
|
221
|
*/
|
222
|
public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs) {
|
223
|
String nameSpace;
|
224
|
Class cdmClass;
|
225
|
Set<String> idSet;
|
226
|
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<Object, Map<String, ? extends CdmBase>>();
|
227
|
|
228
|
try{
|
229
|
Set<String> nameIdSet = new HashSet<String>();
|
230
|
Set<String> referenceIdSet = new HashSet<String>();
|
231
|
while (rs.next()){
|
232
|
// handleForeignKey(rs, nameIdSet, "PTNameFk");
|
233
|
// handleForeignKey(rs, referenceIdSet, "PTRefFk");
|
234
|
}
|
235
|
|
236
|
//reference map
|
237
|
// nameSpace = "Reference";
|
238
|
// cdmClass = ReferenceBase.class;
|
239
|
// Map<String, Person> referenceMap = (Map<String, Person>)getCommonService().getSourcedObjectsByIdInSource(Person.class, teamIdSet, nameSpace);
|
240
|
// result.put(ReferenceBase.class, referenceMap);
|
241
|
|
242
|
} catch (SQLException e) {
|
243
|
throw new RuntimeException(e);
|
244
|
}
|
245
|
return result;
|
246
|
}
|
247
|
|
248
|
|
249
|
/* (non-Javadoc)
|
250
|
* @see eu.etaxonomy.cdm.io.common.mapping.IMappingImport#createObject(java.sql.ResultSet)
|
251
|
*/
|
252
|
public TaxonBase createObject(ResultSet rs, ErmsImportState state) throws SQLException {
|
253
|
int statusId = rs.getInt("status_id");
|
254
|
String tuName = rs.getString("tu_name");
|
255
|
String displayName = rs.getString("tu_displayname");
|
256
|
|
257
|
String parent1Name = rs.getString("parent1name");
|
258
|
Integer parent1Rank = rs.getInt("parent1rank");
|
259
|
|
260
|
String parent2Name = rs.getString("parent2name");
|
261
|
Integer parent2Rank = rs.getInt("parent2rank");
|
262
|
|
263
|
String parent3Name = rs.getString("parent3name");
|
264
|
Integer parent3Rank = rs.getInt("parent3rank");
|
265
|
|
266
|
|
267
|
NonViralName taxonName = getTaxonName(rs, state);
|
268
|
//set epithets
|
269
|
if (taxonName.isGenus() || taxonName.isSupraGeneric()){
|
270
|
taxonName.setGenusOrUninomial(tuName);
|
271
|
}else if (taxonName.isInfraGeneric()){
|
272
|
taxonName.setInfraGenericEpithet(tuName);
|
273
|
taxonName.setGenusOrUninomial(parent1Name);
|
274
|
}else if (taxonName.isSpecies()){
|
275
|
taxonName.setSpecificEpithet(tuName);
|
276
|
getGenusAndInfraGenus(parent1Name, parent2Name, parent1Rank, taxonName);
|
277
|
}else if (taxonName.isInfraSpecific()){
|
278
|
if (parent1Rank < 220){
|
279
|
handleException(parent1Rank, taxonName, displayName);
|
280
|
}
|
281
|
taxonName.setInfraSpecificEpithet(tuName);
|
282
|
taxonName.setSpecificEpithet(parent1Name);
|
283
|
getGenusAndInfraGenus(parent2Name, parent3Name, parent2Rank, taxonName);
|
284
|
}else if (taxonName.getRank()== null){
|
285
|
logger.warn("rank super domain still needs to be implemented. Used domain instead.");
|
286
|
if ("Biota".equalsIgnoreCase(tuName)){
|
287
|
Rank rank = Rank.DOMAIN(); //should be Superdomain
|
288
|
taxonName.setRank(rank);
|
289
|
taxonName.setGenusOrUninomial(tuName);
|
290
|
}else{
|
291
|
String warning = "TaxonName has no rank. Use namecache.";
|
292
|
logger.warn(warning);
|
293
|
taxonName.setNameCache(tuName);
|
294
|
}
|
295
|
|
296
|
}
|
297
|
//e.g. Leucon [Platyhelminthes] ornatus
|
298
|
if (containsBrackets(displayName)){
|
299
|
taxonName.setNameCache(displayName);
|
300
|
logger.warn("Set name cache: " + displayName);
|
301
|
}
|
302
|
|
303
|
//add original source for taxon name (taxon original source is added in mapper
|
304
|
ReferenceBase citation = state.getConfig().getSourceReference();
|
305
|
addOriginalSource(rs, taxonName, "id", NAME_NAMESPACE, citation);
|
306
|
|
307
|
// taxonName.setNameCache("Test");
|
308
|
|
309
|
ErmsImportConfigurator config = state.getConfig();
|
310
|
ReferenceBase sec = config.getSourceReference();
|
311
|
if (statusId == 1){
|
312
|
return Taxon.NewInstance(taxonName, sec);
|
313
|
}else{
|
314
|
return Synonym.NewInstance(taxonName, sec);
|
315
|
}
|
316
|
}
|
317
|
|
318
|
|
319
|
|
320
|
/**
|
321
|
* @param parent1Rank
|
322
|
* @param displayName
|
323
|
* @param taxonName
|
324
|
*/
|
325
|
private void handleException(Integer parent1Rank, NonViralName taxonName, String displayName) {
|
326
|
logger.warn("Parent of infra specific taxon is higher than species. Used nameCache: " + displayName) ;
|
327
|
taxonName.setNameCache(displayName);
|
328
|
}
|
329
|
|
330
|
|
331
|
|
332
|
/**
|
333
|
* @param displayName
|
334
|
* @return
|
335
|
*/
|
336
|
private boolean containsBrackets(String displayName) {
|
337
|
int index = displayName.indexOf("[");
|
338
|
return (index > -1);
|
339
|
}
|
340
|
|
341
|
|
342
|
|
343
|
/**
|
344
|
* @param parent1Name
|
345
|
* @param parent2Name
|
346
|
* @param parent1Rank
|
347
|
* @param taxonName
|
348
|
*/
|
349
|
private void getGenusAndInfraGenus(String parentName, String grandParentName, Integer parent1Rank, NonViralName taxonName) {
|
350
|
if (parent1Rank <220 && parent1Rank > 180){
|
351
|
//parent is infrageneric
|
352
|
taxonName.setInfraGenericEpithet(parentName);
|
353
|
taxonName.setGenusOrUninomial(grandParentName);
|
354
|
}else{
|
355
|
taxonName.setGenusOrUninomial(parentName);
|
356
|
}
|
357
|
}
|
358
|
|
359
|
/**
|
360
|
* @param rs
|
361
|
* @return
|
362
|
* @throws SQLException
|
363
|
*/
|
364
|
private NonViralName getTaxonName(ResultSet rs, ErmsImportState state) throws SQLException {
|
365
|
NonViralName result;
|
366
|
Integer kingdomId = parseKingdomId(rs);
|
367
|
Integer intRank = rs.getInt("tu_rank");
|
368
|
|
369
|
NomenclaturalCode nc = ErmsTransformer.kingdomId2NomCode(kingdomId);
|
370
|
Rank rank = null;
|
371
|
if (kingdomId != null){
|
372
|
rank = state.getRank(intRank, kingdomId);
|
373
|
}
|
374
|
if (nc != null){
|
375
|
result = (NonViralName)nc.getNewTaxonNameInstance(rank);
|
376
|
}else{
|
377
|
result = NonViralName.NewInstance(rank);
|
378
|
}
|
379
|
|
380
|
return result;
|
381
|
}
|
382
|
|
383
|
/**
|
384
|
* Returns the kingdom id by extracting it from the second character in the <code>tu_sp</code>
|
385
|
* attribute. If the attribute can not be parsed to a valid id <code>null</code>
|
386
|
* is returned. If the attribute is <code>null</code> the id of the record is returned.
|
387
|
* @param rs
|
388
|
* @return
|
389
|
* @throws SQLException
|
390
|
*/
|
391
|
private int parseKingdomId(ResultSet rs) throws SQLException {
|
392
|
Integer result = null;
|
393
|
String treeString = rs.getString("tu_sp");
|
394
|
if (treeString != null){
|
395
|
if (CdmUtils.isNotEmpty(treeString) && treeString.length() > 1){
|
396
|
String strKingdom = treeString.substring(1,2);
|
397
|
|
398
|
if (! treeString.substring(0, 1).equals("#") && ! treeString.substring(2, 3).equals("#") ){
|
399
|
logger.warn("Tree string " + treeString + " has no recognized format");
|
400
|
}else{
|
401
|
try {
|
402
|
result = Integer.valueOf(strKingdom);
|
403
|
} catch (NumberFormatException e) {
|
404
|
logger.warn("Kingdom string " + strKingdom + "could not be recognized as a valid number");
|
405
|
}
|
406
|
}
|
407
|
}
|
408
|
}else{
|
409
|
Integer tu_id = rs.getInt("id");
|
410
|
result = tu_id;
|
411
|
}
|
412
|
return result;
|
413
|
}
|
414
|
|
415
|
|
416
|
/* (non-Javadoc)
|
417
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#doCheck(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
418
|
*/
|
419
|
@Override
|
420
|
protected boolean doCheck(ErmsImportState state){
|
421
|
IOValidator<ErmsImportState> validator = new ErmsTaxonImportValidator();
|
422
|
return validator.validate(state);
|
423
|
}
|
424
|
|
425
|
|
426
|
/* (non-Javadoc)
|
427
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
428
|
*/
|
429
|
protected boolean isIgnore(ErmsImportState state){
|
430
|
return ! state.getConfig().isDoTaxa();
|
431
|
}
|
432
|
|
433
|
|
434
|
|
435
|
}
|