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.pesi.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.commons.lang3.StringUtils;
|
21
|
import org.apache.log4j.Logger;
|
22
|
import org.springframework.stereotype.Component;
|
23
|
|
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.DbImportLsidMapper;
|
28
|
import eu.etaxonomy.cdm.io.common.mapping.DbImportMapping;
|
29
|
import eu.etaxonomy.cdm.io.common.mapping.DbImportMarkerMapper;
|
30
|
import eu.etaxonomy.cdm.io.common.mapping.DbImportObjectCreationMapper;
|
31
|
import eu.etaxonomy.cdm.io.common.mapping.DbImportStringMapper;
|
32
|
import eu.etaxonomy.cdm.io.common.mapping.DbNotYetImplementedMapper;
|
33
|
import eu.etaxonomy.cdm.io.common.mapping.IMappingImport;
|
34
|
import eu.etaxonomy.cdm.io.pesi.erms.validation.ErmsTaxonImportValidator;
|
35
|
import eu.etaxonomy.cdm.io.pesi.out.PesiTaxonExport;
|
36
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
37
|
import eu.etaxonomy.cdm.model.common.ExtensionType;
|
38
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
39
|
import eu.etaxonomy.cdm.model.name.Rank;
|
40
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
41
|
import eu.etaxonomy.cdm.model.name.TaxonNameFactory;
|
42
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
43
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
44
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
45
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
46
|
import eu.etaxonomy.cdm.strategy.cache.name.TaxonNameDefaultCacheStrategy;
|
47
|
|
48
|
/**
|
49
|
* @author a.mueller
|
50
|
* @since 20.02.2010
|
51
|
*/
|
52
|
@Component
|
53
|
public class ErmsTaxonImport
|
54
|
extends ErmsImportBase<TaxonBase<?>>
|
55
|
implements IMappingImport<TaxonBase<?>, ErmsImportState>{
|
56
|
|
57
|
private static final long serialVersionUID = -7111568277264140051L;
|
58
|
private static final Logger logger = Logger.getLogger(ErmsTaxonImport.class);
|
59
|
|
60
|
public static final UUID TNS_EXT_UUID = UUID.fromString("41cb0450-ac84-4d73-905e-9c7773c23b05");
|
61
|
|
62
|
private DbImportMapping<ErmsImportState, ErmsImportConfigurator> mapping;
|
63
|
|
64
|
private static final String pluralString = "taxa";
|
65
|
private static final String dbTableName = "tu";
|
66
|
private static final Class<?> cdmTargetClass = TaxonBase.class;
|
67
|
|
68
|
public ErmsTaxonImport(){
|
69
|
super(pluralString, dbTableName, cdmTargetClass);
|
70
|
}
|
71
|
|
72
|
// @Override
|
73
|
// protected String getIdQuery() {
|
74
|
// String strQuery = " SELECT id FROM tu WHERE id < 300000 " ;
|
75
|
// return strQuery;
|
76
|
// }
|
77
|
|
78
|
|
79
|
@Override
|
80
|
protected DbImportMapping<ErmsImportState, ErmsImportConfigurator> getMapping() {
|
81
|
if (mapping == null){
|
82
|
mapping = new DbImportMapping<>();
|
83
|
|
84
|
mapping.addMapper(DbImportObjectCreationMapper.NewInstance(this, "id", TAXON_NAMESPACE)); //id + tu_status
|
85
|
UUID tsnUuid = ErmsTransformer.uuidTsn;
|
86
|
mapping.addMapper(DbImportLsidMapper.NewInstance("GUID", "lsid"));
|
87
|
|
88
|
ExtensionType tsnExtType = getExtensionType(tsnUuid, "TSN", "TSN", "TSN");
|
89
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tsn", tsnExtType));
|
90
|
// mapping.addMapper(DbImportStringMapper.NewInstance("tu_name", "(NonViralName)name.nameCache"));
|
91
|
|
92
|
ExtensionType displayNameExtType = getExtensionType(ErmsTransformer.uuidDisplayName, "display name", "display name", "display name");
|
93
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tu_displayname", displayNameExtType));
|
94
|
ExtensionType fuzzyNameExtType = getExtensionType(ErmsTransformer.uuidFuzzyName, "fuzzy name", "fuzzy name", "fuzzy name");
|
95
|
// mapping.addMapper(DbImportExtensionMapper.NewInstance("tu_fuzzyname", fuzzyNameExtType));
|
96
|
mapping.addMapper(DbImportStringMapper.NewInstance("tu_authority", "name.authorshipCache"));
|
97
|
|
98
|
ExtensionType fossilStatusExtType = getExtensionType(ErmsTransformer.uuidFossilStatus, "fossil status", "fossil status", "fos. stat.");
|
99
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("fossil_name", fossilStatusExtType));
|
100
|
// mapping.addMapper(DbImportExtensionTypeCreationMapper.NewInstance("fossil_name", EXTENSION_TYPE_NAMESPACE, "fossil_name", "fossil_name", "fossil_name"));
|
101
|
|
102
|
ExtensionType unacceptExtType = getExtensionType(ErmsTransformer.uuidUnacceptReason, "unaccept reason", "unaccept reason", "reason");
|
103
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("tu_unacceptreason", unacceptExtType));
|
104
|
|
105
|
ExtensionType qualityStatusExtType = getExtensionType(ErmsTransformer.uuidQualityStatus, "quality status", "quality status", "quality status");
|
106
|
mapping.addMapper(DbImportExtensionMapper.NewInstance("qualitystatus_name", qualityStatusExtType)); //checked by Tax Editor ERMS1.1, Added by db management team (2x), checked by Tax Editor
|
107
|
|
108
|
mapping.addMapper(DbImportMarkerMapper.NewInstance("tu_marine", ErmsTransformer.uuidMarkerMarine, "marine", "marine", "marine", null));
|
109
|
mapping.addMapper(DbImportMarkerMapper.NewInstance("tu_brackish", ErmsTransformer.uuidMarkerBrackish, "brackish", "brackish", "brackish", null));
|
110
|
mapping.addMapper(DbImportMarkerMapper.NewInstance("tu_fresh", ErmsTransformer.uuidMarkerFreshwater, "freshwater", "fresh", "fresh", null));
|
111
|
mapping.addMapper(DbImportMarkerMapper.NewInstance("tu_terrestrial", ErmsTransformer.uuidMarkerTerrestrial, "terrestrial", "terrestrial", "terrestrial", null));
|
112
|
|
113
|
// UUID hiddenUuid = ErmsTransformer.uuidHidden;
|
114
|
// 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
|
115
|
|
116
|
//not yet implemented
|
117
|
mapping.addMapper(DbNotYetImplementedMapper.NewInstance("tu_sp", "included in rank/object creation"));
|
118
|
mapping.addMapper(DbIgnoreMapper.NewInstance("cache_citation", "Needs check if this is needed in PESI"));
|
119
|
|
120
|
//ignore
|
121
|
mapping.addMapper(DbIgnoreMapper.NewInstance("tu_fossil", "tu_fossil implemented as foreign key"));
|
122
|
|
123
|
}
|
124
|
return mapping;
|
125
|
}
|
126
|
|
127
|
@Override
|
128
|
protected String getRecordQuery(ErmsImportConfigurator config) {
|
129
|
String strSelect = " SELECT tu.*, parent1.tu_name AS parent1name, parent2.tu_name AS parent2name, parent3.tu_name AS parent3name, "
|
130
|
+ " parent1.tu_rank AS parent1rank, parent2.tu_rank AS parent2rank, parent3.tu_rank AS parent3rank, " +
|
131
|
" status.status_id as status_id, status.status_name, fossil.fossil_name, qualitystatus.qualitystatus_name";
|
132
|
String strFrom = " FROM tu LEFT OUTER JOIN tu AS parent1 ON parent1.id = tu.tu_parent " +
|
133
|
" LEFT OUTER JOIN tu AS parent2 ON parent2.id = parent1.tu_parent " +
|
134
|
" LEFT OUTER JOIN tu AS parent3 ON parent2.tu_parent = parent3.id " +
|
135
|
" LEFT OUTER JOIN status ON tu.tu_status = status.status_id " +
|
136
|
" LEFT OUTER JOIN fossil ON tu.tu_fossil = fossil.fossil_id " +
|
137
|
" LEFT OUTER JOIN qualitystatus ON tu.tu_qualitystatus = qualitystatus.id ";
|
138
|
String strWhere = " WHERE ( tu.id IN (" + ID_LIST_TOKEN + ") )";
|
139
|
String strRecordQuery = strSelect + strFrom + strWhere;
|
140
|
return strRecordQuery;
|
141
|
}
|
142
|
|
143
|
@Override
|
144
|
protected void doInvoke(ErmsImportState state) {
|
145
|
state.setAcceptedTaxaKeys(getAcceptedTaxaKeys(state));
|
146
|
|
147
|
//first path
|
148
|
super.doInvoke(state);
|
149
|
return;
|
150
|
}
|
151
|
|
152
|
private Set<Integer> getAcceptedTaxaKeys(ErmsImportState state) {
|
153
|
Set<Integer> result = new HashSet<>();
|
154
|
String parentCol = "tu_parent";
|
155
|
String accCol = " tu_acctaxon";
|
156
|
String idCol = " id ";
|
157
|
String tuFk = "tu_id";
|
158
|
String taxonTable = "tu";
|
159
|
String vernacularsTable = "vernaculars";
|
160
|
String distributionTable = "dr";
|
161
|
String sql =
|
162
|
" SELECT DISTINCT %s FROM %s " //fk to parent
|
163
|
+ " UNION SELECT %s FROM %s WHERE %s is NULL" //id of taxa not having accepted taxon
|
164
|
+ " UNION SELECT DISTINCT %s FROM %s " //fk to accepted taxon
|
165
|
+ " UNION SELECT DISTINCT %s FROM %s " //vernaculars
|
166
|
+ " UNION SELECT DISTINCT %s FROM %s "; //distributions
|
167
|
sql = String.format(sql,
|
168
|
parentCol, taxonTable,
|
169
|
idCol, taxonTable, accCol,
|
170
|
accCol, taxonTable,
|
171
|
tuFk, vernacularsTable,
|
172
|
tuFk, distributionTable);
|
173
|
ResultSet rs = state.getConfig().getSource().getResultSet(sql);
|
174
|
try {
|
175
|
while (rs.next()){
|
176
|
Integer id;
|
177
|
id = rs.getInt(parentCol);
|
178
|
result.add(id);
|
179
|
}
|
180
|
return result;
|
181
|
} catch (SQLException e) {
|
182
|
e.printStackTrace();
|
183
|
throw new RuntimeException(e);
|
184
|
}
|
185
|
}
|
186
|
|
187
|
@Override
|
188
|
public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs, ErmsImportState state) {
|
189
|
// String nameSpace;
|
190
|
// Class<?> cdmClass;
|
191
|
// Set<String> idSet;
|
192
|
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<>();
|
193
|
|
194
|
try{
|
195
|
// Set<String> referenceIdSet = new HashSet<>();
|
196
|
while (rs.next()){
|
197
|
// handleForeignKey(rs, referenceIdSet, "PTRefFk");
|
198
|
}
|
199
|
|
200
|
//reference map
|
201
|
// nameSpace = "Reference";
|
202
|
// cdmClass = Reference.class;
|
203
|
// Map<String, Person> referenceMap = (Map<String, Person>)getCommonService().getSourcedObjectsByIdInSource(Person.class, teamIdSet, nameSpace);
|
204
|
// result.put(Reference.class, referenceMap);
|
205
|
|
206
|
} catch (SQLException e) {
|
207
|
throw new RuntimeException(e);
|
208
|
}
|
209
|
return result;
|
210
|
}
|
211
|
|
212
|
@Override
|
213
|
public TaxonBase<?> createObject(ResultSet rs, ErmsImportState state) throws SQLException {
|
214
|
int statusId = rs.getInt("status_id");
|
215
|
// Object accTaxonId = rs.getObject("tu_acctaxon");
|
216
|
Integer meId = rs.getInt("id");
|
217
|
|
218
|
String tuName = rs.getString("tu_name");
|
219
|
String displayName = rs.getString("tu_displayname");
|
220
|
|
221
|
String parent1Name = rs.getString("parent1name");
|
222
|
Integer parent1Rank = rs.getInt("parent1rank");
|
223
|
|
224
|
String parent2Name = rs.getString("parent2name");
|
225
|
Integer parent2Rank = rs.getInt("parent2rank");
|
226
|
|
227
|
String parent3Name = rs.getString("parent3name");
|
228
|
// Integer parent3Rank = rs.getInt("parent3rank");
|
229
|
|
230
|
TaxonName taxonName = getTaxonName(rs, state);
|
231
|
//set epithets
|
232
|
if (taxonName.isGenus() || taxonName.isSupraGeneric()){
|
233
|
taxonName.setGenusOrUninomial(tuName);
|
234
|
}else if (taxonName.isInfraGeneric()){
|
235
|
taxonName.setInfraGenericEpithet(tuName);
|
236
|
taxonName.setGenusOrUninomial(parent1Name);
|
237
|
}else if (taxonName.isSpecies()){
|
238
|
taxonName.setSpecificEpithet(tuName);
|
239
|
getGenusAndInfraGenus(parent1Name, parent2Name, parent1Rank, taxonName);
|
240
|
}else if (taxonName.isInfraSpecific()){
|
241
|
if (parent1Rank < 220){
|
242
|
handleException(parent1Rank, taxonName, displayName, meId);
|
243
|
}
|
244
|
taxonName.setInfraSpecificEpithet(tuName);
|
245
|
taxonName.setSpecificEpithet(parent1Name);
|
246
|
getGenusAndInfraGenus(parent2Name, parent3Name, parent2Rank, taxonName);
|
247
|
}else if (taxonName.getRank()== null){
|
248
|
logger.warn("rank super domain still needs to be implemented. Used domain instead.");
|
249
|
if ("Biota".equalsIgnoreCase(tuName)){
|
250
|
Rank rank = Rank.DOMAIN(); //should be Superdomain
|
251
|
taxonName.setRank(rank);
|
252
|
taxonName.setGenusOrUninomial(tuName);
|
253
|
}else{
|
254
|
String warning = "TaxonName has no rank. Use namecache.";
|
255
|
logger.warn(warning);
|
256
|
taxonName.setNameCache(tuName);
|
257
|
}
|
258
|
|
259
|
}
|
260
|
//e.g. Leucon [Platyhelminthes] ornatus
|
261
|
if (containsBrackets(displayName)){
|
262
|
taxonName.setNameCache(displayName);
|
263
|
logger.warn("Set name cache: " + displayName + ";id =" + meId);
|
264
|
}
|
265
|
|
266
|
//add original source for taxon name (taxon original source is added in mapper
|
267
|
Reference citation = state.getTransactionalSourceReference();
|
268
|
addOriginalSource(rs, taxonName, "id", NAME_NAMESPACE, citation);
|
269
|
|
270
|
//old: if (statusId == 1){
|
271
|
if (state.getAcceptedTaxaKeys().contains(meId)){
|
272
|
Taxon result = Taxon.NewInstance(taxonName, citation);
|
273
|
if (statusId != 1){
|
274
|
logger.info("Taxon created as taxon but has status <> 1 ("+statusId+"): " + meId);
|
275
|
handleNotAcceptedTaxon(result, statusId, state, rs);
|
276
|
}
|
277
|
return result;
|
278
|
}else{
|
279
|
return Synonym.NewInstance(taxonName, citation);
|
280
|
}
|
281
|
}
|
282
|
|
283
|
private void handleNotAcceptedTaxon(Taxon taxon, int statusId, ErmsImportState state, ResultSet rs) throws SQLException {
|
284
|
ExtensionType notAccExtensionType = getExtensionType(state, ErmsTransformer.uuidErmsTaxonStatus, "ERMS taxon status", "ERMS taxon status", "status", null);
|
285
|
String statusName = rs.getString("status_name");
|
286
|
|
287
|
if (statusId > 1){
|
288
|
taxon.addExtension(statusName, notAccExtensionType);
|
289
|
}
|
290
|
}
|
291
|
|
292
|
/**
|
293
|
* @param parent1Rank
|
294
|
* @param displayName
|
295
|
* @param taxonName
|
296
|
* @param meId
|
297
|
*/
|
298
|
private void handleException(Integer parent1Rank, TaxonName taxonName, String displayName, Integer meId) {
|
299
|
logger.warn("Parent of infra specific taxon is higher than species. Used nameCache: " + displayName + "; id=" + meId) ;
|
300
|
taxonName.setNameCache(displayName);
|
301
|
}
|
302
|
|
303
|
/**
|
304
|
* @param displayName
|
305
|
* @return
|
306
|
*/
|
307
|
private boolean containsBrackets(String displayName) {
|
308
|
int index = displayName.indexOf("[");
|
309
|
return (index > -1);
|
310
|
}
|
311
|
|
312
|
/**
|
313
|
* @param parent1Name
|
314
|
* @param parent2Name
|
315
|
* @param parent1Rank
|
316
|
* @param taxonName
|
317
|
*/
|
318
|
private void getGenusAndInfraGenus(String parentName, String grandParentName, Integer parent1Rank, TaxonName taxonName) {
|
319
|
if (parent1Rank <220 && parent1Rank > 180){
|
320
|
//parent is infrageneric
|
321
|
taxonName.setInfraGenericEpithet(parentName);
|
322
|
taxonName.setGenusOrUninomial(grandParentName);
|
323
|
}else{
|
324
|
taxonName.setGenusOrUninomial(parentName);
|
325
|
}
|
326
|
}
|
327
|
|
328
|
/**
|
329
|
* Returns an empty Taxon Name instance according to the given rank and kingdom.
|
330
|
* @param rs
|
331
|
* @return
|
332
|
* @throws SQLException
|
333
|
*/
|
334
|
private TaxonName getTaxonName(ResultSet rs, ErmsImportState state) throws SQLException {
|
335
|
TaxonName result;
|
336
|
Integer kingdomId = parseKingdomId(rs);
|
337
|
Integer intRank = rs.getInt("tu_rank");
|
338
|
|
339
|
NomenclaturalCode nc = ErmsTransformer.kingdomId2NomCode(kingdomId);
|
340
|
Rank rank = null;
|
341
|
if (kingdomId != null){
|
342
|
rank = state.getRank(intRank, kingdomId);
|
343
|
}else{
|
344
|
logger.warn("KingdomId is null");
|
345
|
}
|
346
|
if (rank == null){
|
347
|
logger.warn("Rank is null. KingdomId: " + kingdomId + ", rankId: " + intRank);
|
348
|
}
|
349
|
if (nc != null){
|
350
|
result = nc.getNewTaxonNameInstance(rank);
|
351
|
}else{
|
352
|
result = TaxonNameFactory.NewNonViralInstance(rank);
|
353
|
}
|
354
|
//cache strategy
|
355
|
if (result.isZoological()){
|
356
|
TaxonNameDefaultCacheStrategy cacheStrategy = PesiTaxonExport.zooNameStrategy;
|
357
|
result.setCacheStrategy(cacheStrategy);
|
358
|
}
|
359
|
|
360
|
return result;
|
361
|
}
|
362
|
|
363
|
/**
|
364
|
* Returns the kingdom id by extracting it from the second character in the <code>tu_sp</code>
|
365
|
* attribute. If the attribute can not be parsed to a valid id <code>null</code>
|
366
|
* is returned. If the attribute is <code>null</code> the id of the record is returned.
|
367
|
* @param rs
|
368
|
* @return
|
369
|
* @throws SQLException
|
370
|
*/
|
371
|
private int parseKingdomId(ResultSet rs) throws SQLException {
|
372
|
Integer result = null;
|
373
|
String treeString = rs.getString("tu_sp");
|
374
|
if (treeString != null){
|
375
|
if (StringUtils.isNotBlank(treeString) && treeString.length() > 1){
|
376
|
String strKingdom = treeString.substring(1,2);
|
377
|
|
378
|
if (! treeString.substring(0, 1).equals("#") && ! treeString.substring(2, 3).equals("#") ){
|
379
|
logger.warn("Tree string " + treeString + " has no recognized format");
|
380
|
}else{
|
381
|
try {
|
382
|
result = Integer.valueOf(strKingdom);
|
383
|
} catch (NumberFormatException e) {
|
384
|
logger.warn("Kingdom string " + strKingdom + "could not be recognized as a valid number");
|
385
|
}
|
386
|
}
|
387
|
}
|
388
|
}else{
|
389
|
Integer tu_id = rs.getInt("id");
|
390
|
result = tu_id;
|
391
|
}
|
392
|
return result;
|
393
|
}
|
394
|
|
395
|
@Override
|
396
|
protected boolean doCheck(ErmsImportState state){
|
397
|
IOValidator<ErmsImportState> validator = new ErmsTaxonImportValidator();
|
398
|
return validator.validate(state);
|
399
|
}
|
400
|
|
401
|
@Override
|
402
|
protected boolean isIgnore(ErmsImportState state){
|
403
|
return ! state.getConfig().isDoTaxa();
|
404
|
}
|
405
|
}
|