1 |
9a971015
|
Andreas Müller
|
/**
|
2 |
|
|
* Copyright (C) 2007 EDIT
|
3 |
8422c0cd
|
Andreas Müller
|
* European Distributed Institute of Taxonomy
|
4 |
9a971015
|
Andreas Müller
|
* http://www.e-taxonomy.eu
|
5 |
8422c0cd
|
Andreas Müller
|
*
|
6 |
9a971015
|
Andreas Müller
|
* 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.indexFungorum;
|
11 |
|
|
|
12 |
|
|
import java.sql.ResultSet;
|
13 |
|
|
import java.sql.SQLException;
|
14 |
|
|
import java.util.HashMap;
|
15 |
1e41fbca
|
Andreas Müller
|
import java.util.List;
|
16 |
9a971015
|
Andreas Müller
|
import java.util.Map;
|
17 |
|
|
|
18 |
|
|
import org.apache.log4j.Logger;
|
19 |
|
|
import org.springframework.stereotype.Component;
|
20 |
1e41fbca
|
Andreas Müller
|
import org.springframework.transaction.TransactionStatus;
|
21 |
9a971015
|
Andreas Müller
|
|
22 |
1e41fbca
|
Andreas Müller
|
import eu.etaxonomy.cdm.io.pesi.out.PesiTransformer;
|
23 |
9a971015
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
24 |
e6a3d605
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.common.Marker;
|
25 |
9a971015
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.name.Rank;
|
26 |
51688d00
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
27 |
bbeda5af
|
Katja Luther
|
import eu.etaxonomy.cdm.model.name.TaxonNameFactory;
|
28 |
9a971015
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
29 |
|
|
import eu.etaxonomy.cdm.model.taxon.Classification;
|
30 |
|
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
31 |
|
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
32 |
|
|
|
33 |
|
|
/**
|
34 |
|
|
* @author a.mueller
|
35 |
a13538c8
|
Andreas Müller
|
* @since 27.02.2012
|
36 |
9a971015
|
Andreas Müller
|
*/
|
37 |
|
|
@Component
|
38 |
1e41fbca
|
Andreas Müller
|
public class IndexFungorumHigherClassificationImport extends IndexFungorumImportBase {
|
39 |
ffd6058e
|
Andreas Müller
|
|
40 |
|
|
private static final long serialVersionUID = -6841466146506309309L;
|
41 |
|
|
private static final Logger logger = Logger.getLogger(IndexFungorumHigherClassificationImport.class);
|
42 |
8422c0cd
|
Andreas Müller
|
|
43 |
9a971015
|
Andreas Müller
|
private static final String pluralString = "higher classifications";
|
44 |
|
|
private static final String dbTableName = "tblPESIfungi-Classification";
|
45 |
|
|
|
46 |
|
|
public IndexFungorumHigherClassificationImport(){
|
47 |
deed567d
|
Andreas Müller
|
super(pluralString, dbTableName);
|
48 |
9a971015
|
Andreas Müller
|
}
|
49 |
|
|
|
50 |
|
|
@Override
|
51 |
|
|
protected String getRecordQuery(IndexFungorumImportConfigurator config) {
|
52 |
8422c0cd
|
Andreas Müller
|
String strRecordQuery =
|
53 |
|
|
" SELECT DISTINCT [Kingdom name], [Phylum name], [Subphylum name], [Class name], [Subclass name], [Order name], [Family name], g.[NAME OF FUNGUS] as GenusName, c.PreferredName as SpeciesName " +
|
54 |
1e41fbca
|
Andreas Müller
|
" FROM [tblPESIfungi-Classification] c LEFT OUTER JOIN " +
|
55 |
2d481abb
|
Katja Luther
|
" tblGenera g ON c.PreferredNameFDCnumber = g.[RECORD NUMBER]" +
|
56 |
|
|
" ORDER BY [Kingdom name], [Phylum name], [Subphylum name], [Class name], [Subclass name], [Order name], [Family name], GenusName, SpeciesName ";
|
57 |
9a971015
|
Andreas Müller
|
return strRecordQuery;
|
58 |
|
|
}
|
59 |
|
|
|
60 |
|
|
@Override
|
61 |
|
|
protected void doInvoke(IndexFungorumImportState state) {
|
62 |
ffd6058e
|
Andreas Müller
|
|
63 |
|
|
logger.info("Start higher classification ...");
|
64 |
|
|
|
65 |
9a971015
|
Andreas Müller
|
String sql = getRecordQuery(state.getConfig());
|
66 |
|
|
ResultSet rs = state.getConfig().getSource().getResultSet(sql);
|
67 |
8422c0cd
|
Andreas Müller
|
|
68 |
1e41fbca
|
Andreas Müller
|
//only 1 partition here
|
69 |
8422c0cd
|
Andreas Müller
|
|
70 |
9a971015
|
Andreas Müller
|
String lastKingdom = "";
|
71 |
|
|
String lastPhylum = "";
|
72 |
|
|
String lastSubphylum = "";
|
73 |
|
|
String lastClassname = "";
|
74 |
|
|
String lastSubclass = "";
|
75 |
|
|
String lastOrder = "";
|
76 |
|
|
String lastFamily = "";
|
77 |
8422c0cd
|
Andreas Müller
|
|
78 |
9a971015
|
Andreas Müller
|
Taxon taxonKingdom = null;
|
79 |
|
|
Taxon taxonPhylum = null;
|
80 |
|
|
Taxon taxonSubphylum = null;
|
81 |
|
|
Taxon taxonClass = null;
|
82 |
|
|
Taxon taxonSubclass = null;
|
83 |
|
|
Taxon taxonOrder = null;
|
84 |
|
|
Taxon taxonFamily = null;
|
85 |
8422c0cd
|
Andreas Müller
|
|
86 |
1e41fbca
|
Andreas Müller
|
Taxon higherTaxon = null;
|
87 |
8422c0cd
|
Andreas Müller
|
|
88 |
1e41fbca
|
Andreas Müller
|
TransactionStatus tx = startTransaction();
|
89 |
|
|
ResultSet rsRelatedObjects = state.getConfig().getSource().getResultSet(sql);
|
90 |
8422c0cd
|
Andreas Müller
|
state.setRelatedObjects(getRelatedObjectsForPartition(rsRelatedObjects, state));
|
91 |
|
|
|
92 |
1e41fbca
|
Andreas Müller
|
Classification classification = getClassification(state);
|
93 |
8422c0cd
|
Andreas Müller
|
|
94 |
9a971015
|
Andreas Müller
|
try {
|
95 |
|
|
while (rs.next()){
|
96 |
2d481abb
|
Katja Luther
|
String kingdom = rs.getString("Kingdom name");
|
97 |
|
|
String phylum = rs.getString("Phylum name");
|
98 |
|
|
String subphylum = rs.getString("Subphylum name");
|
99 |
|
|
String classname = rs.getString("Class name");
|
100 |
|
|
String subclass = rs.getString("Subclass name");
|
101 |
|
|
String order = rs.getString("Order name");
|
102 |
|
|
String family = rs.getString("Family name");
|
103 |
8422c0cd
|
Andreas Müller
|
|
104 |
1e41fbca
|
Andreas Müller
|
if (isNewTaxon(family, lastFamily)){
|
105 |
|
|
if (isNewTaxon(order,lastOrder)){
|
106 |
|
|
if (isNewTaxon(subclass,lastSubclass)){
|
107 |
|
|
if (isNewTaxon(classname,lastClassname)){
|
108 |
|
|
if (isNewTaxon(subphylum, lastSubphylum)){
|
109 |
|
|
if (isNewTaxon(phylum,lastPhylum)){
|
110 |
|
|
if (isNewTaxon(kingdom,lastKingdom)){
|
111 |
9a971015
|
Andreas Müller
|
taxonKingdom = makeTaxon(state, kingdom, Rank.KINGDOM());
|
112 |
1e41fbca
|
Andreas Müller
|
lastKingdom = kingdom;
|
113 |
|
|
logger.info("Import kingdom " + kingdom);
|
114 |
|
|
getTaxonService().saveOrUpdate(taxonKingdom);
|
115 |
|
|
}else{
|
116 |
|
|
higherTaxon = taxonKingdom;
|
117 |
9a971015
|
Andreas Müller
|
}
|
118 |
1e41fbca
|
Andreas Müller
|
higherTaxon = isIncertisSedis(kingdom) ? higherTaxon : taxonKingdom;
|
119 |
e6a3d605
|
Andreas Müller
|
Rank newRank = (lastKingdom.equals("Fungi") ? null : Rank.PHYLUM());
|
120 |
|
|
taxonPhylum = makeTaxon(state, phylum, newRank);
|
121 |
1e41fbca
|
Andreas Müller
|
if (taxonPhylum != null){
|
122 |
41d91065
|
Andreas Müller
|
classification.addParentChild(higherTaxon, taxonPhylum, null, null);
|
123 |
1e41fbca
|
Andreas Müller
|
}
|
124 |
|
|
higherTaxon = isIncertisSedis(phylum) ? higherTaxon : taxonPhylum;
|
125 |
|
|
lastPhylum = phylum;
|
126 |
|
|
logger.info("Import Phylum " + phylum);
|
127 |
|
|
}else{
|
128 |
|
|
higherTaxon = taxonPhylum;
|
129 |
9a971015
|
Andreas Müller
|
}
|
130 |
e6a3d605
|
Andreas Müller
|
Rank newRank = (lastKingdom.equals("Fungi") ? null : Rank.SUBPHYLUM());
|
131 |
|
|
taxonSubphylum = makeTaxon(state, subphylum, newRank);
|
132 |
1e41fbca
|
Andreas Müller
|
if (taxonSubphylum != null){
|
133 |
41d91065
|
Andreas Müller
|
getClassification(state).addParentChild(higherTaxon,taxonSubphylum, null, null);
|
134 |
1e41fbca
|
Andreas Müller
|
}
|
135 |
|
|
higherTaxon = isIncertisSedis(subphylum) ? higherTaxon : taxonSubphylum;
|
136 |
|
|
lastSubphylum = subphylum;
|
137 |
|
|
}else{
|
138 |
|
|
higherTaxon = taxonSubphylum;
|
139 |
9a971015
|
Andreas Müller
|
}
|
140 |
|
|
taxonClass = makeTaxon(state, classname, Rank.CLASS());
|
141 |
1e41fbca
|
Andreas Müller
|
if (taxonClass != null){
|
142 |
41d91065
|
Andreas Müller
|
getClassification(state).addParentChild(higherTaxon, taxonClass, null, null);
|
143 |
1e41fbca
|
Andreas Müller
|
}
|
144 |
|
|
higherTaxon = isIncertisSedis(classname) ? higherTaxon : taxonClass;
|
145 |
|
|
lastClassname = classname;
|
146 |
|
|
}else{
|
147 |
|
|
higherTaxon = taxonClass;
|
148 |
9a971015
|
Andreas Müller
|
}
|
149 |
|
|
taxonSubclass = makeTaxon(state, subclass, Rank.SUBCLASS());
|
150 |
1e41fbca
|
Andreas Müller
|
if (taxonSubclass != null){
|
151 |
41d91065
|
Andreas Müller
|
getClassification(state).addParentChild(higherTaxon, taxonSubclass,null, null);
|
152 |
1e41fbca
|
Andreas Müller
|
}
|
153 |
|
|
higherTaxon = isIncertisSedis(subclass) ? higherTaxon : taxonSubclass;
|
154 |
|
|
lastSubclass = subclass;
|
155 |
|
|
}else{
|
156 |
|
|
higherTaxon = taxonSubclass;
|
157 |
9a971015
|
Andreas Müller
|
}
|
158 |
|
|
taxonOrder = makeTaxon(state, order, Rank.ORDER());
|
159 |
1e41fbca
|
Andreas Müller
|
if (taxonOrder != null){
|
160 |
41d91065
|
Andreas Müller
|
getClassification(state).addParentChild(higherTaxon, taxonOrder, null, null);
|
161 |
1e41fbca
|
Andreas Müller
|
}
|
162 |
|
|
higherTaxon = isIncertisSedis(order) ? higherTaxon : taxonOrder;
|
163 |
|
|
lastOrder = order;
|
164 |
|
|
}else{
|
165 |
|
|
higherTaxon = taxonOrder;
|
166 |
9a971015
|
Andreas Müller
|
}
|
167 |
|
|
taxonFamily = makeTaxon(state, family, Rank.FAMILY());
|
168 |
1e41fbca
|
Andreas Müller
|
if (taxonFamily != null){
|
169 |
8fb964f0
|
Andreas Müller
|
try{
|
170 |
bb5bac58
|
Andreas Müller
|
//if this shows a warning see single issue in #2826 about Glomerellaceae (which has 2 different parents)
|
171 |
925afb79
|
Andreas Müller
|
getClassification(state).addParentChild(higherTaxon, taxonFamily, null, null);
|
172 |
2d481abb
|
Katja Luther
|
}catch(IllegalStateException e){
|
173 |
|
|
if (e.getMessage().startsWith("The child taxon is already part of the tree")){
|
174 |
|
|
logger.warn(e.getMessage() + taxonFamily.getTitleCache() + " " + higherTaxon.getTitleCache());
|
175 |
bb5bac58
|
Andreas Müller
|
}
|
176 |
2d481abb
|
Katja Luther
|
}
|
177 |
1e41fbca
|
Andreas Müller
|
}
|
178 |
|
|
higherTaxon = isIncertisSedis(family) ? higherTaxon : taxonFamily;
|
179 |
|
|
lastFamily = family;
|
180 |
|
|
getTaxonService().saveOrUpdate(higherTaxon);
|
181 |
9a971015
|
Andreas Müller
|
}
|
182 |
1e41fbca
|
Andreas Müller
|
getTaxonService().saveOrUpdate(higherTaxon);
|
183 |
9a971015
|
Andreas Müller
|
}
|
184 |
|
|
} catch (SQLException e) {
|
185 |
|
|
e.printStackTrace();
|
186 |
|
|
logger.error(e.getMessage());
|
187 |
1e41fbca
|
Andreas Müller
|
tx.setRollbackOnly();
|
188 |
|
|
state.setSuccess(false);
|
189 |
9a971015
|
Andreas Müller
|
}
|
190 |
1e41fbca
|
Andreas Müller
|
commitTransaction(tx);
|
191 |
ffd6058e
|
Andreas Müller
|
logger.info("End higher classification ...");
|
192 |
|
|
|
193 |
9a971015
|
Andreas Müller
|
return;
|
194 |
|
|
}
|
195 |
|
|
|
196 |
1e41fbca
|
Andreas Müller
|
private boolean isIncertisSedis(String uninomial) {
|
197 |
b015905a
|
Andreas Müller
|
return uninomial.equalsIgnoreCase(INCERTAE_SEDIS) || uninomial.equalsIgnoreCase(FOSSIL_FUNGI);
|
198 |
9a971015
|
Andreas Müller
|
}
|
199 |
|
|
|
200 |
1e41fbca
|
Andreas Müller
|
private boolean isNewTaxon(String uninomial, String lastUninomial) {
|
201 |
|
|
boolean result = !uninomial.equalsIgnoreCase(lastUninomial);
|
202 |
|
|
result |= lastUninomial.equalsIgnoreCase(INCERTAE_SEDIS);
|
203 |
b015905a
|
Andreas Müller
|
result |= lastUninomial.equalsIgnoreCase(FOSSIL_FUNGI);
|
204 |
1e41fbca
|
Andreas Müller
|
return result;
|
205 |
|
|
}
|
206 |
|
|
|
207 |
e6a3d605
|
Andreas Müller
|
private Taxon makeTaxon(IndexFungorumImportState state, String uninomial, Rank newRank) {
|
208 |
b015905a
|
Andreas Müller
|
if (uninomial.equalsIgnoreCase(INCERTAE_SEDIS) || uninomial.equalsIgnoreCase(FOSSIL_FUNGI)){
|
209 |
1e41fbca
|
Andreas Müller
|
return null;
|
210 |
|
|
}
|
211 |
bb5bac58
|
Andreas Müller
|
Taxon taxon = state.getRelatedObject(IndexFungorumImportBase.NAMESPACE_SUPRAGENERIC_NAMES, uninomial, Taxon.class);
|
212 |
1e41fbca
|
Andreas Müller
|
if (taxon == null){
|
213 |
e6a3d605
|
Andreas Müller
|
if (! newRank.equals(Rank.KINGDOM())){
|
214 |
2d481abb
|
Katja Luther
|
logger.warn("Taxon not found for uninomial " + uninomial);
|
215 |
1e41fbca
|
Andreas Müller
|
}
|
216 |
51688d00
|
Andreas Müller
|
TaxonName name = TaxonNameFactory.NewBotanicalInstance(newRank);
|
217 |
1e41fbca
|
Andreas Müller
|
name.setGenusOrUninomial(uninomial);
|
218 |
8422c0cd
|
Andreas Müller
|
Reference sourceReference = state.getRelatedObject(NAMESPACE_REFERENCE, SOURCE_REFERENCE, Reference.class);
|
219 |
1e41fbca
|
Andreas Müller
|
taxon = Taxon.NewInstance(name, sourceReference);
|
220 |
e6a3d605
|
Andreas Müller
|
taxon.addMarker(Marker.NewInstance(getMissingGUIDMarkerType(state), true));
|
221 |
|
|
}else if (newRank != null){
|
222 |
|
|
taxon.getName().setRank(newRank);
|
223 |
1e41fbca
|
Andreas Müller
|
}
|
224 |
|
|
return taxon;
|
225 |
9a971015
|
Andreas Müller
|
}
|
226 |
|
|
|
227 |
9900f429
|
Andreas Müller
|
@Override
|
228 |
|
|
public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs, IndexFungorumImportState state) {
|
229 |
9a971015
|
Andreas Müller
|
String nameSpace;
|
230 |
bb5bac58
|
Andreas Müller
|
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<>();
|
231 |
8422c0cd
|
Andreas Müller
|
|
232 |
9a971015
|
Andreas Müller
|
try{
|
233 |
|
|
//taxon map
|
234 |
bb5bac58
|
Andreas Müller
|
nameSpace = IndexFungorumImportBase.NAMESPACE_SUPRAGENERIC_NAMES ;
|
235 |
|
|
Map<String, TaxonBase<?>> taxonMap = new HashMap<>();
|
236 |
40bb5f10
|
Andreas Müller
|
@SuppressWarnings("unchecked")
|
237 |
|
|
List<Taxon> list = getCommonService().getHqlResult("SELECT t FROM Taxon t JOIN t.sources s WHERE s.citation.uuid = ?0", new Object[]{ PesiTransformer.uuidSourceRefIndexFungorum});
|
238 |
ba8ecd5d
|
Andreas Müller
|
for (Taxon taxon : list){
|
239 |
40bb5f10
|
Andreas Müller
|
String uninomial = CdmBase.deproxy(taxon.getName()).getGenusOrUninomial();
|
240 |
|
|
TaxonBase<?> existing = taxonMap.put(uninomial, taxon);
|
241 |
|
|
if (existing != null){
|
242 |
|
|
logger.warn("There seem to be duplicate taxa for uninomial: " + uninomial);
|
243 |
|
|
}
|
244 |
1e41fbca
|
Andreas Müller
|
}
|
245 |
9a971015
|
Andreas Müller
|
result.put(nameSpace, taxonMap);
|
246 |
8422c0cd
|
Andreas Müller
|
|
247 |
1e41fbca
|
Andreas Müller
|
//source reference
|
248 |
8422c0cd
|
Andreas Müller
|
Reference sourceReference = getReferenceService().find(PesiTransformer.uuidSourceRefIndexFungorum);
|
249 |
bb5bac58
|
Andreas Müller
|
Map<String, Reference> referenceMap = new HashMap<>();
|
250 |
1e41fbca
|
Andreas Müller
|
referenceMap.put(SOURCE_REFERENCE, sourceReference);
|
251 |
|
|
result.put(NAMESPACE_REFERENCE, referenceMap);
|
252 |
8422c0cd
|
Andreas Müller
|
|
253 |
1e41fbca
|
Andreas Müller
|
} catch (Exception e) {
|
254 |
9a971015
|
Andreas Müller
|
throw new RuntimeException(e);
|
255 |
|
|
}
|
256 |
|
|
return result;
|
257 |
|
|
}
|
258 |
8422c0cd
|
Andreas Müller
|
|
259 |
9a971015
|
Andreas Müller
|
@Override
|
260 |
|
|
protected boolean doCheck(IndexFungorumImportState state){
|
261 |
|
|
return true;
|
262 |
|
|
}
|
263 |
8422c0cd
|
Andreas Müller
|
|
264 |
9900f429
|
Andreas Müller
|
@Override
|
265 |
9a971015
|
Andreas Müller
|
protected boolean isIgnore(IndexFungorumImportState state){
|
266 |
1e41fbca
|
Andreas Müller
|
return ! state.getConfig().isDoRelTaxa();
|
267 |
9a971015
|
Andreas Müller
|
}
|
268 |
|
|
}
|