1 |
285d1929
|
p.kelbert
|
// $Id$
|
2 |
|
|
/**
|
3 |
|
|
* Copyright (C) 2013 EDIT
|
4 |
|
|
* European Distributed Institute of Taxonomy
|
5 |
|
|
* http://www.e-taxonomy.eu
|
6 |
|
|
*
|
7 |
|
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
8 |
|
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
9 |
|
|
*/
|
10 |
|
|
package eu.etaxonomy.cdm.io.taxonx2013;
|
11 |
|
|
|
12 |
2b5a7979
|
p.kelbert
|
import java.io.BufferedWriter;
|
13 |
285d1929
|
p.kelbert
|
import java.io.File;
|
14 |
|
|
import java.io.FileWriter;
|
15 |
|
|
import java.io.IOException;
|
16 |
|
|
import java.net.URI;
|
17 |
|
|
import java.util.ArrayList;
|
18 |
|
|
import java.util.HashMap;
|
19 |
|
|
import java.util.List;
|
20 |
|
|
import java.util.Map;
|
21 |
|
|
import java.util.Set;
|
22 |
2b5a7979
|
p.kelbert
|
import java.util.UUID;
|
23 |
285d1929
|
p.kelbert
|
import java.util.regex.Pattern;
|
24 |
|
|
|
25 |
|
|
import javax.xml.transform.TransformerException;
|
26 |
|
|
import javax.xml.transform.TransformerFactoryConfigurationError;
|
27 |
|
|
|
28 |
|
|
import org.apache.commons.lang.StringUtils;
|
29 |
6c41cae2
|
p.kelbert
|
import org.apache.log4j.Logger;
|
30 |
285d1929
|
p.kelbert
|
import org.w3c.dom.Node;
|
31 |
|
|
import org.w3c.dom.NodeList;
|
32 |
|
|
|
33 |
136e16b1
|
Andreas Müller
|
import com.ibm.lsid.MalformedLSIDException;
|
34 |
|
|
|
35 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade;
|
36 |
2b5a7979
|
p.kelbert
|
import eu.etaxonomy.cdm.api.service.pager.Pager;
|
37 |
|
|
import eu.etaxonomy.cdm.model.agent.AgentBase;
|
38 |
|
|
import eu.etaxonomy.cdm.model.agent.Person;
|
39 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
40 |
|
|
import eu.etaxonomy.cdm.model.common.IdentifiableSource;
|
41 |
136e16b1
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.common.LSID;
|
42 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.model.common.Language;
|
43 |
59c2c696
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.common.OriginalSourceType;
|
44 |
2b5a7979
|
p.kelbert
|
import eu.etaxonomy.cdm.model.common.UuidAndTitleCache;
|
45 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.model.description.Feature;
|
46 |
2b5a7979
|
p.kelbert
|
import eu.etaxonomy.cdm.model.description.FeatureNode;
|
47 |
|
|
import eu.etaxonomy.cdm.model.description.FeatureTree;
|
48 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.model.description.IndividualsAssociation;
|
49 |
|
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
50 |
|
|
import eu.etaxonomy.cdm.model.description.TaxonNameDescription;
|
51 |
|
|
import eu.etaxonomy.cdm.model.description.TextData;
|
52 |
136e16b1
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.name.BacterialName;
|
53 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.model.name.BotanicalName;
|
54 |
|
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
55 |
1de96f38
|
p.kelbert
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatus;
|
56 |
|
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType;
|
57 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.model.name.NonViralName;
|
58 |
|
|
import eu.etaxonomy.cdm.model.name.Rank;
|
59 |
|
|
import eu.etaxonomy.cdm.model.name.TaxonNameBase;
|
60 |
|
|
import eu.etaxonomy.cdm.model.name.ZoologicalName;
|
61 |
59c2c696
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.occurrence.DerivedUnit;
|
62 |
136e16b1
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationType;
|
63 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
64 |
|
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
65 |
|
|
import eu.etaxonomy.cdm.model.taxon.Classification;
|
66 |
|
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
67 |
|
|
import eu.etaxonomy.cdm.model.taxon.SynonymRelationshipType;
|
68 |
|
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
69 |
136e16b1
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
70 |
2b5a7979
|
p.kelbert
|
import eu.etaxonomy.cdm.model.taxon.TaxonNode;
|
71 |
|
|
import eu.etaxonomy.cdm.persistence.query.MatchMode;
|
72 |
285d1929
|
p.kelbert
|
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException;
|
73 |
|
|
import eu.etaxonomy.cdm.strategy.parser.INonViralNameParser;
|
74 |
|
|
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl;
|
75 |
|
|
|
76 |
|
|
/**
|
77 |
|
|
* @author pkelbert
|
78 |
|
|
* @date 2 avr. 2013
|
79 |
|
|
*
|
80 |
|
|
*/
|
81 |
|
|
public class TaxonXTreatmentExtractor extends TaxonXExtractor{
|
82 |
|
|
|
83 |
6c41cae2
|
p.kelbert
|
private static final Logger logger = Logger.getLogger(TaxonXTreatmentExtractor.class);
|
84 |
|
|
|
85 |
2b5a7979
|
p.kelbert
|
private static final String notMarkedUp = "Not marked-up";
|
86 |
|
|
private static final UUID proIbioTreeUUID = UUID.fromString("2c49f506-c7f7-44de-a8b9-2e695de3769c");
|
87 |
|
|
private static final UUID OtherUUID = UUID.fromString("6465f8aa-2175-446f-807e-7163994b120f");
|
88 |
|
|
private static final UUID NotMarkedUpUUID = UUID.fromString("796fe3a5-2c9c-4a89-b298-7598ca944063");
|
89 |
|
|
private static final boolean skippQuestion = true;
|
90 |
|
|
|
91 |
285d1929
|
p.kelbert
|
private final NomenclaturalCode nomenclaturalCode;
|
92 |
|
|
private Classification classification;
|
93 |
|
|
|
94 |
|
|
private String treatmentMainName,originalTreatmentName;
|
95 |
|
|
|
96 |
|
|
private final HashMap<String,Map<String,String>> namesMap = new HashMap<String, Map<String,String>>();
|
97 |
|
|
|
98 |
|
|
|
99 |
|
|
private final Pattern keypattern = Pattern.compile("^(\\d+.*|-\\d+.*)");
|
100 |
|
|
private final Pattern keypatternend = Pattern.compile("^.+?\\d$");
|
101 |
|
|
|
102 |
|
|
private boolean maxRankRespected =false;
|
103 |
2b5a7979
|
p.kelbert
|
private Map<String, Feature> featuresMap;
|
104 |
|
|
|
105 |
6b2c2013
|
p.kelbert
|
private MyName currentMyName;
|
106 |
2b5a7979
|
p.kelbert
|
|
107 |
6b2c2013
|
p.kelbert
|
private Reference<?> sourceUrlRef;
|
108 |
2b5a7979
|
p.kelbert
|
|
109 |
|
|
private final TaxonXAddSources sourceHandler = new TaxonXAddSources();
|
110 |
285d1929
|
p.kelbert
|
|
111 |
|
|
/**
|
112 |
|
|
* @param nomenclaturalCode
|
113 |
|
|
* @param classification
|
114 |
|
|
* @param importer
|
115 |
|
|
* @param configState
|
116 |
|
|
*/
|
117 |
|
|
public TaxonXTreatmentExtractor(NomenclaturalCode nomenclaturalCode, Classification classification, TaxonXImport importer,
|
118 |
2b5a7979
|
p.kelbert
|
TaxonXImportState configState,Map<String, Feature> featuresMap, Reference<?> urlSource ) {
|
119 |
285d1929
|
p.kelbert
|
this.nomenclaturalCode=nomenclaturalCode;
|
120 |
|
|
this.classification = classification;
|
121 |
|
|
this.importer=importer;
|
122 |
|
|
this.configState=configState;
|
123 |
2b5a7979
|
p.kelbert
|
this.featuresMap=featuresMap;
|
124 |
|
|
this.sourceUrlRef =urlSource;
|
125 |
285d1929
|
p.kelbert
|
prepareCollectors(configState, importer.getAgentService());
|
126 |
2b5a7979
|
p.kelbert
|
this.sourceHandler.setSourceUrlRef(sourceUrlRef);
|
127 |
|
|
this.sourceHandler.setImporter(importer);
|
128 |
|
|
this.sourceHandler.setConfigState(configState);
|
129 |
285d1929
|
p.kelbert
|
}
|
130 |
|
|
|
131 |
|
|
/**
|
132 |
|
|
* extracts all the treament information and save them
|
133 |
|
|
* @param treatmentnode: the XML Node
|
134 |
|
|
* @param tosave: the list of object to save into the CDM
|
135 |
|
|
* @param refMods: the reference extracted from the MODS
|
136 |
|
|
* @param sourceName: the URI of the document
|
137 |
|
|
*/
|
138 |
|
|
@SuppressWarnings({ "rawtypes", "unused" })
|
139 |
|
|
protected void extractTreatment(Node treatmentnode, List<Object> tosave, Reference<?> refMods, URI sourceName) {
|
140 |
|
|
logger.info("extractTreatment");
|
141 |
|
|
List<TaxonNameBase> nametosave = new ArrayList<TaxonNameBase>();
|
142 |
|
|
NodeList children = treatmentnode.getChildNodes();
|
143 |
|
|
Taxon acceptedTaxon =null;
|
144 |
|
|
Taxon defaultTaxon =null;
|
145 |
|
|
boolean refgroup=false;
|
146 |
|
|
|
147 |
|
|
for (int i=0;i<children.getLength();i++){
|
148 |
|
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:ref_group")) {
|
149 |
|
|
refgroup=true;
|
150 |
|
|
}
|
151 |
|
|
}
|
152 |
|
|
|
153 |
|
|
for (int i=0;i<children.getLength();i++){
|
154 |
136e16b1
|
Andreas Müller
|
|
155 |
285d1929
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:nomenclature")){
|
156 |
|
|
NodeList nomenclature = children.item(i).getChildNodes();
|
157 |
|
|
boolean containsName=false;
|
158 |
|
|
for(int k=0;k<nomenclature.getLength();k++){
|
159 |
|
|
if(nomenclature.item(k).getNodeName().equalsIgnoreCase("tax:name")){
|
160 |
|
|
containsName=true;
|
161 |
|
|
break;
|
162 |
|
|
}
|
163 |
|
|
}
|
164 |
|
|
if (containsName){
|
165 |
|
|
reloadClassification();
|
166 |
|
|
//extract "main" the scientific name
|
167 |
2b5a7979
|
p.kelbert
|
try{
|
168 |
|
|
acceptedTaxon = extractNomenclature(children.item(i),nametosave,refMods);
|
169 |
|
|
}catch(ClassCastException e){e.printStackTrace();System.exit(0);}
|
170 |
|
|
// System.out.println("acceptedTaxon : "+acceptedTaxon);
|
171 |
285d1929
|
p.kelbert
|
}
|
172 |
|
|
}
|
173 |
|
|
else if (children.item(i).getNodeName().equalsIgnoreCase("tax:ref_group") && maxRankRespected){
|
174 |
|
|
reloadClassification();
|
175 |
|
|
//extract the References within the document
|
176 |
|
|
extractReferences(children.item(i),nametosave,acceptedTaxon,refMods);
|
177 |
|
|
}
|
178 |
|
|
else if (children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
179 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("multiple") && maxRankRespected){
|
180 |
|
|
File file = new File("/home/pkelbert/Bureau/multipleTaxonX.txt");
|
181 |
|
|
FileWriter writer;
|
182 |
|
|
try {
|
183 |
|
|
writer = new FileWriter(file ,true);
|
184 |
|
|
writer.write(sourceName+"\n");
|
185 |
|
|
writer.flush();
|
186 |
|
|
writer.close();
|
187 |
|
|
} catch (IOException e1) {
|
188 |
|
|
// TODO Auto-generated catch block
|
189 |
|
|
e1.printStackTrace();
|
190 |
|
|
}
|
191 |
2b5a7979
|
p.kelbert
|
// String multiple = askMultiple(children.item(i));
|
192 |
|
|
String multiple = "Other";
|
193 |
|
|
if (multiple.equalsIgnoreCase("other")) {
|
194 |
|
|
extractSpecificFeatureNotStructured(children.item(i),acceptedTaxon, defaultTaxon,nametosave, refMods,multiple);
|
195 |
285d1929
|
p.kelbert
|
}
|
196 |
|
|
else
|
197 |
2b5a7979
|
p.kelbert
|
if (multiple.equalsIgnoreCase("synonyms")) {
|
198 |
|
|
try{
|
199 |
|
|
extractSynonyms(children.item(i),acceptedTaxon, refMods);
|
200 |
|
|
}catch(NullPointerException e){
|
201 |
|
|
logger.warn("the accepted taxon is maybe null");
|
202 |
|
|
}
|
203 |
285d1929
|
p.kelbert
|
}
|
204 |
|
|
else
|
205 |
2b5a7979
|
p.kelbert
|
if(multiple.equalsIgnoreCase("material examined")){
|
206 |
|
|
extractMaterials(children.item(i),acceptedTaxon, refMods, nametosave);
|
207 |
285d1929
|
p.kelbert
|
}
|
208 |
1de96f38
|
p.kelbert
|
else
|
209 |
2b5a7979
|
p.kelbert
|
if (multiple.equalsIgnoreCase("distribution")){
|
210 |
|
|
extractDistribution(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods);
|
211 |
1de96f38
|
p.kelbert
|
}
|
212 |
|
|
else
|
213 |
2b5a7979
|
p.kelbert
|
if (multiple.equalsIgnoreCase("type status")){
|
214 |
|
|
extractDescriptionWithReference(children.item(i),acceptedTaxon,defaultTaxon,refMods, "TypeStatus");
|
215 |
1de96f38
|
p.kelbert
|
}
|
216 |
2b5a7979
|
p.kelbert
|
else
|
217 |
|
|
if (multiple.equalsIgnoreCase("vernacular name")){
|
218 |
|
|
extractDescriptionWithReference(children.item(i),acceptedTaxon,defaultTaxon,refMods, Feature.COMMON_NAME().getTitleCache());
|
219 |
|
|
|
220 |
|
|
}
|
221 |
|
|
else{
|
222 |
|
|
extractSpecificFeature(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods,multiple);
|
223 |
|
|
}
|
224 |
1de96f38
|
p.kelbert
|
|
225 |
285d1929
|
p.kelbert
|
}
|
226 |
|
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
227 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("biology_ecology") && maxRankRespected){
|
228 |
|
|
extractFeature(children.item(i),acceptedTaxon,defaultTaxon, nametosave, refMods, Feature.BIOLOGY_ECOLOGY());
|
229 |
|
|
}
|
230 |
1de96f38
|
p.kelbert
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
231 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("vernacularName") && maxRankRespected){
|
232 |
2b5a7979
|
p.kelbert
|
extractDescriptionWithReference(children.item(i),acceptedTaxon,defaultTaxon,refMods, Feature.COMMON_NAME().getTitleCache());
|
233 |
1de96f38
|
p.kelbert
|
}
|
234 |
285d1929
|
p.kelbert
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
235 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("description") && maxRankRespected){
|
236 |
|
|
extractFeature(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods, Feature.DESCRIPTION());
|
237 |
|
|
}
|
238 |
|
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
239 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("diagnosis") && maxRankRespected){
|
240 |
|
|
extractFeature(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods,Feature.DIAGNOSIS());
|
241 |
|
|
}
|
242 |
|
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
243 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("discussion") && maxRankRespected){
|
244 |
|
|
extractFeature(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods, Feature.DISCUSSION());
|
245 |
|
|
}
|
246 |
1de96f38
|
p.kelbert
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
247 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("note") && maxRankRespected){
|
248 |
|
|
extractFeature(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods, Feature.DESCRIPTION());
|
249 |
|
|
}
|
250 |
285d1929
|
p.kelbert
|
|
251 |
|
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
252 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("distribution") && maxRankRespected){
|
253 |
136e16b1
|
Andreas Müller
|
extractDistribution(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods);
|
254 |
285d1929
|
p.kelbert
|
}
|
255 |
|
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
256 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("etymology") && maxRankRespected){
|
257 |
|
|
extractFeature(children.item(i),acceptedTaxon,defaultTaxon,nametosave,refMods,Feature.ETYMOLOGY());
|
258 |
|
|
}
|
259 |
|
|
|
260 |
|
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
261 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("materials_examined") && maxRankRespected){
|
262 |
136e16b1
|
Andreas Müller
|
extractMaterials(children.item(i),acceptedTaxon, refMods, nametosave);
|
263 |
285d1929
|
p.kelbert
|
}
|
264 |
1de96f38
|
p.kelbert
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:figure") && maxRankRespected){
|
265 |
2b5a7979
|
p.kelbert
|
extractSpecificFeature(children.item(i),acceptedTaxon,defaultTaxon, nametosave, refMods, "Figure");
|
266 |
1de96f38
|
p.kelbert
|
}
|
267 |
|
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
268 |
2b5a7979
|
p.kelbert
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("Other") && maxRankRespected){
|
269 |
1de96f38
|
p.kelbert
|
extractSpecificFeature(children.item(i),acceptedTaxon,defaultTaxon, nametosave, refMods, "table");
|
270 |
|
|
}
|
271 |
285d1929
|
p.kelbert
|
|
272 |
|
|
else if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
273 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("key") && maxRankRespected){
|
274 |
|
|
//TODO IGNORE keys for the moment
|
275 |
|
|
//extractKey(children.item(i),acceptedTaxon, nametosave,source, refMods);
|
276 |
2b5a7979
|
p.kelbert
|
extractSpecificFeatureNotStructured(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods,"Keys - unparsed");
|
277 |
285d1929
|
p.kelbert
|
}
|
278 |
|
|
else{
|
279 |
1de96f38
|
p.kelbert
|
if (!children.item(i).getNodeName().equalsIgnoreCase("tax:pb")){
|
280 |
6c41cae2
|
p.kelbert
|
//logger.info("ANOTHER KIND OF NODES: "+children.item(i).getNodeName()+", "+children.item(i).getAttributes());
|
281 |
1de96f38
|
p.kelbert
|
if (children.item(i).getAttributes() !=null) {
|
282 |
6c41cae2
|
p.kelbert
|
//logger.info(children.item(i).getAttributes().item(0));
|
283 |
1de96f38
|
p.kelbert
|
}
|
284 |
2b5a7979
|
p.kelbert
|
extractSpecificFeatureNotStructured(children.item(i),acceptedTaxon,defaultTaxon,nametosave, refMods,notMarkedUp);
|
285 |
285d1929
|
p.kelbert
|
}
|
286 |
|
|
}
|
287 |
|
|
}
|
288 |
|
|
// logger.info("saveUpdateNames");
|
289 |
|
|
if (maxRankRespected){
|
290 |
|
|
importer.getNameService().saveOrUpdate(nametosave);
|
291 |
|
|
importer.getClassificationService().saveOrUpdate(classification);
|
292 |
6c41cae2
|
p.kelbert
|
//logger.info("saveUpdateNames-ok");
|
293 |
285d1929
|
p.kelbert
|
}
|
294 |
2b5a7979
|
p.kelbert
|
|
295 |
|
|
buildFeatureTree();
|
296 |
|
|
}
|
297 |
|
|
|
298 |
|
|
|
299 |
|
|
protected Map<String,Feature> getFeaturesUsed(){
|
300 |
|
|
return featuresMap;
|
301 |
|
|
}
|
302 |
|
|
/**
|
303 |
|
|
*
|
304 |
|
|
*/
|
305 |
|
|
private void buildFeatureTree() {
|
306 |
6c41cae2
|
p.kelbert
|
logger.info("buildFeatureTree");
|
307 |
2b5a7979
|
p.kelbert
|
FeatureTree proibiospheretree = importer.getFeatureTreeService().find(proIbioTreeUUID);
|
308 |
|
|
if (proibiospheretree == null){
|
309 |
|
|
List<FeatureTree> trees = importer.getFeatureTreeService().list(FeatureTree.class, null, null, null, null);
|
310 |
|
|
if (trees.size()==1) {
|
311 |
|
|
FeatureTree ft = trees.get(0);
|
312 |
|
|
if (featuresMap==null) {
|
313 |
|
|
featuresMap=new HashMap<String, Feature>();
|
314 |
|
|
}
|
315 |
|
|
for (Feature feature: ft.getDistinctFeatures()){
|
316 |
|
|
if(feature!=null) {
|
317 |
|
|
featuresMap.put(feature.getTitleCache(), feature);
|
318 |
|
|
}
|
319 |
|
|
}
|
320 |
|
|
}
|
321 |
|
|
proibiospheretree = FeatureTree.NewInstance();
|
322 |
|
|
proibiospheretree.setUuid(proIbioTreeUUID);
|
323 |
|
|
}
|
324 |
|
|
// FeatureNode root = proibiospheretree.getRoot();
|
325 |
|
|
FeatureNode root2 = proibiospheretree.getRoot();
|
326 |
|
|
if (root2 != null){
|
327 |
|
|
int nbChildren = root2.getChildCount()-1;
|
328 |
|
|
while (nbChildren>-1){
|
329 |
|
|
try{
|
330 |
|
|
root2.removeChild(nbChildren);
|
331 |
|
|
}catch(Exception e){logger.warn("Can't remove child from FeatureTree "+e);}
|
332 |
|
|
nbChildren --;
|
333 |
|
|
}
|
334 |
|
|
|
335 |
|
|
}
|
336 |
|
|
|
337 |
|
|
for (Feature feature:featuresMap.values()) {
|
338 |
|
|
root2.addChild(FeatureNode.NewInstance(feature));
|
339 |
|
|
}
|
340 |
|
|
importer.getFeatureTreeService().saveOrUpdate(proibiospheretree);
|
341 |
|
|
|
342 |
285d1929
|
p.kelbert
|
}
|
343 |
|
|
|
344 |
|
|
|
345 |
|
|
/**
|
346 |
|
|
* @param keys
|
347 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
348 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
349 |
|
|
* @param refMods: the current reference extracted from the MODS
|
350 |
|
|
*/
|
351 |
2b5a7979
|
p.kelbert
|
/* @SuppressWarnings("rawtypes")
|
352 |
285d1929
|
p.kelbert
|
private void extractKey(Node keys, Taxon acceptedTaxon,List<TaxonNameBase> nametosave, Reference<?> refMods) {
|
353 |
|
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
354 |
|
|
|
355 |
|
|
NodeList children = keys.getChildNodes();
|
356 |
|
|
String key="";
|
357 |
|
|
PolytomousKey poly = PolytomousKey.NewInstance();
|
358 |
59c2c696
|
Andreas Müller
|
poly.addSource(OriginalSourceType.Import, null,null,refMods,null);
|
359 |
2b5a7979
|
p.kelbert
|
poly.addSource(OriginalSourceType.Import, null,null,sourceUrlRef,null);
|
360 |
285d1929
|
p.kelbert
|
poly.addTaxonomicScope(acceptedTaxon);
|
361 |
d03aed8f
|
Andreas Müller
|
poly.setTitleCache("bloup", true);
|
362 |
285d1929
|
p.kelbert
|
// poly.addCoveredTaxon(acceptedTaxon);
|
363 |
|
|
PolytomousKeyNode root = poly.getRoot();
|
364 |
|
|
PolytomousKeyNode previous = null,tmpKey=null;
|
365 |
|
|
Taxon taxonKey=null;
|
366 |
|
|
List<PolytomousKeyNode> polyNodes = new ArrayList<PolytomousKeyNode>();
|
367 |
|
|
|
368 |
|
|
// String fullContent = keys.getTextContent();
|
369 |
|
|
for (int i=0;i<children.getLength();i++){
|
370 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
371 |
|
|
NodeList paragraph = children.item(i).getChildNodes();
|
372 |
|
|
key="";
|
373 |
|
|
taxonKey=null;
|
374 |
|
|
for (int j=0;j<paragraph.getLength();j++){
|
375 |
|
|
if (paragraph.item(j).getNodeName().equalsIgnoreCase("#text")){
|
376 |
|
|
if (! paragraph.item(j).getTextContent().trim().isEmpty()){
|
377 |
|
|
key+=paragraph.item(j).getTextContent().trim();
|
378 |
|
|
// logger.info("KEY: "+j+"--"+key);
|
379 |
|
|
}
|
380 |
|
|
}
|
381 |
|
|
if(paragraph.item(j).getNodeName().equalsIgnoreCase("tax:name")){
|
382 |
|
|
taxonKey=getTaxonFromXML(paragraph.item(j),nametosave,refMods);
|
383 |
|
|
}
|
384 |
|
|
}
|
385 |
|
|
// logger.info("keypattern.matcher(key).matches(): "+keypattern.matcher(key).matches());
|
386 |
|
|
if (keypattern.matcher(key).matches()){
|
387 |
|
|
tmpKey = PolytomousKeyNode.NewInstance(key);
|
388 |
|
|
if (taxonKey!=null) {
|
389 |
|
|
tmpKey.setTaxon(taxonKey);
|
390 |
|
|
}
|
391 |
|
|
polyNodes.add(tmpKey);
|
392 |
|
|
if (previous == null) {
|
393 |
|
|
root.addChild(tmpKey);
|
394 |
|
|
} else {
|
395 |
|
|
previous.addChild(tmpKey);
|
396 |
|
|
}
|
397 |
|
|
}else{
|
398 |
|
|
if (!key.isEmpty()){
|
399 |
|
|
tmpKey=PolytomousKeyNode.NewInstance(key);
|
400 |
|
|
if (taxonKey!=null) {
|
401 |
|
|
tmpKey.setTaxon(taxonKey);
|
402 |
|
|
}
|
403 |
|
|
polyNodes.add(tmpKey);
|
404 |
|
|
if (keypatternend.matcher(key).matches()) {
|
405 |
|
|
root.addChild(tmpKey);
|
406 |
|
|
previous=tmpKey;
|
407 |
|
|
} else{
|
408 |
|
|
previous.addChild(tmpKey);
|
409 |
|
|
}
|
410 |
|
|
|
411 |
|
|
}
|
412 |
|
|
}
|
413 |
|
|
}
|
414 |
|
|
}
|
415 |
|
|
importer.getPolytomousKeyNodeService().saveOrUpdate(polyNodes);
|
416 |
|
|
importer.getPolytomousKeyService().saveOrUpdate(poly);
|
417 |
|
|
}
|
418 |
|
|
*/
|
419 |
2b5a7979
|
p.kelbert
|
// /**
|
420 |
|
|
// * @param taxons: the XML Nodegroup
|
421 |
|
|
// * @param nametosave: the list of objects to save into the CDM
|
422 |
|
|
// * @param acceptedTaxon: the current accepted Taxon
|
423 |
|
|
// * @param refMods: the current reference extracted from the MODS
|
424 |
|
|
// *
|
425 |
|
|
// * @return Taxon object built
|
426 |
|
|
// */
|
427 |
|
|
// @SuppressWarnings({ "rawtypes", "unchecked" })
|
428 |
|
|
// private Taxon getTaxonFromXML(Node taxons, List<TaxonNameBase> nametosave, Reference<?> refMods) {
|
429 |
|
|
// // logger.info("getTaxonFromXML");
|
430 |
|
|
// // logger.info("acceptedTaxon: "+acceptedTaxon);
|
431 |
|
|
//
|
432 |
|
|
// // TaxonNameBase nameToBeFilled = null;
|
433 |
|
|
//
|
434 |
|
|
// currentMyName = new MyName();
|
435 |
|
|
// NomenclaturalStatusType statusType = null;
|
436 |
|
|
//
|
437 |
|
|
// try {
|
438 |
|
|
// currentMyName = extractScientificName(taxons);
|
439 |
|
|
// if (!currentMyName.getStatus().isEmpty()){
|
440 |
|
|
// try {
|
441 |
|
|
// statusType = nomStatusString2NomStatus(currentMyName.getStatus());
|
442 |
|
|
// } catch (UnknownCdmTypeException e) {
|
443 |
|
|
// addProblematicStatusToFile(currentMyName.getStatus());
|
444 |
|
|
// logger.warn("Problem with status");
|
445 |
|
|
// }
|
446 |
|
|
// }
|
447 |
|
|
//
|
448 |
|
|
// } catch (TransformerFactoryConfigurationError e1) {
|
449 |
|
|
// logger.warn(e1);
|
450 |
|
|
// } catch (TransformerException e1) {
|
451 |
|
|
// logger.warn(e1);
|
452 |
|
|
// }
|
453 |
|
|
// /* INonViralNameParser parser = NonViralNameParserImpl.NewInstance();
|
454 |
|
|
//
|
455 |
|
|
// nameToBeFilled = parser.parseFullName(currentMyName.getName(), nomenclaturalCode, currentMyName.getRank());
|
456 |
|
|
// if (nameToBeFilled.hasProblem() &&
|
457 |
|
|
// !((nameToBeFilled.getParsingProblems().size()==1) && nameToBeFilled.getParsingProblems().contains(ParserProblem.CheckRank)) ) {
|
458 |
|
|
// // if (nameToBeFilled.hasProblem() && nameToBeFilled.getParsingProblems().contains(ParserProblem.UnparsableNamePart)){
|
459 |
|
|
// addProblemNameToFile(currentMyName.getName(),"",nomenclaturalCode,currentMyName.getRank());
|
460 |
|
|
// nameToBeFilled=solveNameProblem(currentMyName.getOriginalName(), currentMyName.getName(),parser, currentMyName.getAuthor(), currentMyName.getRank());
|
461 |
|
|
// }
|
462 |
|
|
//
|
463 |
|
|
// nameToBeFilled = getTaxonNameBase(nameToBeFilled,nametosave,statusType);
|
464 |
|
|
// */
|
465 |
|
|
// TaxonNameBase nameToBeFilled = currentMyName.getTaxonNameBase();
|
466 |
|
|
// Taxon t = currentMyName.getTaxon();
|
467 |
|
|
// // importer.getNameService().saveOrUpdate(nametosave);
|
468 |
|
|
// /* Taxon t = importer.getTaxonService().findBestMatchingTaxon(nameToBeFilled.getTitleCache());
|
469 |
|
|
// */
|
470 |
|
|
// boolean statusMatch=false;
|
471 |
|
|
// if(t !=null ){
|
472 |
|
|
// statusMatch=compareStatus(t, statusType);
|
473 |
|
|
// }
|
474 |
|
|
// if (t ==null || (t != null && !statusMatch)){
|
475 |
|
|
// if(statusType != null) {
|
476 |
|
|
// nameToBeFilled.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
477 |
|
|
// }
|
478 |
|
|
// t= new Taxon(nameToBeFilled,(Reference<?>) nameToBeFilled.getNomenclaturalReference() );//TODO TOFIX reference
|
479 |
|
|
// if (t.getSec() == null) {
|
480 |
|
|
// t.setSec(refMods);
|
481 |
|
|
// }
|
482 |
|
|
// if(!configState.getConfig().doKeepOriginalSecundum()) {
|
483 |
|
|
// t.setSec(configState.getConfig().getSecundum());
|
484 |
|
|
// logger.info("SET SECUNDUM "+configState.getConfig().getSecundum());
|
485 |
|
|
// }
|
486 |
|
|
// t.addSource(OriginalSourceType.Import,null,null,refMods,null);
|
487 |
|
|
// t.addSource(OriginalSourceType.Import, null,null,sourceUrlRef,null);
|
488 |
|
|
//
|
489 |
|
|
//
|
490 |
|
|
// if (!currentMyName.getIdentifier().isEmpty() && (currentMyName.getIdentifier().length()>2)){
|
491 |
|
|
// setLSID(currentMyName.getIdentifier(), t);
|
492 |
|
|
// }
|
493 |
|
|
//
|
494 |
|
|
// // Taxon parentTaxon = currentMyName.getHigherTaxa();
|
495 |
|
|
// // if (parentTaxon == null && !skippQuestion) {
|
496 |
|
|
// // parentTaxon = askParent(t, classification);
|
497 |
|
|
// // }
|
498 |
|
|
// // if (parentTaxon ==null){
|
499 |
|
|
// // while (parentTaxon == null) {
|
500 |
|
|
// // System.out.println("parent is null");
|
501 |
|
|
// // parentTaxon = createParent(t, refMods);
|
502 |
|
|
// // classification.addParentChild(parentTaxon, t, refMods, null);
|
503 |
|
|
// // }
|
504 |
|
|
// // }else{
|
505 |
|
|
// // classification.addParentChild(parentTaxon, t, refMods, null);
|
506 |
|
|
// // }
|
507 |
|
|
// }
|
508 |
|
|
// else{
|
509 |
|
|
// t = CdmBase.deproxy(t, Taxon.class);
|
510 |
|
|
// }
|
511 |
|
|
// if (!configState.getConfig().doKeepOriginalSecundum()) {
|
512 |
|
|
// t.setSec(configState.getConfig().getSecundum());
|
513 |
|
|
// logger.info("SET SECUNDUM "+configState.getConfig().getSecundum());
|
514 |
|
|
// }
|
515 |
|
|
// return t;
|
516 |
|
|
// }
|
517 |
|
|
|
518 |
|
|
|
519 |
|
|
|
520 |
|
|
|
521 |
|
|
// private Taxon getTaxonFromTaxonNameBase(TaxonNameBase tnb,Reference<?> ref){
|
522 |
|
|
// Taxon taxon = null;
|
523 |
|
|
//// System.out.println(tnb.getTitleCache());
|
524 |
|
|
// Taxon cc= importer.getTaxonService().findBestMatchingTaxon(tnb.getTitleCache());
|
525 |
|
|
// if (cc != null){
|
526 |
|
|
// if ((cc.getSec() == null || cc.getSec().toString().isEmpty()) || (cc.getSec() != null &&
|
527 |
|
|
// cc.getSec().getTitleCache().equalsIgnoreCase(ref.getTitleCache()))) {
|
528 |
|
|
// if(cc.getSec() == null || cc.getSec().toString().isEmpty()){
|
529 |
|
|
// cc.setSec(ref);
|
530 |
|
|
// importer.getTaxonService().saveOrUpdate(cc);
|
531 |
|
|
// }
|
532 |
|
|
// taxon=cc;
|
533 |
|
|
// }
|
534 |
|
|
// }
|
535 |
|
|
// else{
|
536 |
|
|
// // List<TaxonBase> c = importer.getTaxonService().searchTaxaByName(tnb.getTitleCache(), ref);
|
537 |
|
|
// List<TaxonBase> c = importer.getTaxonService().list(TaxonBase.class, 0, 0, null, null);
|
538 |
|
|
// for (TaxonBase b : c) {
|
539 |
|
|
// try{
|
540 |
|
|
// taxon = (Taxon) b;
|
541 |
|
|
// }catch(ClassCastException e){logger.warn("error while casting existing taxonnamebase");}
|
542 |
|
|
// }
|
543 |
|
|
// }
|
544 |
|
|
// if (taxon == null){
|
545 |
|
|
//// System.out.println("NEW TAXON HERE "+tnb.toString()+", "+ref.toString());
|
546 |
|
|
// taxon = Taxon.NewInstance(tnb, ref); //sec set null
|
547 |
|
|
// importer.getTaxonService().save(taxon);
|
548 |
|
|
//
|
549 |
|
|
// }
|
550 |
|
|
// taxon = (Taxon) importer.getTaxonService().find(taxon.getUuid());
|
551 |
|
|
//
|
552 |
|
|
// boolean exist = false;
|
553 |
|
|
// for (TaxonNode p : classification.getAllNodes()){
|
554 |
|
|
// if(p.getTaxon().equals(taxon)) {
|
555 |
|
|
// exist =true;
|
556 |
|
|
// }
|
557 |
|
|
// }
|
558 |
|
|
// if (!exist){
|
559 |
|
|
// taxon = (Taxon) importer.getTaxonService().find(taxon.getUuid());
|
560 |
|
|
// Taxon parentTaxon = currentMyName.getHigherTaxa();
|
561 |
|
|
// if (parentTaxon != null) {
|
562 |
|
|
// classification.addParentChild(parentTaxon, taxon, ref, null);
|
563 |
|
|
// } else {
|
564 |
|
|
// System.out.println("HERE???");
|
565 |
|
|
// classification.addChildTaxon(taxon, ref, null);
|
566 |
|
|
// }
|
567 |
|
|
// importer.getClassificationService().saveOrUpdate(classification);
|
568 |
|
|
// // refreshTransaction();
|
569 |
|
|
// }
|
570 |
|
|
// taxon = CdmBase.deproxy(taxon, Taxon.class);
|
571 |
|
|
// // System.out.println("TAXON RETOURNE : "+taxon.getTitleCache());
|
572 |
|
|
// return taxon;
|
573 |
|
|
// }
|
574 |
136e16b1
|
Andreas Müller
|
/**
|
575 |
|
|
* @param taxons: the XML Nodegroup
|
576 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
577 |
|
|
* @param acceptedTaxon: the current accepted Taxon
|
578 |
|
|
* @param refMods: the current reference extracted from the MODS
|
579 |
|
|
*
|
580 |
|
|
* @return Taxon object built
|
581 |
|
|
*/
|
582 |
2b5a7979
|
p.kelbert
|
@SuppressWarnings({ "rawtypes", "unused" })
|
583 |
6b2c2013
|
p.kelbert
|
private TaxonNameBase getTaxonNameBaseFromXML(Node taxons, List<TaxonNameBase> nametosave, Reference<?> refMods, boolean isSynonym) {
|
584 |
136e16b1
|
Andreas Müller
|
// logger.info("getTaxonFromXML");
|
585 |
|
|
// logger.info("acceptedTaxon: "+acceptedTaxon);
|
586 |
6c41cae2
|
p.kelbert
|
logger.info("getTaxonNameBaseFromXML");
|
587 |
136e16b1
|
Andreas Müller
|
TaxonNameBase nameToBeFilled = null;
|
588 |
|
|
|
589 |
6b2c2013
|
p.kelbert
|
currentMyName=new MyName(isSynonym);
|
590 |
136e16b1
|
Andreas Müller
|
|
591 |
1de96f38
|
p.kelbert
|
NomenclaturalStatusType statusType = null;
|
592 |
136e16b1
|
Andreas Müller
|
try {
|
593 |
2b5a7979
|
p.kelbert
|
currentMyName = extractScientificName(taxons,refMods);
|
594 |
136e16b1
|
Andreas Müller
|
} catch (TransformerFactoryConfigurationError e1) {
|
595 |
|
|
logger.warn(e1);
|
596 |
|
|
} catch (TransformerException e1) {
|
597 |
|
|
logger.warn(e1);
|
598 |
|
|
}
|
599 |
2b5a7979
|
p.kelbert
|
/* INonViralNameParser parser = NonViralNameParserImpl.NewInstance();
|
600 |
136e16b1
|
Andreas Müller
|
|
601 |
2b5a7979
|
p.kelbert
|
nameToBeFilled = parser.parseFullName(currentMyName.getName(), nomenclaturalCode, currentMyName.getRank());
|
602 |
136e16b1
|
Andreas Müller
|
if (nameToBeFilled.hasProblem() &&
|
603 |
|
|
!((nameToBeFilled.getParsingProblems().size()==1) && nameToBeFilled.getParsingProblems().contains(ParserProblem.CheckRank)) ) {
|
604 |
|
|
// if (nameToBeFilled.hasProblem() && nameToBeFilled.getParsingProblems().contains(ParserProblem.UnparsableNamePart)){
|
605 |
2b5a7979
|
p.kelbert
|
addProblemNameToFile(currentMyName.getName(),"",nomenclaturalCode,currentMyName.getRank());
|
606 |
|
|
nameToBeFilled=solveNameProblem(currentMyName.getOriginalName(), currentMyName.getName(),parser,currentMyName.getAuthor(), currentMyName.getRank());
|
607 |
136e16b1
|
Andreas Müller
|
}
|
608 |
|
|
|
609 |
1de96f38
|
p.kelbert
|
nameToBeFilled = getTaxonNameBase(nameToBeFilled,nametosave,statusType);
|
610 |
2b5a7979
|
p.kelbert
|
*/
|
611 |
|
|
nameToBeFilled = currentMyName.getTaxonNameBase();
|
612 |
136e16b1
|
Andreas Müller
|
return nameToBeFilled;
|
613 |
|
|
|
614 |
|
|
}
|
615 |
|
|
|
616 |
2b5a7979
|
p.kelbert
|
// @SuppressWarnings("rawtypes")
|
617 |
|
|
// private TaxonNameBase getTaxonNameBase (TaxonNameBase name, List<TaxonNameBase> nametosave, NomenclaturalStatusType statusType){
|
618 |
|
|
// List<TaxonNameBase> names = importer.getNameService().list(TaxonNameBase.class, null, null, null, null);
|
619 |
|
|
// for (TaxonNameBase tb : names){
|
620 |
|
|
// if (tb.getTitleCache().equalsIgnoreCase(name.getTitleCache())) {
|
621 |
|
|
// boolean statusMatch=false;
|
622 |
|
|
// if(tb !=null ){
|
623 |
|
|
// statusMatch=compareStatus(tb, statusType);
|
624 |
|
|
// }
|
625 |
|
|
// if (!statusMatch){
|
626 |
|
|
// if(statusType != null) {
|
627 |
|
|
// name.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
628 |
|
|
// }
|
629 |
|
|
// }else
|
630 |
|
|
// {
|
631 |
|
|
// logger.info("TaxonNameBase FOUND"+name.getTitleCache());
|
632 |
|
|
// return CdmBase.deproxy(tb, TaxonNameBase.class);
|
633 |
|
|
// }
|
634 |
|
|
// }
|
635 |
|
|
// }
|
636 |
|
|
// // logger.info("TaxonNameBase NOT FOUND "+name.getTitleCache());
|
637 |
|
|
// // System.out.println("add name "+name);
|
638 |
|
|
// nametosave.add(name);
|
639 |
|
|
// name = CdmBase.deproxy(name, TaxonNameBase.class);
|
640 |
|
|
// return name;
|
641 |
|
|
//
|
642 |
|
|
// }
|
643 |
|
|
|
644 |
|
|
|
645 |
|
|
|
646 |
|
|
// /**
|
647 |
|
|
// * @param tb
|
648 |
|
|
// * @param statusType
|
649 |
|
|
// * @return
|
650 |
|
|
// */
|
651 |
|
|
// private boolean compareStatus(TaxonNameBase tb, NomenclaturalStatusType statusType) {
|
652 |
|
|
// boolean statusMatch=false;
|
653 |
|
|
// //found one taxon
|
654 |
|
|
// Set<NomenclaturalStatus> status = tb.getStatus();
|
655 |
|
|
// if (statusType!=null && status.size()>0){ //the statusType is known for both taxon
|
656 |
|
|
// for (NomenclaturalStatus st:status){
|
657 |
|
|
// NomenclaturalStatusType stype = st.getType();
|
658 |
|
|
// if (stype.toString().equalsIgnoreCase(statusType.toString())) {
|
659 |
|
|
// statusMatch=true;
|
660 |
|
|
// }
|
661 |
|
|
// }
|
662 |
|
|
// }
|
663 |
|
|
// else{
|
664 |
|
|
// if(statusType == null && status.size()==0) {//there is no statusType, we can assume it's the same
|
665 |
|
|
// statusMatch=true;
|
666 |
|
|
// }
|
667 |
|
|
// }
|
668 |
|
|
// return statusMatch;
|
669 |
|
|
// }
|
670 |
1de96f38
|
p.kelbert
|
|
671 |
285d1929
|
p.kelbert
|
/**
|
672 |
|
|
*
|
673 |
|
|
*/
|
674 |
|
|
private void reloadClassification() {
|
675 |
6c41cae2
|
p.kelbert
|
logger.info("reloadClassification");
|
676 |
285d1929
|
p.kelbert
|
Classification cl = importer.getClassificationService().find(classification.getUuid());
|
677 |
|
|
if (cl != null){
|
678 |
|
|
classification=cl;
|
679 |
|
|
}else{
|
680 |
|
|
importer.getClassificationService().saveOrUpdate(classification);
|
681 |
|
|
classification = importer.getClassificationService().find(classification.getUuid());
|
682 |
|
|
}
|
683 |
|
|
}
|
684 |
|
|
|
685 |
2b5a7979
|
p.kelbert
|
// /**
|
686 |
|
|
// * Create a Taxon for the current NameBase, based on the current reference
|
687 |
|
|
// * @param taxonNameBase
|
688 |
|
|
// * @param refMods: the current reference extracted from the MODS
|
689 |
|
|
// * @return Taxon
|
690 |
|
|
// */
|
691 |
|
|
// @SuppressWarnings({ "unused", "rawtypes" })
|
692 |
|
|
// private Taxon getTaxon(TaxonNameBase taxonNameBase, Reference<?> refMods) {
|
693 |
|
|
// Taxon t = new Taxon(taxonNameBase,null );
|
694 |
|
|
// if (!configState.getConfig().doKeepOriginalSecundum() || (t.getSec() == null)) {
|
695 |
|
|
// t.setSec(configState.getConfig().getSecundum());
|
696 |
|
|
// logger.info("SET SECUNDUM "+configState.getConfig().getSecundum());
|
697 |
|
|
// }
|
698 |
|
|
// /*<<<<<<< .courant
|
699 |
|
|
// boolean sourceExists=false;
|
700 |
|
|
// Set<IdentifiableSource> sources = t.getSources();
|
701 |
|
|
// for (IdentifiableSource src : sources){
|
702 |
|
|
// String micro = src.getCitationMicroReference();
|
703 |
|
|
// Reference r = src.getCitation();
|
704 |
|
|
// if (r.equals(refMods) && micro == null) {
|
705 |
|
|
// sourceExists=true;
|
706 |
|
|
// }
|
707 |
|
|
// }
|
708 |
|
|
// if(!sourceExists) {
|
709 |
|
|
// t.addSource(null,null,refMods,null);
|
710 |
|
|
// }
|
711 |
|
|
//=======*/
|
712 |
|
|
// t.addSource(OriginalSourceType.Import,null,null,refMods,null);
|
713 |
|
|
// t.addSource(OriginalSourceType.Import, null,null,sourceUrlRef,null);
|
714 |
|
|
// return t;
|
715 |
|
|
// }
|
716 |
|
|
|
717 |
|
|
private void extractDescriptionWithReference(Node typestatus, Taxon acceptedTaxon, Taxon defaultTaxon, Reference<?> refMods,
|
718 |
|
|
String featureName) {
|
719 |
|
|
// System.out.println("extractDescriptionWithReference !");
|
720 |
6c41cae2
|
p.kelbert
|
logger.info("extractDescriptionWithReference");
|
721 |
1de96f38
|
p.kelbert
|
NodeList children = typestatus.getChildNodes();
|
722 |
|
|
|
723 |
2b5a7979
|
p.kelbert
|
Feature currentFeature=getFeatureObjectFromString(featureName);
|
724 |
1de96f38
|
p.kelbert
|
|
725 |
|
|
String r="";String s="";
|
726 |
|
|
for (int i=0;i<children.getLength();i++){
|
727 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
728 |
|
|
s+=children.item(i).getTextContent().trim();
|
729 |
|
|
}
|
730 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:bibref")){
|
731 |
|
|
r+= children.item(i).getTextContent().trim();
|
732 |
|
|
}
|
733 |
|
|
if (s.indexOf(r)>-1) {
|
734 |
|
|
s=s.split(r)[0];
|
735 |
|
|
}
|
736 |
|
|
}
|
737 |
2b5a7979
|
p.kelbert
|
|
738 |
1de96f38
|
p.kelbert
|
Reference<?> currentref = ReferenceFactory.newGeneric();
|
739 |
|
|
if(!r.isEmpty()) {
|
740 |
6e861bce
|
Andreas Müller
|
currentref.setTitleCache(r, true);
|
741 |
1de96f38
|
p.kelbert
|
} else {
|
742 |
|
|
currentref=refMods;
|
743 |
|
|
}
|
744 |
|
|
setParticularDescription(s,acceptedTaxon,defaultTaxon, currentref, refMods,currentFeature);
|
745 |
|
|
}
|
746 |
|
|
|
747 |
285d1929
|
p.kelbert
|
/**
|
748 |
136e16b1
|
Andreas Müller
|
* @param nametosave
|
749 |
285d1929
|
p.kelbert
|
* @param distribution: the XML node group
|
750 |
|
|
* @param acceptedTaxon: the current accepted Taxon
|
751 |
|
|
* @param defaultTaxon: the current defaultTaxon, only used if there is no accepted name
|
752 |
|
|
* @param refMods: the current reference extracted from the MODS
|
753 |
|
|
*/
|
754 |
|
|
@SuppressWarnings("rawtypes")
|
755 |
136e16b1
|
Andreas Müller
|
private void extractDistribution(Node distribution, Taxon acceptedTaxon, Taxon defaultTaxon, List<TaxonNameBase> nametosave, Reference<?> refMods) {
|
756 |
6c41cae2
|
p.kelbert
|
logger.info("extractDistribution");
|
757 |
285d1929
|
p.kelbert
|
// logger.info("acceptedTaxon: "+acceptedTaxon);
|
758 |
|
|
NodeList children = distribution.getChildNodes();
|
759 |
|
|
Map<Integer,List<MySpecimenOrObservation>> specimenOrObservations = new HashMap<Integer, List<MySpecimenOrObservation>>();
|
760 |
|
|
Map<Integer,String> descriptionsFulltext = new HashMap<Integer,String>();
|
761 |
|
|
|
762 |
|
|
for (int i=0;i<children.getLength();i++){
|
763 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
764 |
|
|
NodeList paragraph = children.item(i).getChildNodes();
|
765 |
|
|
for (int j=0;j<paragraph.getLength();j++){
|
766 |
|
|
if (paragraph.item(j).getNodeName().equalsIgnoreCase("#text")){
|
767 |
2b5a7979
|
p.kelbert
|
extractText(descriptionsFulltext, i, paragraph.item(j));
|
768 |
285d1929
|
p.kelbert
|
}
|
769 |
136e16b1
|
Andreas Müller
|
else if (paragraph.item(j).getNodeName().equalsIgnoreCase("tax:name")){
|
770 |
2b5a7979
|
p.kelbert
|
extractInLine(nametosave, refMods, descriptionsFulltext, i,paragraph.item(j));
|
771 |
136e16b1
|
Andreas Müller
|
}
|
772 |
285d1929
|
p.kelbert
|
else if (paragraph.item(j).getNodeName().equalsIgnoreCase("tax:collection_event")){
|
773 |
|
|
MySpecimenOrObservation specimenOrObservation = new MySpecimenOrObservation();
|
774 |
59c2c696
|
Andreas Müller
|
DerivedUnit derivedUnitBase = null;
|
775 |
136e16b1
|
Andreas Müller
|
specimenOrObservation = extractSpecimenOrObservation(paragraph.item(j), derivedUnitBase, SpecimenOrObservationType.DerivedUnit);
|
776 |
2b5a7979
|
p.kelbert
|
extractTextFromSpecimenOrObservation(specimenOrObservations, descriptionsFulltext, i, specimenOrObservation);
|
777 |
136e16b1
|
Andreas Müller
|
}
|
778 |
285d1929
|
p.kelbert
|
}
|
779 |
|
|
}
|
780 |
136e16b1
|
Andreas Müller
|
}
|
781 |
|
|
|
782 |
|
|
int m=0;
|
783 |
|
|
for (int k:descriptionsFulltext.keySet()) {
|
784 |
|
|
if (k>m) {
|
785 |
|
|
m=k;
|
786 |
|
|
}
|
787 |
|
|
}
|
788 |
|
|
for (int k:specimenOrObservations.keySet()) {
|
789 |
|
|
if (k>m) {
|
790 |
|
|
m=k;
|
791 |
285d1929
|
p.kelbert
|
}
|
792 |
136e16b1
|
Andreas Müller
|
}
|
793 |
285d1929
|
p.kelbert
|
|
794 |
|
|
|
795 |
2b5a7979
|
p.kelbert
|
if(acceptedTaxon!=null){
|
796 |
|
|
TaxonDescription td =importer.getTaxonDescription(acceptedTaxon, false, true);
|
797 |
|
|
Feature currentFeature = Feature.DISTRIBUTION();
|
798 |
|
|
// DerivedUnit derivedUnitBase=null;
|
799 |
|
|
// String descr="";
|
800 |
|
|
for (int k=0;k<=m;k++){
|
801 |
|
|
if(specimenOrObservations.keySet().contains(k)){
|
802 |
|
|
for (MySpecimenOrObservation soo:specimenOrObservations.get(k) ) {
|
803 |
|
|
handleAssociation(acceptedTaxon, refMods, td, soo);
|
804 |
|
|
}
|
805 |
|
|
}
|
806 |
|
|
|
807 |
|
|
if (descriptionsFulltext.keySet().contains(k)){
|
808 |
6c41cae2
|
p.kelbert
|
if (!stringIsEmpty(descriptionsFulltext.get(k).trim()) && (descriptionsFulltext.get(k).startsWith("Hab.") || descriptionsFulltext.get(k).startsWith("Habitat"))){
|
809 |
2b5a7979
|
p.kelbert
|
setParticularDescription(descriptionsFulltext.get(k),acceptedTaxon,defaultTaxon, refMods, Feature.HABITAT());
|
810 |
|
|
break;
|
811 |
1de96f38
|
p.kelbert
|
}
|
812 |
2b5a7979
|
p.kelbert
|
else{
|
813 |
|
|
handleTextData(refMods, descriptionsFulltext, td, currentFeature, k);
|
814 |
1de96f38
|
p.kelbert
|
}
|
815 |
2b5a7979
|
p.kelbert
|
}
|
816 |
136e16b1
|
Andreas Müller
|
|
817 |
2b5a7979
|
p.kelbert
|
if (descriptionsFulltext.keySet().contains(k) || specimenOrObservations.keySet().contains(k)){
|
818 |
|
|
acceptedTaxon.addDescription(td);
|
819 |
|
|
sourceHandler.addAndSaveSource(refMods, td, null);
|
820 |
|
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
821 |
|
|
}
|
822 |
|
|
}
|
823 |
|
|
}
|
824 |
|
|
}
|
825 |
136e16b1
|
Andreas Müller
|
|
826 |
2b5a7979
|
p.kelbert
|
/**
|
827 |
|
|
* @param refMods
|
828 |
|
|
* @param descriptionsFulltext
|
829 |
|
|
* @param td
|
830 |
|
|
* @param currentFeature
|
831 |
|
|
* @param k
|
832 |
|
|
*/
|
833 |
|
|
private void handleTextData(Reference<?> refMods, Map<Integer, String> descriptionsFulltext, TaxonDescription td,
|
834 |
|
|
Feature currentFeature, int k) {
|
835 |
6c41cae2
|
p.kelbert
|
//logger.info("handleTextData");
|
836 |
2b5a7979
|
p.kelbert
|
TextData textData = TextData.NewInstance();
|
837 |
|
|
textData.setFeature(currentFeature);
|
838 |
|
|
textData.putText(Language.UNKNOWN_LANGUAGE(), descriptionsFulltext.get(k));
|
839 |
|
|
sourceHandler.addSource(refMods, textData);
|
840 |
|
|
td.addElement(textData);
|
841 |
|
|
}
|
842 |
285d1929
|
p.kelbert
|
|
843 |
2b5a7979
|
p.kelbert
|
/**
|
844 |
|
|
* @param acceptedTaxon
|
845 |
|
|
* @param refMods
|
846 |
|
|
* @param td
|
847 |
|
|
* @param soo
|
848 |
|
|
*/
|
849 |
|
|
private void handleAssociation(Taxon acceptedTaxon, Reference<?> refMods, TaxonDescription td, MySpecimenOrObservation soo) {
|
850 |
6c41cae2
|
p.kelbert
|
logger.info("handleAssociation");
|
851 |
2b5a7979
|
p.kelbert
|
String descr=soo.getDescr();
|
852 |
|
|
DerivedUnit derivedUnitBase = soo.getDerivedUnitBase();
|
853 |
285d1929
|
p.kelbert
|
|
854 |
2b5a7979
|
p.kelbert
|
sourceHandler.addAndSaveSource(refMods, derivedUnitBase);
|
855 |
136e16b1
|
Andreas Müller
|
|
856 |
2b5a7979
|
p.kelbert
|
TaxonDescription taxonDescription = importer.getTaxonDescription(acceptedTaxon, false, true);
|
857 |
6e861bce
|
Andreas Müller
|
|
858 |
2b5a7979
|
p.kelbert
|
Feature feature=null;
|
859 |
|
|
feature = makeFeature(derivedUnitBase);
|
860 |
|
|
if(!StringUtils.isEmpty(descr)) {
|
861 |
|
|
derivedUnitBase.setTitleCache(descr, true);
|
862 |
|
|
}
|
863 |
285d1929
|
p.kelbert
|
|
864 |
2b5a7979
|
p.kelbert
|
IndividualsAssociation indAssociation = createIndividualAssociation(refMods, derivedUnitBase, feature);
|
865 |
285d1929
|
p.kelbert
|
|
866 |
2b5a7979
|
p.kelbert
|
taxonDescription.addElement(indAssociation);
|
867 |
|
|
sourceHandler.addAndSaveSource(refMods, taxonDescription,null);
|
868 |
|
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
869 |
|
|
td.setDescribedSpecimenOrObservation(soo.getDerivedUnitBase());
|
870 |
|
|
}
|
871 |
285d1929
|
p.kelbert
|
|
872 |
2b5a7979
|
p.kelbert
|
/**
|
873 |
|
|
* create an individualAssociation
|
874 |
|
|
* @param refMods
|
875 |
|
|
* @param derivedUnitBase
|
876 |
|
|
* @param feature
|
877 |
|
|
* @return
|
878 |
|
|
*/
|
879 |
|
|
private IndividualsAssociation createIndividualAssociation(Reference<?> refMods, DerivedUnit derivedUnitBase,
|
880 |
|
|
Feature feature) {
|
881 |
6c41cae2
|
p.kelbert
|
logger.info("createIndividualAssociation");
|
882 |
2b5a7979
|
p.kelbert
|
IndividualsAssociation indAssociation = IndividualsAssociation.NewInstance();
|
883 |
|
|
indAssociation.setAssociatedSpecimenOrObservation(derivedUnitBase);
|
884 |
|
|
indAssociation.setFeature(feature);
|
885 |
|
|
indAssociation = sourceHandler.addSource(refMods, indAssociation);
|
886 |
|
|
return indAssociation;
|
887 |
|
|
}
|
888 |
136e16b1
|
Andreas Müller
|
|
889 |
2b5a7979
|
p.kelbert
|
/**
|
890 |
|
|
* @param specimenOrObservations
|
891 |
|
|
* @param descriptionsFulltext
|
892 |
|
|
* @param i
|
893 |
|
|
* @param specimenOrObservation
|
894 |
|
|
*/
|
895 |
|
|
private void extractTextFromSpecimenOrObservation(Map<Integer, List<MySpecimenOrObservation>> specimenOrObservations,
|
896 |
|
|
Map<Integer, String> descriptionsFulltext, int i, MySpecimenOrObservation specimenOrObservation) {
|
897 |
6c41cae2
|
p.kelbert
|
logger.info("extractTextFromSpecimenOrObservation");
|
898 |
2b5a7979
|
p.kelbert
|
List<MySpecimenOrObservation> speObsList = specimenOrObservations.get(i);
|
899 |
|
|
if (speObsList == null) {
|
900 |
|
|
speObsList=new ArrayList<MySpecimenOrObservation>();
|
901 |
|
|
}
|
902 |
|
|
speObsList.add(specimenOrObservation);
|
903 |
|
|
specimenOrObservations.put(i,speObsList);
|
904 |
136e16b1
|
Andreas Müller
|
|
905 |
2b5a7979
|
p.kelbert
|
String s = specimenOrObservation.getDerivedUnitBase().toString();
|
906 |
|
|
if (descriptionsFulltext.get(i) !=null){
|
907 |
|
|
s = descriptionsFulltext.get(i)+" "+s;
|
908 |
|
|
}
|
909 |
|
|
descriptionsFulltext.put(i, s);
|
910 |
|
|
}
|
911 |
|
|
|
912 |
|
|
/**
|
913 |
|
|
* Extract the text with the inline link to a taxon
|
914 |
|
|
* @param nametosave
|
915 |
|
|
* @param refMods
|
916 |
|
|
* @param descriptionsFulltext
|
917 |
|
|
* @param i
|
918 |
|
|
* @param paragraph
|
919 |
|
|
*/
|
920 |
|
|
@SuppressWarnings("rawtypes")
|
921 |
|
|
private void extractInLine(List<TaxonNameBase> nametosave, Reference<?> refMods, Map<Integer, String> descriptionsFulltext,
|
922 |
|
|
int i, Node paragraph) {
|
923 |
6c41cae2
|
p.kelbert
|
//logger.info("extractInLine");
|
924 |
2b5a7979
|
p.kelbert
|
String inLine=getInlineText(nametosave, refMods, paragraph);
|
925 |
|
|
if (descriptionsFulltext.get(i) !=null){
|
926 |
|
|
inLine = descriptionsFulltext.get(i)+inLine;
|
927 |
|
|
}
|
928 |
|
|
descriptionsFulltext.put(i, inLine);
|
929 |
|
|
}
|
930 |
|
|
|
931 |
|
|
/**
|
932 |
|
|
* Extract the raw text from a Node
|
933 |
|
|
* @param descriptionsFulltext
|
934 |
|
|
* @param node
|
935 |
|
|
* @param j
|
936 |
|
|
*/
|
937 |
|
|
private void extractText(Map<Integer, String> descriptionsFulltext, int i, Node node) {
|
938 |
6c41cae2
|
p.kelbert
|
//logger.info("extractText");
|
939 |
2b5a7979
|
p.kelbert
|
if(!node.getTextContent().trim().isEmpty()) {
|
940 |
|
|
String s =node.getTextContent().trim();
|
941 |
|
|
if (descriptionsFulltext.get(i) !=null){
|
942 |
|
|
s = descriptionsFulltext.get(i)+" "+s;
|
943 |
285d1929
|
p.kelbert
|
}
|
944 |
2b5a7979
|
p.kelbert
|
descriptionsFulltext.put(i, s);
|
945 |
285d1929
|
p.kelbert
|
}
|
946 |
|
|
}
|
947 |
|
|
|
948 |
|
|
|
949 |
|
|
/**
|
950 |
|
|
* @param materials: the XML node group
|
951 |
|
|
* @param acceptedTaxon: the current accepted Taxon
|
952 |
|
|
* @param refMods: the current reference extracted from the MODS
|
953 |
|
|
*/
|
954 |
|
|
@SuppressWarnings("rawtypes")
|
955 |
136e16b1
|
Andreas Müller
|
private void extractMaterials(Node materials, Taxon acceptedTaxon, Reference<?> refMods,List<TaxonNameBase> nametosave) {
|
956 |
6c41cae2
|
p.kelbert
|
logger.info("EXTRACTMATERIALS");
|
957 |
285d1929
|
p.kelbert
|
// logger.info("acceptedTaxon: "+acceptedTaxon);
|
958 |
|
|
NodeList children = materials.getChildNodes();
|
959 |
|
|
NodeList events = null;
|
960 |
2b5a7979
|
p.kelbert
|
// String descr="";
|
961 |
285d1929
|
p.kelbert
|
|
962 |
|
|
|
963 |
|
|
for (int i=0;i<children.getLength();i++){
|
964 |
136e16b1
|
Andreas Müller
|
String rawAssociation="";
|
965 |
|
|
boolean added=false;
|
966 |
285d1929
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
967 |
|
|
events = children.item(i).getChildNodes();
|
968 |
|
|
for(int k=0;k<events.getLength();k++){
|
969 |
136e16b1
|
Andreas Müller
|
if (events.item(k).getNodeName().equalsIgnoreCase("tax:name")){
|
970 |
2b5a7979
|
p.kelbert
|
String inLine= getInlineText(nametosave, refMods, events.item(k));
|
971 |
|
|
if(!inLine.isEmpty()) {
|
972 |
|
|
rawAssociation+=inLine;
|
973 |
|
|
}
|
974 |
136e16b1
|
Andreas Müller
|
}
|
975 |
|
|
if (! events.item(k).getNodeName().equalsIgnoreCase("tax:name")
|
976 |
|
|
&& !events.item(k).getNodeName().equalsIgnoreCase("tax:collection_event")){
|
977 |
|
|
rawAssociation+= events.item(k).getTextContent().trim();
|
978 |
|
|
}
|
979 |
285d1929
|
p.kelbert
|
if(events.item(k).getNodeName().equalsIgnoreCase("tax:collection_event")){
|
980 |
136e16b1
|
Andreas Müller
|
if (!containsDistinctLetters(rawAssociation.replaceAll(";",""))) {
|
981 |
|
|
rawAssociation="no description text";
|
982 |
|
|
}
|
983 |
|
|
added=true;
|
984 |
2b5a7979
|
p.kelbert
|
handleDerivedUnitFacadeAndBase(acceptedTaxon, refMods, events.item(k), rawAssociation);
|
985 |
|
|
}
|
986 |
|
|
if (!rawAssociation.isEmpty() && !added){
|
987 |
285d1929
|
p.kelbert
|
|
988 |
2b5a7979
|
p.kelbert
|
Feature feature = Feature.MATERIALS_EXAMINED();
|
989 |
|
|
featuresMap.put(feature.getTitleCache(),feature);
|
990 |
285d1929
|
p.kelbert
|
|
991 |
2b5a7979
|
p.kelbert
|
TextData textData = createTextData(rawAssociation, refMods, feature);
|
992 |
285d1929
|
p.kelbert
|
|
993 |
2b5a7979
|
p.kelbert
|
if(! rawAssociation.isEmpty() && (acceptedTaxon!=null)){
|
994 |
|
|
TaxonDescription td =importer.getTaxonDescription(acceptedTaxon, false, true);
|
995 |
|
|
td.addElement(textData);
|
996 |
|
|
acceptedTaxon.addDescription(td);
|
997 |
|
|
sourceHandler.addAndSaveSource(refMods, td, null);
|
998 |
1de96f38
|
p.kelbert
|
}
|
999 |
2b5a7979
|
p.kelbert
|
// DerivedUnitFacade derivedUnitFacade = getFacade(rawAssociation.replaceAll(";",""),SpecimenOrObservationType.DerivedUnit);
|
1000 |
|
|
// derivedUnitBase = derivedUnitFacade.innerDerivedUnit();
|
1001 |
|
|
//
|
1002 |
|
|
// TaxonDescription taxonDescription = importer.getTaxonDescription(acceptedTaxon, false, true);
|
1003 |
|
|
// acceptedTaxon.addDescription(taxonDescription);
|
1004 |
|
|
//
|
1005 |
|
|
// IndividualsAssociation indAssociation = IndividualsAssociation.NewInstance();
|
1006 |
|
|
//
|
1007 |
|
|
// Feature feature = Feature.MATERIALS_EXAMINED();
|
1008 |
|
|
// featuresMap.put(feature.getTitleCache(),feature);
|
1009 |
|
|
// if(!StringUtils.isEmpty(rawAssociation)) {
|
1010 |
|
|
// derivedUnitBase.setTitleCache(rawAssociation, true);
|
1011 |
|
|
// }
|
1012 |
|
|
// indAssociation.setAssociatedSpecimenOrObservation(derivedUnitBase);
|
1013 |
|
|
// indAssociation.setFeature(feature);
|
1014 |
|
|
// indAssociation.addSource(OriginalSourceType.Import, null, null, refMods, null);
|
1015 |
|
|
//
|
1016 |
|
|
// /*boolean sourceExists=false;
|
1017 |
|
|
// Set<DescriptionElementSource> dsources = indAssociation.getSources();
|
1018 |
|
|
// for (DescriptionElementSource src : dsources){
|
1019 |
|
|
// String micro = src.getCitationMicroReference();
|
1020 |
|
|
// Reference r = src.getCitation();
|
1021 |
|
|
// if (r.equals(refMods) && micro == null) {
|
1022 |
|
|
// sourceExists=true;
|
1023 |
|
|
// }
|
1024 |
|
|
// }
|
1025 |
|
|
// if(!sourceExists) {
|
1026 |
|
|
// indAssociation.addSource(null, null, refMods, null);
|
1027 |
|
|
// }*/
|
1028 |
|
|
// taxonDescription.addElement(indAssociation);
|
1029 |
|
|
// taxonDescription.setTaxon(acceptedTaxon);
|
1030 |
|
|
// taxonDescription.addSource(OriginalSourceType.Import, null,null,refMods,null);
|
1031 |
|
|
//
|
1032 |
|
|
// /*sourceExists=false;
|
1033 |
|
|
// Set<IdentifiableSource> sources = taxonDescription.getSources();
|
1034 |
|
|
// for (IdentifiableSource src : sources){
|
1035 |
|
|
// String micro = src.getCitationMicroReference();
|
1036 |
|
|
// Reference r = src.getCitation();
|
1037 |
|
|
// if (r.equals(refMods) && micro == null) {
|
1038 |
|
|
// sourceExists=true;
|
1039 |
|
|
// }
|
1040 |
|
|
// }
|
1041 |
|
|
// if(!sourceExists) {
|
1042 |
|
|
// taxonDescription.addSource(OriginalSourceType.Import,null,null,refMods,null);
|
1043 |
|
|
// }*/
|
1044 |
|
|
//
|
1045 |
|
|
// importer.getDescriptionService().saveOrUpdate(taxonDescription);
|
1046 |
|
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
1047 |
1de96f38
|
p.kelbert
|
|
1048 |
2b5a7979
|
p.kelbert
|
rawAssociation="";
|
1049 |
285d1929
|
p.kelbert
|
}
|
1050 |
2b5a7979
|
p.kelbert
|
}
|
1051 |
|
|
}
|
1052 |
|
|
}
|
1053 |
|
|
}
|
1054 |
1de96f38
|
p.kelbert
|
|
1055 |
2b5a7979
|
p.kelbert
|
/**
|
1056 |
|
|
* @param acceptedTaxon
|
1057 |
|
|
* @param refMods
|
1058 |
|
|
* @param events
|
1059 |
|
|
* @param rawAssociation
|
1060 |
|
|
* @param k
|
1061 |
|
|
*/
|
1062 |
|
|
private void handleDerivedUnitFacadeAndBase(Taxon acceptedTaxon, Reference<?> refMods, Node event,
|
1063 |
|
|
String rawAssociation) {
|
1064 |
6c41cae2
|
p.kelbert
|
logger.info("handleDerivedUnitFacadeAndBase");
|
1065 |
2b5a7979
|
p.kelbert
|
String descr;
|
1066 |
|
|
DerivedUnit derivedUnitBase;
|
1067 |
|
|
MySpecimenOrObservation myspecimenOrObservation;
|
1068 |
|
|
DerivedUnitFacade derivedUnitFacade = getFacade(rawAssociation.replaceAll(";",""),SpecimenOrObservationType.DerivedUnit);
|
1069 |
|
|
derivedUnitBase = derivedUnitFacade.innerDerivedUnit();
|
1070 |
285d1929
|
p.kelbert
|
|
1071 |
2b5a7979
|
p.kelbert
|
sourceHandler.addAndSaveSource(refMods, derivedUnitBase);
|
1072 |
285d1929
|
p.kelbert
|
|
1073 |
2b5a7979
|
p.kelbert
|
myspecimenOrObservation = extractSpecimenOrObservation(event,derivedUnitBase,SpecimenOrObservationType.DerivedUnit);
|
1074 |
|
|
derivedUnitBase = myspecimenOrObservation.getDerivedUnitBase();
|
1075 |
|
|
descr=myspecimenOrObservation.getDescr();
|
1076 |
1de96f38
|
p.kelbert
|
|
1077 |
2b5a7979
|
p.kelbert
|
sourceHandler.addAndSaveSource(refMods, derivedUnitBase);
|
1078 |
285d1929
|
p.kelbert
|
|
1079 |
2b5a7979
|
p.kelbert
|
TaxonDescription taxonDescription = importer.getTaxonDescription(acceptedTaxon, false, true);
|
1080 |
6e861bce
|
Andreas Müller
|
|
1081 |
2b5a7979
|
p.kelbert
|
Feature feature = makeFeature(derivedUnitBase);
|
1082 |
|
|
featuresMap.put(feature.getTitleCache(),feature);
|
1083 |
|
|
if(!StringUtils.isEmpty(descr)) {
|
1084 |
|
|
derivedUnitBase.setTitleCache(descr, true);
|
1085 |
285d1929
|
p.kelbert
|
}
|
1086 |
2b5a7979
|
p.kelbert
|
|
1087 |
|
|
IndividualsAssociation indAssociation = createIndividualAssociation(refMods, derivedUnitBase, feature);
|
1088 |
|
|
|
1089 |
|
|
taxonDescription.addElement(indAssociation);
|
1090 |
|
|
sourceHandler.addAndSaveSource(refMods, taxonDescription,null);
|
1091 |
|
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
1092 |
285d1929
|
p.kelbert
|
}
|
1093 |
|
|
|
1094 |
2b5a7979
|
p.kelbert
|
|
1095 |
|
|
|
1096 |
136e16b1
|
Andreas Müller
|
/**
|
1097 |
|
|
* @param materials: the XML node group
|
1098 |
|
|
* @param acceptedTaxon: the current accepted Taxon
|
1099 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1100 |
|
|
*/
|
1101 |
1de96f38
|
p.kelbert
|
private String extractMaterialsDirect(Node materials, Taxon acceptedTaxon, Reference<?> refMods, String event) {
|
1102 |
6c41cae2
|
p.kelbert
|
logger.info("extractMaterialsDirect");
|
1103 |
136e16b1
|
Andreas Müller
|
// logger.info("acceptedTaxon: "+acceptedTaxon);
|
1104 |
|
|
String descr="";
|
1105 |
|
|
|
1106 |
|
|
DerivedUnit derivedUnitBase=null;
|
1107 |
2b5a7979
|
p.kelbert
|
MySpecimenOrObservation myspecimenOrObservation = extractSpecimenOrObservation(materials,derivedUnitBase, SpecimenOrObservationType.DerivedUnit);
|
1108 |
136e16b1
|
Andreas Müller
|
derivedUnitBase = myspecimenOrObservation.getDerivedUnitBase();
|
1109 |
|
|
|
1110 |
2b5a7979
|
p.kelbert
|
sourceHandler.addAndSaveSource(refMods, derivedUnitBase);
|
1111 |
136e16b1
|
Andreas Müller
|
|
1112 |
|
|
TaxonDescription taxonDescription = importer.getTaxonDescription(acceptedTaxon, false, true);
|
1113 |
|
|
|
1114 |
|
|
Feature feature=null;
|
1115 |
|
|
if (event.equalsIgnoreCase("collection")){
|
1116 |
|
|
feature = makeFeature(derivedUnitBase);
|
1117 |
|
|
}
|
1118 |
|
|
else{
|
1119 |
|
|
feature = Feature.MATERIALS_EXAMINED();
|
1120 |
|
|
}
|
1121 |
2b5a7979
|
p.kelbert
|
featuresMap.put(feature.getTitleCache(), feature);
|
1122 |
|
|
|
1123 |
|
|
descr=myspecimenOrObservation.getDescr();
|
1124 |
136e16b1
|
Andreas Müller
|
if(!StringUtils.isEmpty(descr)) {
|
1125 |
6e861bce
|
Andreas Müller
|
derivedUnitBase.setTitleCache(descr, true);
|
1126 |
136e16b1
|
Andreas Müller
|
}
|
1127 |
|
|
|
1128 |
2b5a7979
|
p.kelbert
|
IndividualsAssociation indAssociation = createIndividualAssociation(refMods, derivedUnitBase, feature);
|
1129 |
|
|
|
1130 |
136e16b1
|
Andreas Müller
|
taxonDescription.addElement(indAssociation);
|
1131 |
2b5a7979
|
p.kelbert
|
sourceHandler.addAndSaveSource(refMods, taxonDescription,null);
|
1132 |
136e16b1
|
Andreas Müller
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
1133 |
|
|
|
1134 |
1de96f38
|
p.kelbert
|
return derivedUnitBase.getTitleCache();
|
1135 |
136e16b1
|
Andreas Müller
|
|
1136 |
|
|
}
|
1137 |
|
|
|
1138 |
285d1929
|
p.kelbert
|
|
1139 |
|
|
/**
|
1140 |
|
|
* @param description: the XML node group
|
1141 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
1142 |
|
|
* @param defaultTaxon: the current defaultTaxon, only used if there is no accepted name
|
1143 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
1144 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1145 |
|
|
* @param featureName: the feature name
|
1146 |
|
|
*/
|
1147 |
6e861bce
|
Andreas Müller
|
@SuppressWarnings({ "rawtypes"})
|
1148 |
1de96f38
|
p.kelbert
|
private String extractSpecificFeature(Node description, Taxon acceptedTaxon, Taxon defaultTaxon,
|
1149 |
285d1929
|
p.kelbert
|
List<TaxonNameBase> nametosave, Reference<?> refMods, String featureName ) {
|
1150 |
6c41cae2
|
p.kelbert
|
logger.info("extractSpecificFeature "+featureName);
|
1151 |
1de96f38
|
p.kelbert
|
// System.out.println("GRUUUUuu");
|
1152 |
285d1929
|
p.kelbert
|
NodeList children = description.getChildNodes();
|
1153 |
|
|
NodeList insideNodes ;
|
1154 |
1de96f38
|
p.kelbert
|
NodeList trNodes;
|
1155 |
2b5a7979
|
p.kelbert
|
// String descr ="";
|
1156 |
285d1929
|
p.kelbert
|
String localdescr="";
|
1157 |
1de96f38
|
p.kelbert
|
List<String> blabla=null;
|
1158 |
|
|
List<String> text = new ArrayList<String>();
|
1159 |
|
|
|
1160 |
|
|
String table="<table>";
|
1161 |
|
|
String head="";
|
1162 |
|
|
String line="";
|
1163 |
285d1929
|
p.kelbert
|
|
1164 |
2b5a7979
|
p.kelbert
|
Feature currentFeature=getFeatureObjectFromString(featureName);
|
1165 |
|
|
|
1166 |
285d1929
|
p.kelbert
|
// String fullContent = description.getTextContent();
|
1167 |
|
|
for (int i=0;i<children.getLength();i++){
|
1168 |
2b5a7979
|
p.kelbert
|
// localdescr="";
|
1169 |
285d1929
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("#text") && !children.item(i).getTextContent().trim().isEmpty()){
|
1170 |
2b5a7979
|
p.kelbert
|
text.add(children.item(i).getTextContent().trim());
|
1171 |
285d1929
|
p.kelbert
|
}
|
1172 |
1de96f38
|
p.kelbert
|
if (featureName.equalsIgnoreCase("table")){
|
1173 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
1174 |
|
|
children.item(i).getAttributes().getNamedItem("otherType").getNodeValue().equalsIgnoreCase("thead")){
|
1175 |
2b5a7979
|
p.kelbert
|
head = extractTableHead(children.item(i));
|
1176 |
1de96f38
|
p.kelbert
|
table+=head;
|
1177 |
2b5a7979
|
p.kelbert
|
line = extractTableLine(children.item(i));
|
1178 |
1de96f38
|
p.kelbert
|
if (!line.equalsIgnoreCase("<tr></tr>")) {
|
1179 |
|
|
table+=line;
|
1180 |
|
|
}
|
1181 |
|
|
}
|
1182 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
1183 |
|
|
children.item(i).getAttributes().getNamedItem("otherType").getNodeValue().equalsIgnoreCase("tr")){
|
1184 |
2b5a7979
|
p.kelbert
|
line = extractTableLineWithColumn(children.item(i).getChildNodes());
|
1185 |
1de96f38
|
p.kelbert
|
if(!line.equalsIgnoreCase("<tr></tr>")) {
|
1186 |
|
|
table+=line;
|
1187 |
|
|
}
|
1188 |
|
|
}
|
1189 |
|
|
}
|
1190 |
285d1929
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
1191 |
|
|
insideNodes=children.item(i).getChildNodes();
|
1192 |
1de96f38
|
p.kelbert
|
blabla= new ArrayList<String>();
|
1193 |
285d1929
|
p.kelbert
|
for (int j=0;j<insideNodes.getLength();j++){
|
1194 |
|
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("tax:name")){
|
1195 |
2b5a7979
|
p.kelbert
|
String inlinetext = getInlineText(nametosave, refMods, insideNodes.item(j));
|
1196 |
|
|
if (!inlinetext.isEmpty()) {
|
1197 |
|
|
blabla.add(inlinetext);
|
1198 |
|
|
}
|
1199 |
285d1929
|
p.kelbert
|
}
|
1200 |
|
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("#text")) {
|
1201 |
|
|
if(!insideNodes.item(j).getTextContent().trim().isEmpty()){
|
1202 |
|
|
blabla.add(insideNodes.item(j).getTextContent().trim());
|
1203 |
2b5a7979
|
p.kelbert
|
// localdescr += insideNodes.item(j).getTextContent().trim();
|
1204 |
285d1929
|
p.kelbert
|
}
|
1205 |
|
|
}
|
1206 |
|
|
}
|
1207 |
|
|
if (!blabla.isEmpty()) {
|
1208 |
2b5a7979
|
p.kelbert
|
String blaStr = StringUtils.join(blabla," ").trim();
|
1209 |
6c41cae2
|
p.kelbert
|
if(!stringIsEmpty(blaStr)) {
|
1210 |
2b5a7979
|
p.kelbert
|
setParticularDescription(blaStr,acceptedTaxon,defaultTaxon, refMods,currentFeature);
|
1211 |
6c41cae2
|
p.kelbert
|
text.add(blaStr);
|
1212 |
285d1929
|
p.kelbert
|
}
|
1213 |
|
|
}
|
1214 |
6c41cae2
|
p.kelbert
|
|
1215 |
285d1929
|
p.kelbert
|
}
|
1216 |
2b5a7979
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("#text")){
|
1217 |
|
|
if(!children.item(i).getTextContent().trim().isEmpty()){
|
1218 |
|
|
localdescr = children.item(i).getTextContent().trim();
|
1219 |
6c41cae2
|
p.kelbert
|
if(!stringIsEmpty(localdescr)) {
|
1220 |
|
|
setParticularDescription(localdescr,acceptedTaxon,defaultTaxon, refMods,currentFeature);
|
1221 |
|
|
}
|
1222 |
2b5a7979
|
p.kelbert
|
}
|
1223 |
|
|
}
|
1224 |
1de96f38
|
p.kelbert
|
}
|
1225 |
285d1929
|
p.kelbert
|
|
1226 |
1de96f38
|
p.kelbert
|
table+="</table>";
|
1227 |
|
|
if (!table.equalsIgnoreCase("<table></table>")){
|
1228 |
2b5a7979
|
p.kelbert
|
// System.out.println("TABLE : "+table);
|
1229 |
1de96f38
|
p.kelbert
|
text.add(table);
|
1230 |
|
|
}
|
1231 |
|
|
|
1232 |
|
|
if (text !=null && !text.isEmpty()) {
|
1233 |
|
|
return StringUtils.join(text," ");
|
1234 |
|
|
} else {
|
1235 |
|
|
return "";
|
1236 |
285d1929
|
p.kelbert
|
}
|
1237 |
|
|
|
1238 |
|
|
}
|
1239 |
|
|
|
1240 |
2b5a7979
|
p.kelbert
|
/**
|
1241 |
|
|
* @param children
|
1242 |
|
|
* @param i
|
1243 |
|
|
* @return
|
1244 |
|
|
*/
|
1245 |
|
|
private String extractTableLine(Node child) {
|
1246 |
6c41cae2
|
p.kelbert
|
//logger.info("extractTableLine");
|
1247 |
2b5a7979
|
p.kelbert
|
String line;
|
1248 |
|
|
line="<tr>";
|
1249 |
|
|
if (child.getAttributes().getNamedItem("otherType").getNodeValue().equalsIgnoreCase("tr")){
|
1250 |
|
|
line = extractTableLineWithColumn(child.getChildNodes());
|
1251 |
|
|
}
|
1252 |
|
|
line+="</tr>";
|
1253 |
|
|
return line;
|
1254 |
|
|
}
|
1255 |
|
|
|
1256 |
|
|
/**
|
1257 |
|
|
* @param children
|
1258 |
|
|
* @param i
|
1259 |
|
|
* @return
|
1260 |
|
|
*/
|
1261 |
|
|
private String extractTableHead(Node child) {
|
1262 |
6c41cae2
|
p.kelbert
|
//logger.info("extractTableHead");
|
1263 |
2b5a7979
|
p.kelbert
|
String head;
|
1264 |
|
|
String line;
|
1265 |
|
|
head="<th>";
|
1266 |
|
|
NodeList trNodes = child.getChildNodes();
|
1267 |
|
|
for (int k=0;k<trNodes.getLength();k++){
|
1268 |
|
|
if (trNodes.item(k).getNodeName().equalsIgnoreCase("tax:div")
|
1269 |
|
|
&& trNodes.item(k).getAttributes().getNamedItem("otherType").getNodeValue().equalsIgnoreCase("tr")){
|
1270 |
|
|
line = extractTableLineWithColumn(trNodes.item(k).getChildNodes());
|
1271 |
|
|
head+=line;
|
1272 |
|
|
}
|
1273 |
|
|
}
|
1274 |
|
|
head+="</th>";
|
1275 |
|
|
return head;
|
1276 |
|
|
}
|
1277 |
|
|
|
1278 |
|
|
/**
|
1279 |
|
|
* build a html table line, with td columns
|
1280 |
|
|
* @param tdNodes
|
1281 |
|
|
* @return an html coded line
|
1282 |
|
|
*/
|
1283 |
|
|
private String extractTableLineWithColumn(NodeList tdNodes) {
|
1284 |
6c41cae2
|
p.kelbert
|
//logger.info("extractTableLineWithColumn");
|
1285 |
2b5a7979
|
p.kelbert
|
String line;
|
1286 |
|
|
line="<tr>";
|
1287 |
|
|
for (int l=0;l<tdNodes.getLength();l++){
|
1288 |
|
|
if (tdNodes.item(l).getNodeName().equalsIgnoreCase("tax:p")){
|
1289 |
|
|
line+="<td>"+tdNodes.item(l).getTextContent()+"</td>";
|
1290 |
|
|
}
|
1291 |
|
|
}
|
1292 |
|
|
line+="</tr>";
|
1293 |
|
|
return line;
|
1294 |
|
|
}
|
1295 |
|
|
|
1296 |
|
|
/**
|
1297 |
|
|
* @param description: the XML node group
|
1298 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
1299 |
|
|
* @param defaultTaxon: the current defaultTaxon, only used if there is no accepted name
|
1300 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
1301 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1302 |
|
|
* @param featureName: the feature name
|
1303 |
|
|
*/
|
1304 |
|
|
@SuppressWarnings({ "unused", "rawtypes" })
|
1305 |
|
|
private String extractSpecificFeatureNotStructured(Node description, Taxon acceptedTaxon, Taxon defaultTaxon,
|
1306 |
|
|
List<TaxonNameBase> nametosave, Reference<?> refMods, String featureName ) {
|
1307 |
6c41cae2
|
p.kelbert
|
logger.info("extractSpecificFeatureNotStructured "+featureName);
|
1308 |
2b5a7979
|
p.kelbert
|
NodeList children = description.getChildNodes();
|
1309 |
|
|
NodeList insideNodes ;
|
1310 |
|
|
List<String> blabla= new ArrayList<String>();
|
1311 |
|
|
|
1312 |
|
|
|
1313 |
|
|
Feature currentFeature = getFeatureObjectFromString(featureName);
|
1314 |
|
|
|
1315 |
|
|
String fullContent = description.getTextContent();
|
1316 |
|
|
for (int i=0;i<children.getLength();i++){
|
1317 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
1318 |
|
|
insideNodes=children.item(i).getChildNodes();
|
1319 |
|
|
for (int j=0;j<insideNodes.getLength();j++){
|
1320 |
|
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("tax:name")){
|
1321 |
|
|
String inlineText =getInlineText(nametosave, refMods, insideNodes.item(j));
|
1322 |
|
|
if(!inlineText.isEmpty()) {
|
1323 |
|
|
blabla.add(inlineText);
|
1324 |
|
|
}
|
1325 |
|
|
}
|
1326 |
|
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("#text")) {
|
1327 |
|
|
if(!insideNodes.item(j).getTextContent().trim().isEmpty()){
|
1328 |
|
|
blabla.add(insideNodes.item(j).getTextContent().trim());
|
1329 |
|
|
}
|
1330 |
|
|
}
|
1331 |
|
|
}
|
1332 |
|
|
}
|
1333 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("#text")){
|
1334 |
|
|
if(!children.item(i).getTextContent().trim().isEmpty()){
|
1335 |
|
|
String localdescr = children.item(i).getTextContent().trim();
|
1336 |
|
|
if(!localdescr.isEmpty())
|
1337 |
|
|
{
|
1338 |
|
|
blabla.add(localdescr);
|
1339 |
|
|
}
|
1340 |
|
|
}
|
1341 |
|
|
}
|
1342 |
|
|
}
|
1343 |
|
|
|
1344 |
|
|
if (blabla !=null && !blabla.isEmpty()) {
|
1345 |
|
|
String blaStr = StringUtils.join(blabla," ").trim();
|
1346 |
6c41cae2
|
p.kelbert
|
if (! stringIsEmpty(blaStr)) {
|
1347 |
2b5a7979
|
p.kelbert
|
setParticularDescription(blaStr,acceptedTaxon,defaultTaxon, refMods,currentFeature);
|
1348 |
6c41cae2
|
p.kelbert
|
return blaStr;
|
1349 |
|
|
} else {
|
1350 |
|
|
return "";
|
1351 |
2b5a7979
|
p.kelbert
|
}
|
1352 |
|
|
} else {
|
1353 |
|
|
return "";
|
1354 |
|
|
}
|
1355 |
|
|
|
1356 |
|
|
}
|
1357 |
|
|
|
1358 |
6c41cae2
|
p.kelbert
|
/**
|
1359 |
|
|
* @param blaStr
|
1360 |
|
|
* @return
|
1361 |
|
|
*/
|
1362 |
|
|
private boolean stringIsEmpty(String blaStr) {
|
1363 |
|
|
if (!StringUtils.isEmpty(blaStr)) {
|
1364 |
|
|
if (!blaStr.equalsIgnoreCase(".")) {
|
1365 |
|
|
if (!blaStr.equalsIgnoreCase(",")) {
|
1366 |
|
|
if (!blaStr.equalsIgnoreCase(";")) {
|
1367 |
|
|
return false;
|
1368 |
|
|
}
|
1369 |
|
|
}
|
1370 |
|
|
}
|
1371 |
|
|
}
|
1372 |
|
|
return true;
|
1373 |
|
|
}
|
1374 |
|
|
|
1375 |
2b5a7979
|
p.kelbert
|
/**
|
1376 |
|
|
* @param nametosave
|
1377 |
|
|
* @param refMods
|
1378 |
|
|
* @param insideNodes
|
1379 |
|
|
* @param blabla
|
1380 |
|
|
* @param j
|
1381 |
|
|
*/
|
1382 |
|
|
@SuppressWarnings({ "rawtypes" })
|
1383 |
|
|
private String getInlineText(List<TaxonNameBase> nametosave, Reference<?> refMods, Node insideNode) {
|
1384 |
6c41cae2
|
p.kelbert
|
//logger.info("getInlineText");
|
1385 |
6b2c2013
|
p.kelbert
|
TaxonNameBase tnb = getTaxonNameBaseFromXML(insideNode, nametosave,refMods,false);
|
1386 |
2b5a7979
|
p.kelbert
|
// Taxon tax = getTaxonFromTxonNameBase(tnb, refMods);
|
1387 |
|
|
Taxon tax = currentMyName.getTaxon();
|
1388 |
|
|
if(tnb !=null){
|
1389 |
|
|
String linkedTaxon = tnb.toString().split("sec")[0];//TODO NOT IMPLEMENTED IN THE CDM YET
|
1390 |
|
|
return "<cdm:taxon uuid='"+tax.getUuid()+"'>"+linkedTaxon+"</cdm:taxon>";
|
1391 |
|
|
}
|
1392 |
|
|
return "";
|
1393 |
|
|
}
|
1394 |
|
|
|
1395 |
|
|
/**
|
1396 |
|
|
* @param featureName
|
1397 |
|
|
* @return
|
1398 |
|
|
*/
|
1399 |
|
|
@SuppressWarnings("rawtypes")
|
1400 |
|
|
private Feature getFeatureObjectFromString(String featureName) {
|
1401 |
6c41cae2
|
p.kelbert
|
logger.info("getFeatureObjectFromString");
|
1402 |
fdf9eec2
|
Andreas Müller
|
List<Feature> features = importer.getTermService().list(Feature.class, null,null,null,null);
|
1403 |
2b5a7979
|
p.kelbert
|
Feature currentFeature=null;
|
1404 |
fdf9eec2
|
Andreas Müller
|
for (Feature feature: features){
|
1405 |
|
|
String tmpF = feature.getTitleCache();
|
1406 |
2b5a7979
|
p.kelbert
|
if (tmpF.equalsIgnoreCase(featureName)) {
|
1407 |
fdf9eec2
|
Andreas Müller
|
currentFeature=feature;
|
1408 |
2b5a7979
|
p.kelbert
|
// System.out.println("currentFeatureFromList "+currentFeature.getUuid());
|
1409 |
|
|
}
|
1410 |
|
|
}
|
1411 |
|
|
if (currentFeature == null) {
|
1412 |
|
|
currentFeature=Feature.NewInstance(featureName, featureName, featureName);
|
1413 |
|
|
if(featureName.equalsIgnoreCase("Other")){
|
1414 |
|
|
currentFeature.setUuid(OtherUUID);
|
1415 |
|
|
}
|
1416 |
|
|
if(featureName.equalsIgnoreCase(notMarkedUp)){
|
1417 |
|
|
currentFeature.setUuid(NotMarkedUpUUID);
|
1418 |
|
|
}
|
1419 |
|
|
importer.getTermService().saveOrUpdate(currentFeature);
|
1420 |
|
|
}
|
1421 |
|
|
return currentFeature;
|
1422 |
|
|
}
|
1423 |
|
|
|
1424 |
285d1929
|
p.kelbert
|
|
1425 |
|
|
|
1426 |
|
|
|
1427 |
|
|
/**
|
1428 |
|
|
* @param children: the XML node group
|
1429 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
1430 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
1431 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1432 |
|
|
* @param fullContent :the parsed XML content
|
1433 |
|
|
* @return a list of description (text)
|
1434 |
|
|
*/
|
1435 |
2b5a7979
|
p.kelbert
|
@SuppressWarnings({ "unused", "rawtypes" })
|
1436 |
285d1929
|
p.kelbert
|
private List<String> parseParagraph(List<TaxonNameBase> nametosave, Taxon acceptedTaxon, Reference<?> refMods, Node paragraph, Feature feature){
|
1437 |
6c41cae2
|
p.kelbert
|
logger.info("parseParagraph "+feature.toString());
|
1438 |
285d1929
|
p.kelbert
|
List<String> fullDescription= new ArrayList<String>();
|
1439 |
136e16b1
|
Andreas Müller
|
// String localdescr;
|
1440 |
285d1929
|
p.kelbert
|
String descr="";
|
1441 |
|
|
NodeList insideNodes ;
|
1442 |
|
|
boolean collectionEvent = false;
|
1443 |
136e16b1
|
Andreas Müller
|
List<Node>collectionEvents = new ArrayList<Node>();
|
1444 |
285d1929
|
p.kelbert
|
|
1445 |
|
|
NodeList children = paragraph.getChildNodes();
|
1446 |
|
|
|
1447 |
|
|
for (int i=0;i<children.getLength();i++){
|
1448 |
136e16b1
|
Andreas Müller
|
// localdescr="";
|
1449 |
285d1929
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("#text") && !children.item(i).getTextContent().trim().isEmpty()){
|
1450 |
|
|
descr += children.item(i).getTextContent().trim();
|
1451 |
|
|
}
|
1452 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
1453 |
|
|
insideNodes=children.item(i).getChildNodes();
|
1454 |
|
|
List<String> blabla= new ArrayList<String>();
|
1455 |
|
|
for (int j=0;j<insideNodes.getLength();j++){
|
1456 |
136e16b1
|
Andreas Müller
|
boolean nodeKnown = false;
|
1457 |
2b5a7979
|
p.kelbert
|
// System.out.println("insideNodes.item(j).getNodeName() : "+insideNodes.item(j).getNodeName());
|
1458 |
285d1929
|
p.kelbert
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("tax:name")){
|
1459 |
2b5a7979
|
p.kelbert
|
String inlineText = getInlineText(nametosave, refMods, insideNodes.item(j));
|
1460 |
|
|
if (!inlineText.isEmpty()) {
|
1461 |
|
|
blabla.add(inlineText);
|
1462 |
|
|
}
|
1463 |
136e16b1
|
Andreas Müller
|
nodeKnown=true;
|
1464 |
285d1929
|
p.kelbert
|
}
|
1465 |
|
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("#text")) {
|
1466 |
|
|
if(!insideNodes.item(j).getTextContent().trim().isEmpty()){
|
1467 |
|
|
blabla.add(insideNodes.item(j).getTextContent().trim());
|
1468 |
136e16b1
|
Andreas Müller
|
// localdescr += insideNodes.item(j).getTextContent().trim();
|
1469 |
|
|
}
|
1470 |
|
|
nodeKnown=true;
|
1471 |
|
|
}
|
1472 |
|
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("tax:bibref")) {
|
1473 |
|
|
String ref = insideNodes.item(j).getTextContent().trim();
|
1474 |
|
|
if (ref.endsWith(";") && ((ref.length())>1)) {
|
1475 |
|
|
ref=ref.substring(0, ref.length()-1)+".";
|
1476 |
285d1929
|
p.kelbert
|
}
|
1477 |
136e16b1
|
Andreas Müller
|
Reference<?> reference = ReferenceFactory.newGeneric();
|
1478 |
|
|
reference.setTitleCache(ref, true);
|
1479 |
|
|
blabla.add(reference.getTitleCache());
|
1480 |
|
|
nodeKnown=true;
|
1481 |
285d1929
|
p.kelbert
|
}
|
1482 |
1de96f38
|
p.kelbert
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("tax:figure")){
|
1483 |
|
|
String figure = extractSpecificFeature(insideNodes.item(j),acceptedTaxon,acceptedTaxon, nametosave, refMods, "figure");
|
1484 |
|
|
blabla.add(figure);
|
1485 |
|
|
}
|
1486 |
|
|
if(insideNodes.item(j).getNodeName().equalsIgnoreCase("tax:div") &&
|
1487 |
2b5a7979
|
p.kelbert
|
insideNodes.item(j).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("Other") &&
|
1488 |
1de96f38
|
p.kelbert
|
insideNodes.item(j).getAttributes().getNamedItem("otherType").getNodeValue().equalsIgnoreCase("table")){
|
1489 |
|
|
String table = extractSpecificFeature(insideNodes.item(j),acceptedTaxon,acceptedTaxon, nametosave, refMods, "table");
|
1490 |
|
|
blabla.add(table);
|
1491 |
|
|
}
|
1492 |
285d1929
|
p.kelbert
|
if (insideNodes.item(j).getNodeName().equalsIgnoreCase("tax:collection_event")) {
|
1493 |
2b5a7979
|
p.kelbert
|
// logger.warn("SEEMS TO BE COLLECTION EVENT INSIDE A "+feature.toString());
|
1494 |
1de96f38
|
p.kelbert
|
String titlecache = extractMaterialsDirect(insideNodes.item(j), acceptedTaxon, refMods, "collection");
|
1495 |
|
|
blabla.add(titlecache);
|
1496 |
285d1929
|
p.kelbert
|
collectionEvent=true;
|
1497 |
136e16b1
|
Andreas Müller
|
collectionEvents.add(insideNodes.item(j));
|
1498 |
|
|
nodeKnown=true;
|
1499 |
|
|
}
|
1500 |
2b5a7979
|
p.kelbert
|
// if (!nodeKnown && !insideNodes.item(j).getNodeName().equalsIgnoreCase("tax:pb")) {
|
1501 |
|
|
// logger.info("Node not handled yet : "+insideNodes.item(j).getNodeName());
|
1502 |
|
|
// }
|
1503 |
285d1929
|
p.kelbert
|
|
1504 |
|
|
}
|
1505 |
6c41cae2
|
p.kelbert
|
if (!StringUtils.isEmpty(StringUtils.join(blabla," "))) {
|
1506 |
285d1929
|
p.kelbert
|
fullDescription.add(StringUtils.join(blabla," "));
|
1507 |
|
|
}
|
1508 |
|
|
}
|
1509 |
1de96f38
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:figure")){
|
1510 |
2b5a7979
|
p.kelbert
|
String figure = extractSpecificFeature(children.item(i),acceptedTaxon,acceptedTaxon, nametosave, refMods, "Figure");
|
1511 |
1de96f38
|
p.kelbert
|
fullDescription.add(figure);
|
1512 |
|
|
}
|
1513 |
|
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:div") &&
|
1514 |
2b5a7979
|
p.kelbert
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("Other") &&
|
1515 |
1de96f38
|
p.kelbert
|
children.item(i).getAttributes().getNamedItem("otherType").getNodeValue().equalsIgnoreCase("table")){
|
1516 |
|
|
String table = extractSpecificFeature(children.item(i),acceptedTaxon,acceptedTaxon, nametosave, refMods, "table");
|
1517 |
|
|
fullDescription.add(table);
|
1518 |
136e16b1
|
Andreas Müller
|
}
|
1519 |
285d1929
|
p.kelbert
|
}
|
1520 |
2b5a7979
|
p.kelbert
|
|
1521 |
6c41cae2
|
p.kelbert
|
if( !stringIsEmpty(descr.trim())){
|
1522 |
2b5a7979
|
p.kelbert
|
Feature currentFeature= getNotMarkedUpFeatureObject();
|
1523 |
6c41cae2
|
p.kelbert
|
setParticularDescription(descr.trim(),acceptedTaxon,acceptedTaxon, refMods,currentFeature);
|
1524 |
2b5a7979
|
p.kelbert
|
}
|
1525 |
1de96f38
|
p.kelbert
|
// if (collectionEvent) {
|
1526 |
|
|
// logger.warn("SEEMS TO BE COLLECTION EVENT INSIDE A "+feature.toString());
|
1527 |
|
|
// for (Node coll:collectionEvents){
|
1528 |
|
|
// = extractMaterialsDirect(coll, acceptedTaxon, refMods, "collection");
|
1529 |
|
|
// }
|
1530 |
|
|
// }
|
1531 |
285d1929
|
p.kelbert
|
return fullDescription;
|
1532 |
|
|
}
|
1533 |
|
|
|
1534 |
|
|
|
1535 |
|
|
/**
|
1536 |
|
|
* @param description: the XML node group
|
1537 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
1538 |
|
|
* @param defaultTaxon: the current defaultTaxon, only used if there is no accepted name
|
1539 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
1540 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1541 |
|
|
* @param feature: the feature to link the data with
|
1542 |
|
|
*/
|
1543 |
2b5a7979
|
p.kelbert
|
@SuppressWarnings("rawtypes")
|
1544 |
285d1929
|
p.kelbert
|
private void extractFeature(Node description, Taxon acceptedTaxon, Taxon defaultTaxon, List<TaxonNameBase> nametosave, Reference<?> refMods, Feature feature){
|
1545 |
6c41cae2
|
p.kelbert
|
logger.info("EXTRACT FEATURE "+feature.toString());
|
1546 |
285d1929
|
p.kelbert
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
1547 |
|
|
List<String> fullDescription= parseParagraph( nametosave, acceptedTaxon, refMods, description,feature);
|
1548 |
|
|
|
1549 |
2b5a7979
|
p.kelbert
|
// System.out.println("Feature : "+feature.toString()+", "+fullDescription.toString());
|
1550 |
6c41cae2
|
p.kelbert
|
if (!fullDescription.isEmpty() &&!stringIsEmpty(StringUtils.join(fullDescription," ").trim())) {
|
1551 |
|
|
setParticularDescription(StringUtils.join(fullDescription," ").trim(),acceptedTaxon,defaultTaxon, refMods,feature);
|
1552 |
285d1929
|
p.kelbert
|
}
|
1553 |
|
|
|
1554 |
|
|
}
|
1555 |
|
|
|
1556 |
|
|
|
1557 |
|
|
/**
|
1558 |
|
|
* @param descr: the XML Nodegroup to parse
|
1559 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
1560 |
|
|
* @param defaultTaxon: the current defaultTaxon, only used if there is no accepted name
|
1561 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1562 |
|
|
* @param currentFeature: the feature name
|
1563 |
|
|
* @return
|
1564 |
|
|
*/
|
1565 |
|
|
private void setParticularDescription(String descr, Taxon acceptedTaxon, Taxon defaultTaxon, Reference<?> refMods, Feature currentFeature) {
|
1566 |
2b5a7979
|
p.kelbert
|
logger.info("setParticularDescription "+currentFeature.getTitleCache()+", \n blabla : "+descr);
|
1567 |
|
|
// System.out.println("setParticularDescription "+currentFeature.getTitleCache()+", \n blabla : "+descr);
|
1568 |
285d1929
|
p.kelbert
|
// logger.info("acceptedTaxon: "+acceptedTaxon);
|
1569 |
|
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
1570 |
|
|
|
1571 |
2b5a7979
|
p.kelbert
|
featuresMap.put(currentFeature.getTitleCache(),currentFeature);
|
1572 |
285d1929
|
p.kelbert
|
|
1573 |
2b5a7979
|
p.kelbert
|
TextData textData = createTextData(descr, refMods, currentFeature);
|
1574 |
285d1929
|
p.kelbert
|
|
1575 |
6c41cae2
|
p.kelbert
|
if(acceptedTaxon!=null){
|
1576 |
285d1929
|
p.kelbert
|
TaxonDescription td =importer.getTaxonDescription(acceptedTaxon, false, true);
|
1577 |
|
|
td.addElement(textData);
|
1578 |
|
|
acceptedTaxon.addDescription(td);
|
1579 |
2b5a7979
|
p.kelbert
|
|
1580 |
|
|
sourceHandler.addAndSaveSource(refMods, td, null);
|
1581 |
285d1929
|
p.kelbert
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
1582 |
|
|
}
|
1583 |
|
|
|
1584 |
136e16b1
|
Andreas Müller
|
if(! descr.isEmpty() && (acceptedTaxon == null) && (defaultTaxon != null)){
|
1585 |
285d1929
|
p.kelbert
|
try{
|
1586 |
|
|
Taxon tmp =(Taxon) importer.getTaxonService().find(defaultTaxon.getUuid());
|
1587 |
|
|
if (tmp!=null) {
|
1588 |
|
|
defaultTaxon=CdmBase.deproxy(tmp,Taxon.class);
|
1589 |
|
|
}else{
|
1590 |
|
|
importer.getTaxonService().saveOrUpdate(defaultTaxon);
|
1591 |
|
|
}
|
1592 |
|
|
}catch(Exception e){
|
1593 |
|
|
logger.debug("TAXON EXISTS"+defaultTaxon);
|
1594 |
|
|
}
|
1595 |
|
|
|
1596 |
|
|
TaxonDescription td =importer.getTaxonDescription(defaultTaxon, false, true);
|
1597 |
|
|
defaultTaxon.addDescription(td);
|
1598 |
|
|
td.addElement(textData);
|
1599 |
2b5a7979
|
p.kelbert
|
sourceHandler.addAndSaveSource(refMods, td, null);
|
1600 |
285d1929
|
p.kelbert
|
importer.getTaxonService().saveOrUpdate(defaultTaxon);
|
1601 |
|
|
}
|
1602 |
|
|
}
|
1603 |
|
|
|
1604 |
2b5a7979
|
p.kelbert
|
/**
|
1605 |
|
|
* @param descr
|
1606 |
|
|
* @param refMods
|
1607 |
|
|
* @param currentFeature
|
1608 |
|
|
* @return
|
1609 |
|
|
*/
|
1610 |
|
|
private TextData createTextData(String descr, Reference<?> refMods, Feature currentFeature) {
|
1611 |
6c41cae2
|
p.kelbert
|
//logger.info("createTextData");
|
1612 |
2b5a7979
|
p.kelbert
|
TextData textData = TextData.NewInstance();
|
1613 |
|
|
textData.setFeature(currentFeature);
|
1614 |
|
|
sourceHandler.addSource(refMods, textData);
|
1615 |
|
|
|
1616 |
|
|
textData.putText(Language.UNKNOWN_LANGUAGE(), descr);
|
1617 |
|
|
return textData;
|
1618 |
|
|
}
|
1619 |
|
|
|
1620 |
|
|
|
1621 |
|
|
|
1622 |
1de96f38
|
p.kelbert
|
/**
|
1623 |
|
|
* @param descr: the XML Nodegroup to parse
|
1624 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
1625 |
|
|
* @param defaultTaxon: the current defaultTaxon, only used if there is no accepted name
|
1626 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1627 |
|
|
* @param currentFeature: the feature name
|
1628 |
|
|
* @return
|
1629 |
|
|
*/
|
1630 |
|
|
private void setParticularDescription(String descr, Taxon acceptedTaxon, Taxon defaultTaxon,Reference<?> currentRef, Reference<?> refMods, Feature currentFeature) {
|
1631 |
2b5a7979
|
p.kelbert
|
// System.out.println("setParticularDescriptionSPecial "+currentFeature);
|
1632 |
1de96f38
|
p.kelbert
|
// logger.info("acceptedTaxon: "+acceptedTaxon);
|
1633 |
6c41cae2
|
p.kelbert
|
logger.info("setParticularDescription");
|
1634 |
1de96f38
|
p.kelbert
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
1635 |
|
|
|
1636 |
2b5a7979
|
p.kelbert
|
featuresMap.put(currentFeature.getTitleCache(),currentFeature);
|
1637 |
|
|
TextData textData = createTextData(descr, refMods, currentFeature);
|
1638 |
1de96f38
|
p.kelbert
|
|
1639 |
|
|
if(! descr.isEmpty() && (acceptedTaxon!=null)){
|
1640 |
|
|
TaxonDescription td =importer.getTaxonDescription(acceptedTaxon, false, true);
|
1641 |
|
|
td.addElement(textData);
|
1642 |
|
|
acceptedTaxon.addDescription(td);
|
1643 |
2b5a7979
|
p.kelbert
|
|
1644 |
|
|
sourceHandler.addAndSaveSource(refMods, td, currentRef);
|
1645 |
1de96f38
|
p.kelbert
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
1646 |
|
|
}
|
1647 |
|
|
|
1648 |
|
|
if(! descr.isEmpty() && (acceptedTaxon == null) && (defaultTaxon != null)){
|
1649 |
|
|
try{
|
1650 |
|
|
Taxon tmp =(Taxon) importer.getTaxonService().find(defaultTaxon.getUuid());
|
1651 |
|
|
if (tmp!=null) {
|
1652 |
|
|
defaultTaxon=CdmBase.deproxy(tmp,Taxon.class);
|
1653 |
|
|
}else{
|
1654 |
|
|
importer.getTaxonService().saveOrUpdate(defaultTaxon);
|
1655 |
|
|
}
|
1656 |
|
|
}catch(Exception e){
|
1657 |
|
|
logger.debug("TAXON EXISTS"+defaultTaxon);
|
1658 |
|
|
}
|
1659 |
|
|
|
1660 |
|
|
TaxonDescription td =importer.getTaxonDescription(defaultTaxon, false, true);
|
1661 |
|
|
defaultTaxon.addDescription(td);
|
1662 |
|
|
td.addElement(textData);
|
1663 |
2b5a7979
|
p.kelbert
|
sourceHandler.addAndSaveSource(currentRef, td,currentRef);
|
1664 |
1de96f38
|
p.kelbert
|
importer.getTaxonService().saveOrUpdate(defaultTaxon);
|
1665 |
|
|
}
|
1666 |
|
|
}
|
1667 |
|
|
|
1668 |
285d1929
|
p.kelbert
|
|
1669 |
|
|
|
1670 |
|
|
/**
|
1671 |
|
|
* @param synonyms: the XML Nodegroup to parse
|
1672 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
1673 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
1674 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1675 |
|
|
*/
|
1676 |
2b5a7979
|
p.kelbert
|
@SuppressWarnings({ "rawtypes" })
|
1677 |
|
|
private void extractSynonyms(Node synonyms, Taxon acceptedTaxon,Reference<?> refMods) {
|
1678 |
6c41cae2
|
p.kelbert
|
logger.info("extractSynonyms");
|
1679 |
|
|
//System.out.println("extractSynonyms for: "+acceptedTaxon);
|
1680 |
285d1929
|
p.kelbert
|
Taxon ttmp = (Taxon) importer.getTaxonService().find(acceptedTaxon.getUuid());
|
1681 |
|
|
if (ttmp != null) {
|
1682 |
|
|
acceptedTaxon = CdmBase.deproxy(ttmp,Taxon.class);
|
1683 |
|
|
}
|
1684 |
|
|
else{
|
1685 |
|
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
1686 |
|
|
}
|
1687 |
|
|
NodeList children = synonyms.getChildNodes();
|
1688 |
|
|
TaxonNameBase nameToBeFilled = null;
|
1689 |
1de96f38
|
p.kelbert
|
List<MyName> names = new ArrayList<MyName>();
|
1690 |
|
|
|
1691 |
|
|
if(synonyms.getNodeName().equalsIgnoreCase("tax:name")){
|
1692 |
|
|
MyName myName;
|
1693 |
|
|
try {
|
1694 |
6b2c2013
|
p.kelbert
|
myName = extractScientificNameSynonym(synonyms,refMods);
|
1695 |
1de96f38
|
p.kelbert
|
names.add(myName);
|
1696 |
|
|
} catch (TransformerFactoryConfigurationError e) {
|
1697 |
|
|
logger.warn(e);
|
1698 |
|
|
} catch (TransformerException e) {
|
1699 |
|
|
logger.warn(e);
|
1700 |
|
|
}
|
1701 |
|
|
}
|
1702 |
285d1929
|
p.kelbert
|
|
1703 |
|
|
|
1704 |
|
|
for (int i=0;i<children.getLength();i++){
|
1705 |
|
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
1706 |
|
|
NodeList tmp = children.item(i).getChildNodes();
|
1707 |
|
|
// String fullContent = children.item(i).getTextContent();
|
1708 |
|
|
for (int j=0; j< tmp.getLength();j++){
|
1709 |
|
|
if(tmp.item(j).getNodeName().equalsIgnoreCase("tax:name")){
|
1710 |
1de96f38
|
p.kelbert
|
MyName myName;
|
1711 |
285d1929
|
p.kelbert
|
try {
|
1712 |
6b2c2013
|
p.kelbert
|
myName = extractScientificNameSynonym(tmp.item(j),refMods);
|
1713 |
1de96f38
|
p.kelbert
|
names.add(myName);
|
1714 |
285d1929
|
p.kelbert
|
} catch (TransformerFactoryConfigurationError e) {
|
1715 |
|
|
logger.warn(e);
|
1716 |
|
|
} catch (TransformerException e) {
|
1717 |
|
|
logger.warn(e);
|
1718 |
|
|
}
|
1719 |
|
|
|
1720 |
|
|
}
|
1721 |
|
|
}
|
1722 |
|
|
}
|
1723 |
1de96f38
|
p.kelbert
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:name")){
|
1724 |
|
|
MyName myName;
|
1725 |
|
|
try {
|
1726 |
6b2c2013
|
p.kelbert
|
myName = extractScientificNameSynonym(children.item(i),refMods);
|
1727 |
1de96f38
|
p.kelbert
|
names.add(myName);
|
1728 |
|
|
} catch (TransformerFactoryConfigurationError e) {
|
1729 |
|
|
logger.warn(e);
|
1730 |
|
|
} catch (TransformerException e) {
|
1731 |
|
|
logger.warn(e);
|
1732 |
|
|
}
|
1733 |
|
|
|
1734 |
|
|
}
|
1735 |
285d1929
|
p.kelbert
|
}
|
1736 |
1de96f38
|
p.kelbert
|
NomenclaturalStatusType statusType = null;
|
1737 |
6c41cae2
|
p.kelbert
|
//System.out.println("names: "+names);
|
1738 |
1de96f38
|
p.kelbert
|
for(MyName name:names){
|
1739 |
6c41cae2
|
p.kelbert
|
//System.out.println("HANDLE NAME "+name);
|
1740 |
136e16b1
|
Andreas Müller
|
|
1741 |
1de96f38
|
p.kelbert
|
statusType = null;
|
1742 |
|
|
|
1743 |
6b2c2013
|
p.kelbert
|
nameToBeFilled = name.getTaxonNameBase();
|
1744 |
2b5a7979
|
p.kelbert
|
|
1745 |
6b2c2013
|
p.kelbert
|
Synonym synonym = name.getSyno();
|
1746 |
2b5a7979
|
p.kelbert
|
/* INonViralNameParser parser = NonViralNameParserImpl.NewInstance();
|
1747 |
1de96f38
|
p.kelbert
|
nameToBeFilled = parser.parseFullName(name.getName(), nomenclaturalCode, name.getRank());
|
1748 |
285d1929
|
p.kelbert
|
if (nameToBeFilled.hasProblem() &&
|
1749 |
136e16b1
|
Andreas Müller
|
!((nameToBeFilled.getParsingProblems().size()==1) && nameToBeFilled.getParsingProblems().contains(ParserProblem.CheckRank)) ) {
|
1750 |
285d1929
|
p.kelbert
|
// if (nameToBeFilled.hasProblem() && nameToBeFilled.getParsingProblems().contains(ParserProblem.UnparsableNamePart)){
|
1751 |
2b5a7979
|
p.kelbert
|
addProblemNameToFile(name.getName(),"",nomenclaturalCode,name.getRank());
|
1752 |
|
|
nameToBeFilled = solveNameProblem(name.getOriginalName(), name.getName(), parser,name.getAuthor(), name.getRank());
|
1753 |
285d1929
|
p.kelbert
|
}
|
1754 |
1de96f38
|
p.kelbert
|
nameToBeFilled = getTaxonNameBase(nameToBeFilled,nametosave,statusType);
|
1755 |
2b5a7979
|
p.kelbert
|
*/
|
1756 |
1de96f38
|
p.kelbert
|
if (!name.getIdentifier().isEmpty() && (name.getIdentifier().length()>2)){
|
1757 |
|
|
setLSID(name.getIdentifier(), synonym);
|
1758 |
136e16b1
|
Andreas Müller
|
}
|
1759 |
|
|
|
1760 |
1de96f38
|
p.kelbert
|
Set<Synonym> synonymsSet= acceptedTaxon.getSynonyms();
|
1761 |
2b5a7979
|
p.kelbert
|
// System.out.println(synonym.getName()+" -- "+synonym.getSec());
|
1762 |
1de96f38
|
p.kelbert
|
boolean synoExist = false;
|
1763 |
|
|
for (Synonym syn: synonymsSet){
|
1764 |
6c41cae2
|
p.kelbert
|
//System.out.println(syn.getName()+" -- "+syn.getSec());
|
1765 |
1de96f38
|
p.kelbert
|
boolean a =syn.getName().equals(synonym.getName());
|
1766 |
|
|
boolean b = syn.getSec().equals(synonym.getSec());
|
1767 |
|
|
if (a && b) {
|
1768 |
|
|
synoExist=true;
|
1769 |
|
|
}
|
1770 |
|
|
}
|
1771 |
|
|
if (!synonymsSet.contains(synonym) && ! (synoExist)) {
|
1772 |
6c41cae2
|
p.kelbert
|
//System.out.println("SYNONYM");
|
1773 |
2b5a7979
|
p.kelbert
|
sourceHandler.addSource(refMods, synonym);
|
1774 |
|
|
|
1775 |
1de96f38
|
p.kelbert
|
acceptedTaxon.addSynonym(synonym, SynonymRelationshipType.SYNONYM_OF(),refMods, null);
|
1776 |
2b5a7979
|
p.kelbert
|
|
1777 |
1de96f38
|
p.kelbert
|
}
|
1778 |
285d1929
|
p.kelbert
|
}
|
1779 |
2b5a7979
|
p.kelbert
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
1780 |
285d1929
|
p.kelbert
|
}
|
1781 |
|
|
|
1782 |
|
|
|
1783 |
|
|
/**
|
1784 |
|
|
* @param refgroup: the XML nodes
|
1785 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
1786 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
1787 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
1788 |
|
|
* @param refMods: the current reference extracted from the MODS
|
1789 |
|
|
* @return the acceptedTaxon (why?)
|
1790 |
1de96f38
|
p.kelbert
|
* handle cases where the bibref are inside <p> and outside
|
1791 |
285d1929
|
p.kelbert
|
*/
|
1792 |
2b5a7979
|
p.kelbert
|
@SuppressWarnings({ "rawtypes" })
|
1793 |
285d1929
|
p.kelbert
|
private Taxon extractReferences(Node refgroup, List<TaxonNameBase> nametosave, Taxon acceptedTaxon, Reference<?> refMods) {
|
1794 |
6c41cae2
|
p.kelbert
|
logger.info("extractReferences");
|
1795 |
285d1929
|
p.kelbert
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
1796 |
|
|
|
1797 |
|
|
NodeList children = refgroup.getChildNodes();
|
1798 |
2b5a7979
|
p.kelbert
|
NonViralName<?> nameToBeFilled = getNonViralNameAccNomenclature();
|
1799 |
1de96f38
|
p.kelbert
|
|
1800 |
6c41cae2
|
p.kelbert
|
ReferenceBuilder refBuild = new ReferenceBuilder(sourceHandler);
|
1801 |
285d1929
|
p.kelbert
|
for (int i=0;i<children.getLength();i++){
|
1802 |
1de96f38
|
p.kelbert
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:bibref")){
|
1803 |
|
|
String ref = children.item(i).getTextContent().trim();
|
1804 |
|
|
refBuild.builReference(ref, treatmentMainName, nomenclaturalCode, acceptedTaxon, refMods);
|
1805 |
|
|
if (!refBuild.isFoundBibref()){
|
1806 |
2b5a7979
|
p.kelbert
|
extractReferenceRawText(children.item(i).getChildNodes(), nameToBeFilled, refMods, acceptedTaxon);
|
1807 |
1de96f38
|
p.kelbert
|
}
|
1808 |
|
|
}
|
1809 |
|
|
|
1810 |
285d1929
|
p.kelbert
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:p")){
|
1811 |
|
|
NodeList references = children.item(i).getChildNodes();
|
1812 |
2b5a7979
|
p.kelbert
|
String descr="";
|
1813 |
285d1929
|
p.kelbert
|
for (int j=0;j<references.getLength();j++){
|
1814 |
|
|
if(references.item(j).getNodeName().equalsIgnoreCase("tax:bibref")){
|
1815 |
|
|
String ref = references.item(j).getTextContent().trim();
|
1816 |
1de96f38
|
p.kelbert
|
refBuild.builReference(ref, treatmentMainName, nomenclaturalCode, acceptedTaxon, refMods);
|
1817 |
|
|
}
|
1818 |
2b5a7979
|
p.kelbert
|
else
|
1819 |
|
|
if (references.item(j).getNodeName().equalsIgnoreCase("#text")
|
1820 |
|
|
&& !references.item(j).getTextContent().trim().isEmpty()){
|
1821 |
|
|
descr += references.item(j).getTextContent().trim();
|
1822 |
|
|
}
|
1823 |
|
|
|
1824 |
1de96f38
|
p.kelbert
|
}
|
1825 |
|
|
if (!refBuild.isFoundBibref()){
|
1826 |
2b5a7979
|
p.kelbert
|
//if it's not tagged, put it as row information.
|
1827 |
|
|
// extractReferenceRawText(references, nameToBeFilled, nametosave, refMods, acceptedTaxon);
|
1828 |
|
|
//then put it as a not markup feature if not empty
|
1829 |
6c41cae2
|
p.kelbert
|
if (!stringIsEmpty(descr.trim())){
|
1830 |
2b5a7979
|
p.kelbert
|
Feature currentFeature= getNotMarkedUpFeatureObject();
|
1831 |
6c41cae2
|
p.kelbert
|
setParticularDescription(descr.trim(),acceptedTaxon,acceptedTaxon, refMods,currentFeature);
|
1832 |
2b5a7979
|
p.kelbert
|
}
|
1833 |
1de96f38
|
p.kelbert
|
}
|
1834 |
|
|
}
|
1835 |
|
|
}
|
1836 |
|
|
// importer.getClassificationService().saveOrUpdate(classification);
|
1837 |
|
|
return acceptedTaxon;
|
1838 |
285d1929
|
p.kelbert
|
|
1839 |
1de96f38
|
p.kelbert
|
}
|
1840 |
285d1929
|
p.kelbert
|
|
1841 |
2b5a7979
|
p.kelbert
|
/**
|
1842 |
|
|
* get the non viral name according to the current nomenclature
|
1843 |
|
|
* @return
|
1844 |
|
|
*/
|
1845 |
|
|
private NonViralName<?> getNonViralNameAccNomenclature() {
|
1846 |
6c41cae2
|
p.kelbert
|
//logger.info("getNonViralNameAccNomenclature");
|
1847 |
2b5a7979
|
p.kelbert
|
NonViralName<?> nameToBeFilled = null;
|
1848 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNAFP)){
|
1849 |
|
|
nameToBeFilled = BotanicalName.NewInstance(null);
|
1850 |
|
|
}
|
1851 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICZN)){
|
1852 |
|
|
nameToBeFilled = ZoologicalName.NewInstance(null);
|
1853 |
|
|
}
|
1854 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNB)){
|
1855 |
|
|
nameToBeFilled = BacterialName.NewInstance(null);
|
1856 |
|
|
}
|
1857 |
|
|
return nameToBeFilled;
|
1858 |
|
|
}
|
1859 |
|
|
|
1860 |
|
|
/**
|
1861 |
|
|
* @return the feature object for the category "not marked up"
|
1862 |
|
|
*/
|
1863 |
|
|
@SuppressWarnings("rawtypes")
|
1864 |
|
|
private Feature getNotMarkedUpFeatureObject() {
|
1865 |
6c41cae2
|
p.kelbert
|
logger.info("getNotMarkedUpFeatureObject");
|
1866 |
fdf9eec2
|
Andreas Müller
|
List<Feature> features = importer.getTermService().list(Feature.class, null,null,null,null);
|
1867 |
2b5a7979
|
p.kelbert
|
Feature currentFeature =null;
|
1868 |
fdf9eec2
|
Andreas Müller
|
for (Feature feat: features){
|
1869 |
|
|
String tmpF = feat.getTitleCache();
|
1870 |
2b5a7979
|
p.kelbert
|
if (tmpF.equalsIgnoreCase(notMarkedUp)) {
|
1871 |
fdf9eec2
|
Andreas Müller
|
currentFeature=feat;
|
1872 |
2b5a7979
|
p.kelbert
|
}
|
1873 |
|
|
}
|
1874 |
|
|
if (currentFeature == null) {
|
1875 |
|
|
currentFeature=Feature.NewInstance(notMarkedUp, notMarkedUp, notMarkedUp);
|
1876 |
|
|
currentFeature.setUuid(NotMarkedUpUUID);
|
1877 |
|
|
importer.getTermService().saveOrUpdate(currentFeature);
|
1878 |
|
|
}
|
1879 |
|
|
return currentFeature;
|
1880 |
|
|
}
|
1881 |
|
|
|
1882 |
1de96f38
|
p.kelbert
|
/**
|
1883 |
|
|
* @param references
|
1884 |
|
|
* handle cases where the bibref are inside <p> and outside
|
1885 |
|
|
*/
|
1886 |
|
|
@SuppressWarnings("rawtypes")
|
1887 |
2b5a7979
|
p.kelbert
|
private void extractReferenceRawText(NodeList references, NonViralName<?> nameToBeFilled, Reference<?> refMods,
|
1888 |
|
|
Taxon acceptedTaxon) {
|
1889 |
6c41cae2
|
p.kelbert
|
logger.info("extractReferenceRawText");
|
1890 |
1de96f38
|
p.kelbert
|
String refString="";
|
1891 |
|
|
NomenclaturalStatusType statusType = null;
|
1892 |
6b2c2013
|
p.kelbert
|
currentMyName= new MyName(true);
|
1893 |
1de96f38
|
p.kelbert
|
for (int j=0;j<references.getLength();j++){
|
1894 |
2b5a7979
|
p.kelbert
|
acceptedTaxon=CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
1895 |
1de96f38
|
p.kelbert
|
//no bibref tag inside
|
1896 |
2b5a7979
|
p.kelbert
|
// System.out.println("references.item(j).getNodeName()"+references.item(j).getNodeName());
|
1897 |
1de96f38
|
p.kelbert
|
if (references.item(j).getNodeName().equalsIgnoreCase("tax:name")){
|
1898 |
285d1929
|
p.kelbert
|
|
1899 |
1de96f38
|
p.kelbert
|
try {
|
1900 |
2b5a7979
|
p.kelbert
|
currentMyName = extractScientificName(references.item(j),refMods);
|
1901 |
1de96f38
|
p.kelbert
|
// if (myName.getNewName().isEmpty()) {
|
1902 |
|
|
// name=myName.getOriginalName()+"---"+myName.getRank()+"---"+myName.getIdentifier()+"---"+myName.getStatus();
|
1903 |
|
|
// } else {
|
1904 |
|
|
// name=myName.getNewName()+"---"+myName.getRank()+"---"+myName.getIdentifier()+"---"+myName.getStatus();
|
1905 |
|
|
// }
|
1906 |
|
|
} catch (TransformerFactoryConfigurationError e) {
|
1907 |
|
|
logger.warn(e);
|
1908 |
|
|
} catch (TransformerException e) {
|
1909 |
|
|
logger.warn(e);
|
1910 |
|
|
}
|
1911 |
285d1929
|
p.kelbert
|
|
1912 |
1de96f38
|
p.kelbert
|
// name=name.trim();
|
1913 |
|
|
}
|
1914 |
|
|
if (references.item(j).getNodeName().equalsIgnoreCase("#text")){
|
1915 |
|
|
refString = references.item(j).getTextContent().trim();
|
1916 |
|
|
}
|
1917 |
|
|
if(references.item(j).getNodeName().equalsIgnoreCase("#text") && !references.item(j).getTextContent().trim().isEmpty()){
|
1918 |
|
|
//
|
1919 |
|
|
statusType = null;
|
1920 |
2b5a7979
|
p.kelbert
|
if (!currentMyName.getStatus().isEmpty()){
|
1921 |
1de96f38
|
p.kelbert
|
try {
|
1922 |
2b5a7979
|
p.kelbert
|
statusType = nomStatusString2NomStatus(currentMyName.getStatus());
|
1923 |
1de96f38
|
p.kelbert
|
} catch (UnknownCdmTypeException e) {
|
1924 |
2b5a7979
|
p.kelbert
|
addProblematicStatusToFile(currentMyName.getStatus());
|
1925 |
1de96f38
|
p.kelbert
|
logger.warn("Problem with status");
|
1926 |
|
|
}
|
1927 |
|
|
}
|
1928 |
285d1929
|
p.kelbert
|
|
1929 |
2b5a7979
|
p.kelbert
|
|
1930 |
|
|
/*INonViralNameParser parser = NonViralNameParserImpl.NewInstance();*/
|
1931 |
1de96f38
|
p.kelbert
|
String fullLineRefName = references.item(j).getTextContent().trim();
|
1932 |
2b5a7979
|
p.kelbert
|
int nameOrRefOrOther=2;
|
1933 |
1de96f38
|
p.kelbert
|
nameOrRefOrOther=askIfNameContained(fullLineRefName);
|
1934 |
2b5a7979
|
p.kelbert
|
// System.out.println("NAMEORREFOR?? "+nameOrRefOrOther);
|
1935 |
1de96f38
|
p.kelbert
|
if (nameOrRefOrOther==0){
|
1936 |
2b5a7979
|
p.kelbert
|
/*TaxonNameBase nameTBF = parser.parseFullName(fullLineRefName, nomenclaturalCode, Rank.UNKNOWN_RANK());
|
1937 |
1de96f38
|
p.kelbert
|
if (nameTBF.hasProblem() &&
|
1938 |
|
|
!((nameTBF.getParsingProblems().size()==1) && nameTBF.getParsingProblems().contains(ParserProblem.CheckRank)) ) {
|
1939 |
2b5a7979
|
p.kelbert
|
addProblemNameToFile(fullLineRefName,"",nomenclaturalCode,Rank.UNKNOWN_RANK());
|
1940 |
|
|
nameTBF=solveNameProblem(fullLineRefName, fullLineRefName,parser,currentMyName.getAuthor(), currentMyName.getRank());
|
1941 |
1de96f38
|
p.kelbert
|
}
|
1942 |
|
|
nameTBF = getTaxonNameBase(nameTBF,nametosave,statusType);
|
1943 |
2b5a7979
|
p.kelbert
|
*/
|
1944 |
|
|
TaxonNameBase nameTBF = currentMyName.getTaxonNameBase();
|
1945 |
|
|
Synonym synonym = null;
|
1946 |
|
|
if (!currentMyName.getStatus().isEmpty()){
|
1947 |
|
|
try {
|
1948 |
|
|
statusType = nomStatusString2NomStatus(currentMyName.getStatus());
|
1949 |
|
|
nameToBeFilled.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
1950 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
1951 |
|
|
} catch (UnknownCdmTypeException e) {
|
1952 |
|
|
addProblematicStatusToFile(currentMyName.getStatus());
|
1953 |
|
|
logger.warn("Problem with status");
|
1954 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
1955 |
|
|
synonym.setAppendedPhrase(currentMyName.getStatus());
|
1956 |
|
|
}
|
1957 |
|
|
}
|
1958 |
|
|
else{
|
1959 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
1960 |
|
|
}
|
1961 |
1de96f38
|
p.kelbert
|
|
1962 |
|
|
Set<Synonym> synonymsSet= acceptedTaxon.getSynonyms();
|
1963 |
2b5a7979
|
p.kelbert
|
// System.out.println(synonym.getName()+" -- "+synonym.getSec());
|
1964 |
1de96f38
|
p.kelbert
|
boolean synoExist = false;
|
1965 |
|
|
for (Synonym syn: synonymsSet){
|
1966 |
2b5a7979
|
p.kelbert
|
// System.out.println(syn.getName()+" -- "+syn.getSec());
|
1967 |
1de96f38
|
p.kelbert
|
boolean a =syn.getName().equals(synonym.getName());
|
1968 |
|
|
boolean b = syn.getSec().equals(synonym.getSec());
|
1969 |
|
|
if (a && b) {
|
1970 |
|
|
synoExist=true;
|
1971 |
285d1929
|
p.kelbert
|
}
|
1972 |
1de96f38
|
p.kelbert
|
}
|
1973 |
|
|
if (!synonymsSet.contains(synonym) && ! (synoExist)) {
|
1974 |
2b5a7979
|
p.kelbert
|
sourceHandler.addSource(refMods, synonym);
|
1975 |
|
|
|
1976 |
1de96f38
|
p.kelbert
|
acceptedTaxon.addSynonym(synonym, SynonymRelationshipType.SYNONYM_OF(),refMods, null);
|
1977 |
285d1929
|
p.kelbert
|
}
|
1978 |
|
|
}
|
1979 |
|
|
|
1980 |
1de96f38
|
p.kelbert
|
if (nameOrRefOrOther==1){
|
1981 |
|
|
Reference<?> re = ReferenceFactory.newGeneric();
|
1982 |
6e861bce
|
Andreas Müller
|
re.setTitleCache(fullLineRefName, true);
|
1983 |
1de96f38
|
p.kelbert
|
|
1984 |
2b5a7979
|
p.kelbert
|
/* TaxonNameBase nameTBF = parser.parseFullName(currentMyName.getName(), nomenclaturalCode, currentMyName.getRank());
|
1985 |
1de96f38
|
p.kelbert
|
if (nameTBF.hasProblem() &&
|
1986 |
|
|
!((nameTBF.getParsingProblems().size()==1) && nameTBF.getParsingProblems().contains(ParserProblem.CheckRank)) ) {
|
1987 |
2b5a7979
|
p.kelbert
|
addProblemNameToFile(currentMyName.getName(),"",nomenclaturalCode,currentMyName.getRank());
|
1988 |
|
|
nameTBF=solveNameProblem(currentMyName.getName(), currentMyName.getName(),parser,currentMyName.getAuthor(), currentMyName.getRank());
|
1989 |
1de96f38
|
p.kelbert
|
}
|
1990 |
|
|
nameTBF = getTaxonNameBase(nameTBF,nametosave,statusType);
|
1991 |
2b5a7979
|
p.kelbert
|
*/
|
1992 |
|
|
TaxonNameBase nameTBF = currentMyName.getTaxonNameBase();
|
1993 |
|
|
Synonym synonym = null;
|
1994 |
|
|
if (!currentMyName.getStatus().isEmpty()){
|
1995 |
|
|
try {
|
1996 |
|
|
statusType = nomStatusString2NomStatus(currentMyName.getStatus());
|
1997 |
|
|
nameToBeFilled.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
1998 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
1999 |
|
|
} catch (UnknownCdmTypeException e) {
|
2000 |
|
|
addProblematicStatusToFile(currentMyName.getStatus());
|
2001 |
|
|
logger.warn("Problem with status");
|
2002 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
2003 |
|
|
synonym.setAppendedPhrase(currentMyName.getStatus());
|
2004 |
|
|
}
|
2005 |
|
|
}
|
2006 |
|
|
else{
|
2007 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
2008 |
|
|
}
|
2009 |
1de96f38
|
p.kelbert
|
|
2010 |
|
|
Set<Synonym> synonymsSet= acceptedTaxon.getSynonyms();
|
2011 |
2b5a7979
|
p.kelbert
|
// System.out.println(synonym.getName()+" -- "+synonym.getSec());
|
2012 |
1de96f38
|
p.kelbert
|
boolean synoExist = false;
|
2013 |
|
|
for (Synonym syn: synonymsSet){
|
2014 |
2b5a7979
|
p.kelbert
|
// System.out.println(syn.getName()+" -- "+syn.getSec());
|
2015 |
1de96f38
|
p.kelbert
|
boolean a =syn.getName().equals(synonym.getName());
|
2016 |
|
|
boolean b = syn.getSec().equals(synonym.getSec());
|
2017 |
|
|
if (a && b) {
|
2018 |
|
|
synoExist=true;
|
2019 |
285d1929
|
p.kelbert
|
}
|
2020 |
1de96f38
|
p.kelbert
|
}
|
2021 |
|
|
if (!synonymsSet.contains(synonym) && ! (synoExist)) {
|
2022 |
2b5a7979
|
p.kelbert
|
sourceHandler.addSource(refMods, synonym);
|
2023 |
|
|
|
2024 |
1de96f38
|
p.kelbert
|
acceptedTaxon.addSynonym(synonym, SynonymRelationshipType.SYNONYM_OF(),re, null);
|
2025 |
|
|
}
|
2026 |
136e16b1
|
Andreas Müller
|
|
2027 |
1de96f38
|
p.kelbert
|
}
|
2028 |
285d1929
|
p.kelbert
|
|
2029 |
1de96f38
|
p.kelbert
|
|
2030 |
2b5a7979
|
p.kelbert
|
if (!currentMyName.getIdentifier().isEmpty() && (currentMyName.getIdentifier().length()>2)){
|
2031 |
|
|
setLSID(currentMyName.getIdentifier(), acceptedTaxon);
|
2032 |
|
|
}
|
2033 |
1de96f38
|
p.kelbert
|
}
|
2034 |
136e16b1
|
Andreas Müller
|
|
2035 |
2b5a7979
|
p.kelbert
|
if(!currentMyName.getName().isEmpty()){
|
2036 |
6c41cae2
|
p.kelbert
|
//logger.info("acceptedTaxon and name: *"+acceptedTaxon.getTitleCache()+"*, *"+currentMyName.getName()+"*");
|
2037 |
2b5a7979
|
p.kelbert
|
if (acceptedTaxon.getTitleCache().split("sec")[0].trim().equalsIgnoreCase(currentMyName.getName().trim())){
|
2038 |
1de96f38
|
p.kelbert
|
Reference<?> refS = ReferenceFactory.newGeneric();
|
2039 |
|
|
refS.setTitleCache(refString, true);
|
2040 |
|
|
// TaxonDescription td =importer.getTaxonDescription(acceptedTaxon, false, true);
|
2041 |
|
|
// acceptedTaxon.addDescription(td);
|
2042 |
|
|
// acceptedTaxon.addSource(refSource);
|
2043 |
|
|
//
|
2044 |
|
|
// TextData textData = TextData.NewInstance(Feature.CITATION());
|
2045 |
|
|
//
|
2046 |
|
|
// textData.addSource(null, null, refS, null);
|
2047 |
|
|
// td.addElement(textData);
|
2048 |
|
|
// td.addSource(refSource);
|
2049 |
|
|
// importer.getDescriptionService().saveOrUpdate(td);
|
2050 |
136e16b1
|
Andreas Müller
|
|
2051 |
|
|
|
2052 |
2b5a7979
|
p.kelbert
|
if (!currentMyName.getIdentifier().isEmpty() && (currentMyName.getIdentifier().length()>2)){
|
2053 |
|
|
setLSID(currentMyName.getIdentifier(), acceptedTaxon);
|
2054 |
285d1929
|
p.kelbert
|
|
2055 |
1de96f38
|
p.kelbert
|
}
|
2056 |
136e16b1
|
Andreas Müller
|
|
2057 |
1de96f38
|
p.kelbert
|
acceptedTaxon.getName().setNomenclaturalReference(refS);
|
2058 |
|
|
}
|
2059 |
|
|
else{
|
2060 |
2b5a7979
|
p.kelbert
|
/* INonViralNameParser parser = NonViralNameParserImpl.NewInstance();
|
2061 |
|
|
TaxonNameBase nameTBF = parser.parseFullName(currentMyName.getName(), nomenclaturalCode, currentMyName.getRank());
|
2062 |
1de96f38
|
p.kelbert
|
if (nameTBF.hasProblem() &&
|
2063 |
|
|
!((nameTBF.getParsingProblems().size()==1) && nameTBF.getParsingProblems().contains(ParserProblem.CheckRank)) ) {
|
2064 |
|
|
// if (nameToBeFilled.hasProblem() && nameToBeFilled.getParsingProblems().contains(ParserProblem.UnparsableNamePart)){
|
2065 |
2b5a7979
|
p.kelbert
|
addProblemNameToFile(currentMyName.getName(),"",nomenclaturalCode,currentMyName.getRank());
|
2066 |
|
|
nameTBF=solveNameProblem(currentMyName.getOriginalName(), currentMyName.getName(),parser,currentMyName.getAuthor(), currentMyName.getRank());
|
2067 |
1de96f38
|
p.kelbert
|
}
|
2068 |
|
|
nameTBF = getTaxonNameBase(nameTBF,nametosave,statusType);
|
2069 |
2b5a7979
|
p.kelbert
|
*/
|
2070 |
|
|
TaxonNameBase nameTBF = currentMyName.getTaxonNameBase();
|
2071 |
|
|
Synonym synonym = null;
|
2072 |
|
|
if (!currentMyName.getStatus().isEmpty()){
|
2073 |
|
|
try {
|
2074 |
|
|
statusType = nomStatusString2NomStatus(currentMyName.getStatus());
|
2075 |
|
|
nameToBeFilled.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
2076 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
2077 |
|
|
} catch (UnknownCdmTypeException e) {
|
2078 |
|
|
addProblematicStatusToFile(currentMyName.getStatus());
|
2079 |
|
|
logger.warn("Problem with status");
|
2080 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
2081 |
|
|
synonym.setAppendedPhrase(currentMyName.getStatus());
|
2082 |
1de96f38
|
p.kelbert
|
}
|
2083 |
|
|
}
|
2084 |
2b5a7979
|
p.kelbert
|
else{
|
2085 |
|
|
synonym = Synonym.NewInstance(nameTBF, refMods);
|
2086 |
|
|
}
|
2087 |
|
|
|
2088 |
|
|
|
2089 |
|
|
if (!currentMyName.getIdentifier().isEmpty() && (currentMyName.getIdentifier().length()>2)){
|
2090 |
|
|
setLSID(currentMyName.getIdentifier(), synonym);
|
2091 |
|
|
}
|
2092 |
1de96f38
|
p.kelbert
|
|
2093 |
|
|
Set<Synonym> synonymsSet= acceptedTaxon.getSynonyms();
|
2094 |
2b5a7979
|
p.kelbert
|
// System.out.println(synonym.getName()+" -- "+synonym.getSec());
|
2095 |
1de96f38
|
p.kelbert
|
boolean synoExist = false;
|
2096 |
|
|
for (Synonym syn: synonymsSet){
|
2097 |
2b5a7979
|
p.kelbert
|
// System.out.println(syn.getName()+" -- "+syn.getSec());
|
2098 |
1de96f38
|
p.kelbert
|
boolean a =syn.getName().equals(synonym.getName());
|
2099 |
|
|
boolean b = syn.getSec().equals(synonym.getSec());
|
2100 |
|
|
if (a && b) {
|
2101 |
|
|
synoExist=true;
|
2102 |
285d1929
|
p.kelbert
|
}
|
2103 |
|
|
}
|
2104 |
1de96f38
|
p.kelbert
|
if (!synonymsSet.contains(synonym) && ! (synoExist)) {
|
2105 |
2b5a7979
|
p.kelbert
|
sourceHandler.addSource(refMods, synonym);
|
2106 |
|
|
|
2107 |
1de96f38
|
p.kelbert
|
acceptedTaxon.addSynonym(synonym, SynonymRelationshipType.SYNONYM_OF(),refMods, null);
|
2108 |
|
|
}
|
2109 |
285d1929
|
p.kelbert
|
}
|
2110 |
|
|
}
|
2111 |
2b5a7979
|
p.kelbert
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
2112 |
285d1929
|
p.kelbert
|
}
|
2113 |
|
|
}
|
2114 |
|
|
|
2115 |
1de96f38
|
p.kelbert
|
|
2116 |
|
|
|
2117 |
136e16b1
|
Andreas Müller
|
/**
|
2118 |
|
|
* @param identifier
|
2119 |
|
|
* @param acceptedTaxon
|
2120 |
|
|
*/
|
2121 |
2b5a7979
|
p.kelbert
|
@SuppressWarnings("rawtypes")
|
2122 |
136e16b1
|
Andreas Müller
|
private void setLSID(String identifier, TaxonBase<?> taxon) {
|
2123 |
6c41cae2
|
p.kelbert
|
//logger.info("setLSID");
|
2124 |
2b5a7979
|
p.kelbert
|
// boolean lsidok=false;
|
2125 |
136e16b1
|
Andreas Müller
|
String id = identifier.split("__")[0];
|
2126 |
|
|
String source = identifier.split("__")[1];
|
2127 |
|
|
if (id.indexOf("lsid")>-1){
|
2128 |
|
|
try {
|
2129 |
|
|
LSID lsid = new LSID(id);
|
2130 |
|
|
taxon.setLsid(lsid);
|
2131 |
2b5a7979
|
p.kelbert
|
// lsidok=true;
|
2132 |
136e16b1
|
Andreas Müller
|
} catch (MalformedLSIDException e) {
|
2133 |
|
|
logger.warn("Malformed LSID");
|
2134 |
|
|
}
|
2135 |
|
|
|
2136 |
|
|
}
|
2137 |
2b5a7979
|
p.kelbert
|
|
2138 |
6c41cae2
|
p.kelbert
|
//logger.info("search reference for LSID");
|
2139 |
2b5a7979
|
p.kelbert
|
// if ((id.indexOf("lsid")<0) || !lsidok){
|
2140 |
|
|
//ADD ORIGINAL SOURCE ID EVEN IF LSID
|
2141 |
|
|
Reference<?> re = null;
|
2142 |
6c41cae2
|
p.kelbert
|
Pager<Reference> references = importer.getReferenceService().findByTitle(Reference.class, source, MatchMode.EXACT, null, 1, null, null, null);
|
2143 |
|
|
if( references !=null && references.getCount()>0){
|
2144 |
|
|
re=references.getRecords().get(0);
|
2145 |
2b5a7979
|
p.kelbert
|
}
|
2146 |
6c41cae2
|
p.kelbert
|
//logger.info("search reference for LSID-end");
|
2147 |
2b5a7979
|
p.kelbert
|
if(re == null){
|
2148 |
|
|
re = ReferenceFactory.newGeneric();
|
2149 |
6e861bce
|
Andreas Müller
|
re.setTitleCache(source, true);
|
2150 |
2b5a7979
|
p.kelbert
|
importer.getReferenceService().saveOrUpdate(re);
|
2151 |
|
|
}
|
2152 |
|
|
re=CdmBase.deproxy(re, Reference.class);
|
2153 |
|
|
|
2154 |
6c41cae2
|
p.kelbert
|
//logger.info("search source for LSID");
|
2155 |
2b5a7979
|
p.kelbert
|
Set<IdentifiableSource> sources = taxon.getSources();
|
2156 |
|
|
boolean lsidinsource=false;
|
2157 |
|
|
boolean urlinsource=false;
|
2158 |
|
|
for (IdentifiableSource src:sources){
|
2159 |
|
|
if (id.equalsIgnoreCase(src.getIdInSource()) && re.getTitleCache().equals(src.getCitation().getTitleCache())) {
|
2160 |
|
|
lsidinsource=true;
|
2161 |
|
|
}
|
2162 |
|
|
if (src.getIdInSource() == null && re.getTitleCache().equals(sourceUrlRef.getTitleCache())) {
|
2163 |
|
|
urlinsource=true;
|
2164 |
|
|
}
|
2165 |
|
|
}
|
2166 |
|
|
if(!lsidinsource) {
|
2167 |
|
|
taxon.addSource(OriginalSourceType.Import, id,null,re,null);
|
2168 |
|
|
}
|
2169 |
|
|
if(!urlinsource)
|
2170 |
|
|
{
|
2171 |
6b2c2013
|
p.kelbert
|
sourceUrlRef=CdmBase.deproxy(sourceUrlRef, Reference.class);
|
2172 |
2b5a7979
|
p.kelbert
|
taxon.addSource(OriginalSourceType.Import, null,null,sourceUrlRef,null);
|
2173 |
|
|
// }
|
2174 |
136e16b1
|
Andreas Müller
|
}
|
2175 |
|
|
|
2176 |
|
|
}
|
2177 |
|
|
|
2178 |
285d1929
|
p.kelbert
|
/**
|
2179 |
|
|
* try to solve a parsing problem for a scientific name
|
2180 |
|
|
* @param original : the name from the OCR document
|
2181 |
|
|
* @param name : the tagged version
|
2182 |
|
|
* @param parser
|
2183 |
|
|
* @return the corrected TaxonNameBase
|
2184 |
|
|
*/
|
2185 |
2b5a7979
|
p.kelbert
|
/* @SuppressWarnings({ "unchecked", "rawtypes" })
|
2186 |
|
|
private TaxonNameBase<?,?> solveNameProblem(String original, String name, INonViralNameParser parser, String author, Rank rank) {
|
2187 |
285d1929
|
p.kelbert
|
Map<String,String> ato = namesMap.get(original);
|
2188 |
2b5a7979
|
p.kelbert
|
if (ato == null) {
|
2189 |
|
|
ato = namesMap.get(original+" "+author);
|
2190 |
|
|
}
|
2191 |
|
|
|
2192 |
285d1929
|
p.kelbert
|
|
2193 |
2b5a7979
|
p.kelbert
|
if (ato == null && rank.equals(Rank.UNKNOWN_RANK())){
|
2194 |
285d1929
|
p.kelbert
|
rank=askForRank(original, Rank.UNKNOWN_RANK(), nomenclaturalCode);
|
2195 |
2b5a7979
|
p.kelbert
|
}
|
2196 |
|
|
if (ato != null && rank.equals(Rank.UNKNOWN_RANK())){
|
2197 |
285d1929
|
p.kelbert
|
rank = getRank(ato);
|
2198 |
|
|
}
|
2199 |
2b5a7979
|
p.kelbert
|
// TaxonNameBase<?,?> nameTBF = parser.parseFullName(name, nomenclaturalCode, rank);
|
2200 |
|
|
TaxonNameBase<?,?> nameTBF = parser.parseSimpleName(name, nomenclaturalCode, rank);
|
2201 |
285d1929
|
p.kelbert
|
// logger.info("RANK: "+rank);
|
2202 |
|
|
int retry=0;
|
2203 |
2b5a7979
|
p.kelbert
|
List<ParserProblem> problems = nameTBF.getParsingProblems();
|
2204 |
|
|
for (ParserProblem pb:problems) {
|
2205 |
|
|
System.out.println(pb.toString());
|
2206 |
|
|
}
|
2207 |
136e16b1
|
Andreas Müller
|
while (nameTBF.hasProblem() && (retry <1) && !((nameTBF.getParsingProblems().size()==1) && nameTBF.getParsingProblems().contains(ParserProblem.CheckRank))){
|
2208 |
2b5a7979
|
p.kelbert
|
addProblemNameToFile(name,author,nomenclaturalCode,rank);
|
2209 |
|
|
String fullname=name;
|
2210 |
|
|
if(! skippQuestion) {
|
2211 |
|
|
fullname = getFullReference(name,nameTBF.getParsingProblems());
|
2212 |
|
|
}
|
2213 |
59c2c696
|
Andreas Müller
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNAFP)){
|
2214 |
285d1929
|
p.kelbert
|
nameTBF = BotanicalName.NewInstance(null);
|
2215 |
|
|
}
|
2216 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICZN)){
|
2217 |
|
|
nameTBF = ZoologicalName.NewInstance(null);
|
2218 |
|
|
}
|
2219 |
136e16b1
|
Andreas Müller
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNB)){
|
2220 |
|
|
nameTBF= BacterialName.NewInstance(null);
|
2221 |
|
|
}
|
2222 |
285d1929
|
p.kelbert
|
parser.parseReferencedName(nameTBF, fullname, rank, false);
|
2223 |
|
|
retry++;
|
2224 |
|
|
}
|
2225 |
136e16b1
|
Andreas Müller
|
if (retry == 1){
|
2226 |
2b5a7979
|
p.kelbert
|
if(author != null){
|
2227 |
|
|
if (name.indexOf(author)>-1) {
|
2228 |
|
|
nameTBF = parser.parseSimpleName(name.substring(0, name.indexOf(author)), nomenclaturalCode, rank);
|
2229 |
|
|
} else {
|
2230 |
|
|
nameTBF = parser.parseSimpleName(name, nomenclaturalCode, rank);
|
2231 |
|
|
}
|
2232 |
|
|
if (nameTBF.hasProblem()){
|
2233 |
|
|
if (name.indexOf(author)>-1) {
|
2234 |
|
|
addProblemNameToFile(name.substring(0, name.indexOf(author)),author,nomenclaturalCode,rank);
|
2235 |
|
|
} else {
|
2236 |
|
|
addProblemNameToFile(name,author,nomenclaturalCode,rank);
|
2237 |
|
|
}
|
2238 |
|
|
// System.out.println("TBF still has problems "+nameTBF.hasProblem());
|
2239 |
|
|
problems = nameTBF.getParsingProblems();
|
2240 |
|
|
for (ParserProblem pb:problems) {
|
2241 |
|
|
System.out.println(pb.toString());
|
2242 |
|
|
}
|
2243 |
|
|
nameTBF.setFullTitleCache(name, true);
|
2244 |
|
|
}else{
|
2245 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNAFP)) {
|
2246 |
|
|
((BotanicalName) nameTBF).setAuthorshipCache(currentMyName.getAuthor());
|
2247 |
|
|
}
|
2248 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICZN)) {
|
2249 |
|
|
((ZoologicalName) nameTBF).setAuthorshipCache(currentMyName.getAuthor());
|
2250 |
|
|
}
|
2251 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNB)) {
|
2252 |
|
|
((BacterialName) nameTBF).setAuthorshipCache(currentMyName.getAuthor());
|
2253 |
|
|
}
|
2254 |
|
|
}
|
2255 |
|
|
// logger.info("FULL TITLE CACHE "+name);
|
2256 |
|
|
}else{
|
2257 |
|
|
nameTBF.setFullTitleCache(name, true);
|
2258 |
|
|
}
|
2259 |
285d1929
|
p.kelbert
|
}
|
2260 |
|
|
return nameTBF;
|
2261 |
|
|
}
|
2262 |
|
|
|
2263 |
2b5a7979
|
p.kelbert
|
*/
|
2264 |
|
|
|
2265 |
285d1929
|
p.kelbert
|
/**
|
2266 |
|
|
* @param nomenclatureNode: the XML nodes
|
2267 |
|
|
* @param nametosave: the list of objects to save into the CDM
|
2268 |
|
|
* @param refMods: the current reference extracted from the MODS
|
2269 |
|
|
* @return
|
2270 |
|
|
*/
|
2271 |
6e861bce
|
Andreas Müller
|
@SuppressWarnings({ "rawtypes" })
|
2272 |
2b5a7979
|
p.kelbert
|
private Taxon extractNomenclature(Node nomenclatureNode, List<TaxonNameBase> nametosave, Reference<?> refMods) throws ClassCastException{
|
2273 |
6b2c2013
|
p.kelbert
|
refMods=CdmBase.deproxy(refMods, Reference.class);
|
2274 |
|
|
|
2275 |
6c41cae2
|
p.kelbert
|
logger.info("extractNomenclature");
|
2276 |
285d1929
|
p.kelbert
|
NodeList children = nomenclatureNode.getChildNodes();
|
2277 |
2b5a7979
|
p.kelbert
|
String freetext="";
|
2278 |
|
|
NonViralName<?> nameToBeFilled = null;
|
2279 |
285d1929
|
p.kelbert
|
Taxon acceptedTaxon = null;
|
2280 |
2b5a7979
|
p.kelbert
|
// INonViralNameParser parser = NonViralNameParserImpl.NewInstance();
|
2281 |
285d1929
|
p.kelbert
|
|
2282 |
|
|
// String fullContent = nomenclatureNode.getTextContent();
|
2283 |
1de96f38
|
p.kelbert
|
|
2284 |
|
|
NomenclaturalStatusType statusType = null;
|
2285 |
|
|
for (int i=0;i<children.getLength();i++){
|
2286 |
|
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:status")){
|
2287 |
|
|
String status = children.item(i).getTextContent().trim();
|
2288 |
|
|
if (!status.isEmpty()){
|
2289 |
|
|
try {
|
2290 |
|
|
statusType = nomStatusString2NomStatus(status);
|
2291 |
|
|
} catch (UnknownCdmTypeException e) {
|
2292 |
6b2c2013
|
p.kelbert
|
addProblematicStatusToFile(status);
|
2293 |
1de96f38
|
p.kelbert
|
logger.warn("Problem with status");
|
2294 |
|
|
}
|
2295 |
|
|
}
|
2296 |
|
|
}
|
2297 |
|
|
}
|
2298 |
|
|
|
2299 |
|
|
boolean containsSynonyms=false;
|
2300 |
285d1929
|
p.kelbert
|
for (int i=0;i<children.getLength();i++){
|
2301 |
1de96f38
|
p.kelbert
|
|
2302 |
285d1929
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("#text")) {
|
2303 |
|
|
freetext=children.item(i).getTextContent();
|
2304 |
|
|
}
|
2305 |
136e16b1
|
Andreas Müller
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:collection_event")) {
|
2306 |
2b5a7979
|
p.kelbert
|
// System.out.println("COLLECTION EVENT INSIDE NOMENCLATURE");
|
2307 |
136e16b1
|
Andreas Müller
|
extractMaterialsDirect(children.item(i), acceptedTaxon, refMods, "collection");
|
2308 |
|
|
}
|
2309 |
285d1929
|
p.kelbert
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:name")){
|
2310 |
6c41cae2
|
p.kelbert
|
//System.out.println("HANDLE FIRST NAME OF THE LIST");
|
2311 |
1de96f38
|
p.kelbert
|
if(!containsSynonyms){
|
2312 |
6c41cae2
|
p.kelbert
|
//System.out.println("I : "+i);
|
2313 |
6b2c2013
|
p.kelbert
|
currentMyName = new MyName(false);
|
2314 |
1de96f38
|
p.kelbert
|
try {
|
2315 |
2b5a7979
|
p.kelbert
|
currentMyName = extractScientificName(children.item(i),refMods);
|
2316 |
|
|
treatmentMainName = currentMyName.getNewName();
|
2317 |
|
|
originalTreatmentName = currentMyName.getOriginalName();
|
2318 |
1de96f38
|
p.kelbert
|
|
2319 |
|
|
} catch (TransformerFactoryConfigurationError e1) {
|
2320 |
|
|
logger.warn(e1);
|
2321 |
|
|
} catch (TransformerException e1) {
|
2322 |
|
|
logger.warn(e1);
|
2323 |
285d1929
|
p.kelbert
|
}
|
2324 |
1de96f38
|
p.kelbert
|
|
2325 |
2b5a7979
|
p.kelbert
|
if (currentMyName.getRank().equals(Rank.UNKNOWN_RANK()) || currentMyName.getRank().isLower(configState.getConfig().getMaxRank()) || currentMyName.getRank().equals(configState.getConfig().getMaxRank())){
|
2326 |
1de96f38
|
p.kelbert
|
maxRankRespected=true;
|
2327 |
6b2c2013
|
p.kelbert
|
|
2328 |
2b5a7979
|
p.kelbert
|
nameToBeFilled=currentMyName.getTaxonNameBase();
|
2329 |
|
|
|
2330 |
|
|
// acceptedTaxon = importer.getTaxonService().findBestMatchingTaxon(treatmentMainName);
|
2331 |
|
|
acceptedTaxon=currentMyName.getTaxon();
|
2332 |
6c41cae2
|
p.kelbert
|
//System.out.println("TreatmentName "+treatmentMainName+" - "+acceptedTaxon);
|
2333 |
1de96f38
|
p.kelbert
|
|
2334 |
|
|
|
2335 |
|
|
boolean statusMatch=false;
|
2336 |
|
|
if(acceptedTaxon !=null ){
|
2337 |
2b5a7979
|
p.kelbert
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
2338 |
1de96f38
|
p.kelbert
|
statusMatch=compareStatus(acceptedTaxon, statusType);
|
2339 |
6c41cae2
|
p.kelbert
|
//System.out.println("statusMatch: "+statusMatch);
|
2340 |
285d1929
|
p.kelbert
|
}
|
2341 |
1de96f38
|
p.kelbert
|
if (acceptedTaxon ==null || (acceptedTaxon != null && !statusMatch)){
|
2342 |
2b5a7979
|
p.kelbert
|
|
2343 |
|
|
nameToBeFilled=currentMyName.getTaxonNameBase();
|
2344 |
|
|
if (nameToBeFilled!=null){
|
2345 |
|
|
if (!originalTreatmentName.isEmpty()) {
|
2346 |
|
|
TaxonNameDescription td = TaxonNameDescription.NewInstance();
|
2347 |
6e861bce
|
Andreas Müller
|
td.setTitleCache(originalTreatmentName, true);
|
2348 |
2b5a7979
|
p.kelbert
|
nameToBeFilled.addDescription(td);
|
2349 |
|
|
}
|
2350 |
|
|
|
2351 |
|
|
if(statusType != null) {
|
2352 |
|
|
nameToBeFilled.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
2353 |
|
|
}
|
2354 |
|
|
sourceHandler.addSource(refMods, nameToBeFilled);
|
2355 |
|
|
|
2356 |
|
|
if (nameToBeFilled.getNomenclaturalReference() == null) {
|
2357 |
|
|
acceptedTaxon= new Taxon(nameToBeFilled,refMods);
|
2358 |
6c41cae2
|
p.kelbert
|
//System.out.println("NEW ACCEPTED HERE "+nameToBeFilled);
|
2359 |
2b5a7979
|
p.kelbert
|
}
|
2360 |
|
|
else {
|
2361 |
|
|
acceptedTaxon= new Taxon(nameToBeFilled,(Reference<?>) nameToBeFilled.getNomenclaturalReference() );//TODO TOFIX reference
|
2362 |
6c41cae2
|
p.kelbert
|
//System.out.println("NEW ACCEPTED HERE2 "+nameToBeFilled);
|
2363 |
2b5a7979
|
p.kelbert
|
}
|
2364 |
|
|
|
2365 |
|
|
sourceHandler.addSource(refMods, acceptedTaxon);
|
2366 |
|
|
|
2367 |
|
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
2368 |
|
|
acceptedTaxon.setSec(configState.getConfig().getSecundum());
|
2369 |
6c41cae2
|
p.kelbert
|
//logger.info("SET SECUNDUM "+configState.getConfig().getSecundum());
|
2370 |
|
|
//System.out.println("SET SECUNDUM "+configState.getConfig().getSecundum());
|
2371 |
2b5a7979
|
p.kelbert
|
}
|
2372 |
285d1929
|
p.kelbert
|
|
2373 |
2b5a7979
|
p.kelbert
|
if (!currentMyName.getIdentifier().isEmpty() && (currentMyName.getIdentifier().length()>2)){
|
2374 |
|
|
setLSID(currentMyName.getIdentifier(), acceptedTaxon);
|
2375 |
1de96f38
|
p.kelbert
|
}
|
2376 |
2b5a7979
|
p.kelbert
|
|
2377 |
|
|
|
2378 |
|
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
2379 |
|
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
2380 |
285d1929
|
p.kelbert
|
}
|
2381 |
2b5a7979
|
p.kelbert
|
|
2382 |
285d1929
|
p.kelbert
|
}else{
|
2383 |
1de96f38
|
p.kelbert
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
2384 |
|
|
Set<IdentifiableSource> sources = acceptedTaxon.getSources();
|
2385 |
|
|
boolean sourcelinked=false;
|
2386 |
|
|
for (IdentifiableSource source:sources){
|
2387 |
2b5a7979
|
p.kelbert
|
if (source.getCitation().getTitleCache().equalsIgnoreCase(refMods.getTitleCache())) {
|
2388 |
1de96f38
|
p.kelbert
|
sourcelinked=true;
|
2389 |
|
|
}
|
2390 |
285d1929
|
p.kelbert
|
}
|
2391 |
1de96f38
|
p.kelbert
|
if (!configState.getConfig().doKeepOriginalSecundum()) {
|
2392 |
|
|
acceptedTaxon.setSec(configState.getConfig().getSecundum());
|
2393 |
6c41cae2
|
p.kelbert
|
//logger.info("SET SECUNDUM "+configState.getConfig().getSecundum());
|
2394 |
|
|
//System.out.println("SET SECUNDUM "+configState.getConfig().getSecundum());
|
2395 |
1de96f38
|
p.kelbert
|
}
|
2396 |
6b2c2013
|
p.kelbert
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
2397 |
|
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
2398 |
1de96f38
|
p.kelbert
|
if (!sourcelinked){
|
2399 |
2b5a7979
|
p.kelbert
|
sourceHandler.addSource(refMods, acceptedTaxon);
|
2400 |
1de96f38
|
p.kelbert
|
}
|
2401 |
|
|
if (!sourcelinked || !configState.getConfig().doKeepOriginalSecundum()){
|
2402 |
136e16b1
|
Andreas Müller
|
|
2403 |
2b5a7979
|
p.kelbert
|
if (!currentMyName.getIdentifier().isEmpty() && (currentMyName.getIdentifier().length()>2)){
|
2404 |
|
|
setLSID(currentMyName.getIdentifier(), acceptedTaxon);
|
2405 |
1de96f38
|
p.kelbert
|
}
|
2406 |
|
|
importer.getTaxonService().saveOrUpdate(acceptedTaxon);
|
2407 |
136e16b1
|
Andreas Müller
|
}
|
2408 |
285d1929
|
p.kelbert
|
}
|
2409 |
1de96f38
|
p.kelbert
|
}else{
|
2410 |
|
|
maxRankRespected=false;
|
2411 |
285d1929
|
p.kelbert
|
}
|
2412 |
1de96f38
|
p.kelbert
|
containsSynonyms=true;
|
2413 |
285d1929
|
p.kelbert
|
}else{
|
2414 |
6c41cae2
|
p.kelbert
|
//System.out.println("YOUHOUUU "+i);
|
2415 |
2b5a7979
|
p.kelbert
|
try{
|
2416 |
|
|
extractSynonyms(children.item(i), acceptedTaxon, refMods);
|
2417 |
|
|
}catch(NullPointerException e){
|
2418 |
|
|
logger.warn("nullpointerexception, the accepted taxon might be null");
|
2419 |
|
|
}
|
2420 |
285d1929
|
p.kelbert
|
}
|
2421 |
6b2c2013
|
p.kelbert
|
containsSynonyms=true;
|
2422 |
285d1929
|
p.kelbert
|
}
|
2423 |
1de96f38
|
p.kelbert
|
if (children.item(i).getNodeName().equalsIgnoreCase("tax:ref_group") && maxRankRespected){
|
2424 |
|
|
reloadClassification();
|
2425 |
|
|
//extract the References within the document
|
2426 |
|
|
extractReferences(children.item(i),nametosave,acceptedTaxon,refMods);
|
2427 |
|
|
}
|
2428 |
6c41cae2
|
p.kelbert
|
if(!stringIsEmpty(freetext.trim())) {
|
2429 |
|
|
setParticularDescription(freetext.trim(),acceptedTaxon,acceptedTaxon, refMods,getNotMarkedUpFeatureObject());
|
2430 |
2b5a7979
|
p.kelbert
|
}
|
2431 |
1de96f38
|
p.kelbert
|
|
2432 |
285d1929
|
p.kelbert
|
}
|
2433 |
|
|
// importer.getClassificationService().saveOrUpdate(classification);
|
2434 |
|
|
return acceptedTaxon;
|
2435 |
|
|
}
|
2436 |
|
|
|
2437 |
2b5a7979
|
p.kelbert
|
|
2438 |
1de96f38
|
p.kelbert
|
/**
|
2439 |
|
|
* @return
|
2440 |
|
|
*/
|
2441 |
6e861bce
|
Andreas Müller
|
|
2442 |
|
|
private boolean compareStatus(TaxonBase<?> t, NomenclaturalStatusType statusType) {
|
2443 |
6c41cae2
|
p.kelbert
|
//logger.info("compareStatus");
|
2444 |
1de96f38
|
p.kelbert
|
boolean statusMatch=false;
|
2445 |
|
|
//found one taxon
|
2446 |
|
|
Set<NomenclaturalStatus> status = t.getName().getStatus();
|
2447 |
|
|
if (statusType!=null && status.size()>0){ //the statusType is known for both taxon
|
2448 |
|
|
for (NomenclaturalStatus st:status){
|
2449 |
|
|
NomenclaturalStatusType stype = st.getType();
|
2450 |
|
|
if (stype.toString().equalsIgnoreCase(statusType.toString())) {
|
2451 |
|
|
statusMatch=true;
|
2452 |
|
|
}
|
2453 |
|
|
}
|
2454 |
|
|
}
|
2455 |
|
|
else{
|
2456 |
|
|
if(statusType == null && status.size()==0) {//there is no statusType, we can assume it's the same
|
2457 |
|
|
statusMatch=true;
|
2458 |
|
|
}
|
2459 |
|
|
}
|
2460 |
|
|
return statusMatch;
|
2461 |
|
|
}
|
2462 |
|
|
|
2463 |
285d1929
|
p.kelbert
|
/**
|
2464 |
|
|
* @param acceptedTaxon: the current acceptedTaxon
|
2465 |
|
|
* @param ref: the current reference extracted from the MODS
|
2466 |
|
|
* @return the parent for the current accepted taxon
|
2467 |
|
|
*/
|
2468 |
2b5a7979
|
p.kelbert
|
/* private Taxon createParent(Taxon acceptedTaxon, Reference<?> ref) {
|
2469 |
285d1929
|
p.kelbert
|
acceptedTaxon = CdmBase.deproxy(acceptedTaxon, Taxon.class);
|
2470 |
|
|
|
2471 |
|
|
List<Rank> rankList = new ArrayList<Rank>();
|
2472 |
|
|
rankList = importer.getTermService().listByTermClass(Rank.class, null, null, null, null);
|
2473 |
|
|
|
2474 |
|
|
List<String> rankListStr = new ArrayList<String>();
|
2475 |
|
|
for (Rank r:rankList) {
|
2476 |
|
|
rankListStr.add(r.toString());
|
2477 |
|
|
}
|
2478 |
|
|
String r="";
|
2479 |
|
|
String s = acceptedTaxon.getTitleCache();
|
2480 |
|
|
Taxon tax = null;
|
2481 |
2b5a7979
|
p.kelbert
|
if(!skippQuestion){
|
2482 |
|
|
int addTaxon = askAddParent(s);
|
2483 |
|
|
logger.info("ADD TAXON: "+addTaxon);
|
2484 |
|
|
if (addTaxon == 0 ){
|
2485 |
|
|
Taxon tmp = askParent(acceptedTaxon, classification);
|
2486 |
|
|
if (tmp == null){
|
2487 |
|
|
s = askSetParent(s);
|
2488 |
|
|
r = askRank(s,rankListStr);
|
2489 |
|
|
|
2490 |
|
|
NonViralName<?> nameToBeFilled = null;
|
2491 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNAFP)){
|
2492 |
|
|
nameToBeFilled = BotanicalName.NewInstance(null);
|
2493 |
|
|
}
|
2494 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICZN)){
|
2495 |
|
|
nameToBeFilled = ZoologicalName.NewInstance(null);
|
2496 |
|
|
}
|
2497 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNB)){
|
2498 |
|
|
nameToBeFilled = BacterialName.NewInstance(null);
|
2499 |
|
|
}
|
2500 |
d03aed8f
|
Andreas Müller
|
nameToBeFilled.setTitleCache(s, true);
|
2501 |
|
|
nameToBeFilled.setRank(getRank(r), true);
|
2502 |
285d1929
|
p.kelbert
|
|
2503 |
2b5a7979
|
p.kelbert
|
tax = Taxon.NewInstance(nameToBeFilled, ref);
|
2504 |
285d1929
|
p.kelbert
|
}
|
2505 |
2b5a7979
|
p.kelbert
|
else{
|
2506 |
|
|
tax=tmp;
|
2507 |
136e16b1
|
Andreas Müller
|
}
|
2508 |
285d1929
|
p.kelbert
|
|
2509 |
2b5a7979
|
p.kelbert
|
createParent(tax, ref);
|
2510 |
|
|
// logger.info("add parent child "+tax.getTitleCache()+", "+acceptedTaxon.getTitleCache());
|
2511 |
|
|
classification.addParentChild(tax, acceptedTaxon, ref, null);
|
2512 |
285d1929
|
p.kelbert
|
}
|
2513 |
|
|
else{
|
2514 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(acceptedTaxon, ref, null);
|
2515 |
|
|
tax=acceptedTaxon;
|
2516 |
285d1929
|
p.kelbert
|
}
|
2517 |
2b5a7979
|
p.kelbert
|
} else{
|
2518 |
59c2c696
|
Andreas Müller
|
classification.addChildTaxon(acceptedTaxon, ref, null);
|
2519 |
285d1929
|
p.kelbert
|
tax=acceptedTaxon;
|
2520 |
|
|
}
|
2521 |
|
|
// logger.info("RETURN: "+tax );
|
2522 |
|
|
return tax;
|
2523 |
|
|
|
2524 |
|
|
}
|
2525 |
|
|
|
2526 |
2b5a7979
|
p.kelbert
|
*/
|
2527 |
285d1929
|
p.kelbert
|
|
2528 |
6b2c2013
|
p.kelbert
|
|
2529 |
|
|
private MyName extractScientificNameSynonym(Node name, Reference<?> refMods) throws TransformerFactoryConfigurationError, TransformerException {
|
2530 |
6c41cae2
|
p.kelbert
|
//System.out.println("extractScientificNameSynonym");
|
2531 |
|
|
logger.info("extractScientificNameSynonym");
|
2532 |
6b2c2013
|
p.kelbert
|
String[] rankListToPrint_tmp ={"dwc:genus","dwc:specificepithet","dwc:species","dwc:subspecies", "dwc:infraspecificepithet","dwc:scientificnameauthorship"};
|
2533 |
|
|
List<String> rankListToPrint = new ArrayList<String>();
|
2534 |
|
|
for (String r : rankListToPrint_tmp) {
|
2535 |
|
|
rankListToPrint.add(r.toLowerCase());
|
2536 |
|
|
}
|
2537 |
|
|
|
2538 |
|
|
Rank rank = Rank.UNKNOWN_RANK();
|
2539 |
|
|
NodeList children = name.getChildNodes();
|
2540 |
6c41cae2
|
p.kelbert
|
String originalName="";
|
2541 |
6b2c2013
|
p.kelbert
|
String fullName = "";
|
2542 |
|
|
String newName="";
|
2543 |
|
|
String identifier="";
|
2544 |
|
|
HashMap<String, String> atomisedMap = new HashMap<String, String>();
|
2545 |
|
|
List<String> atomisedName= new ArrayList<String>();
|
2546 |
|
|
|
2547 |
|
|
String rankStr = "";
|
2548 |
|
|
Rank tmpRank ;
|
2549 |
|
|
|
2550 |
|
|
String status= extractStatus(children);
|
2551 |
|
|
|
2552 |
|
|
for (int i=0;i<children.getLength();i++){
|
2553 |
|
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:xmldata")){
|
2554 |
|
|
NodeList atom = children.item(i).getChildNodes();
|
2555 |
|
|
for (int k=0;k<atom.getLength();k++){
|
2556 |
|
|
identifier = extractIdentifier(identifier, atom.item(k));
|
2557 |
|
|
tmpRank = null;
|
2558 |
|
|
rankStr = atom.item(k).getNodeName().toLowerCase();
|
2559 |
|
|
// logger.info("RANKSTR:*"+rankStr+"*");
|
2560 |
|
|
if (rankStr.equalsIgnoreCase("dwc:taxonRank")) {
|
2561 |
|
|
rankStr=atom.item(k).getTextContent().trim();
|
2562 |
|
|
tmpRank = getRank(rankStr);
|
2563 |
|
|
}
|
2564 |
|
|
// if ((tmpRank != null) && (tmpRank.isLower(rank) || rank.equals(Rank.UNKNOWN_RANK()))) {
|
2565 |
|
|
if (tmpRank != null){
|
2566 |
|
|
rank=tmpRank;
|
2567 |
|
|
}
|
2568 |
|
|
atomisedMap.put(rankStr.toLowerCase(),atom.item(k).getTextContent().trim());
|
2569 |
|
|
}
|
2570 |
|
|
addAtomisedNamesToMap(rankListToPrint, rank, atomisedName, atom);
|
2571 |
|
|
}
|
2572 |
|
|
if(children.item(i).getNodeName().equalsIgnoreCase("#text") && !StringUtils.isBlank(children.item(i).getTextContent())){
|
2573 |
|
|
// logger.info("name non atomised: "+children.item(i).getTextContent());
|
2574 |
|
|
fullName = children.item(i).getTextContent().trim();
|
2575 |
|
|
// logger.info("fullname: "+fullName);
|
2576 |
|
|
}
|
2577 |
|
|
}
|
2578 |
6c41cae2
|
p.kelbert
|
originalName=fullName;
|
2579 |
6b2c2013
|
p.kelbert
|
fullName = cleanName(fullName, atomisedName);
|
2580 |
|
|
namesMap.put(fullName,atomisedMap);
|
2581 |
|
|
|
2582 |
|
|
String atomisedNameStr = getAtomisedNameStr(atomisedName);
|
2583 |
|
|
|
2584 |
|
|
if (fullName != null){
|
2585 |
|
|
// System.out.println("fullname: "+fullName);
|
2586 |
|
|
// System.out.println("atomised: "+atomisedNameStr);
|
2587 |
|
|
if (!fullName.equalsIgnoreCase(atomisedNameStr)) {
|
2588 |
|
|
if (skippQuestion){
|
2589 |
|
|
// String defaultN = "";
|
2590 |
|
|
if (atomisedNameStr.length()>fullName.length()) {
|
2591 |
|
|
newName=atomisedNameStr;
|
2592 |
|
|
} else {
|
2593 |
|
|
if (fullName.length()>atomisedNameStr.length() && (rank.isLower(Rank.SPECIES()) && fullName.length()>2 && !fullName.substring(0, 1).equals("."))) {
|
2594 |
6c41cae2
|
p.kelbert
|
newName=askWhichScientificName(fullName,atomisedNameStr,classification.getTitleCache(),name);
|
2595 |
6b2c2013
|
p.kelbert
|
} else {
|
2596 |
|
|
newName=fullName;
|
2597 |
|
|
}
|
2598 |
|
|
}
|
2599 |
|
|
} else {
|
2600 |
6c41cae2
|
p.kelbert
|
newName=askWhichScientificName(fullName,atomisedNameStr,classification.getTitleCache(),name);
|
2601 |
6b2c2013
|
p.kelbert
|
}
|
2602 |
|
|
} else {
|
2603 |
|
|
newName=fullName;
|
2604 |
|
|
}
|
2605 |
|
|
}
|
2606 |
|
|
//not really needed
|
2607 |
|
|
// rank = askForRank(newName, rank, nomenclaturalCode);
|
2608 |
|
|
// System.out.println("atomised: "+atomisedMap.toString());
|
2609 |
|
|
|
2610 |
|
|
// String[] names = new String[5];
|
2611 |
|
|
MyName myname = new MyName(true);
|
2612 |
|
|
|
2613 |
6c41cae2
|
p.kelbert
|
//System.out.println("Handle "+newName+ "(rank: "+rank+")");
|
2614 |
6b2c2013
|
p.kelbert
|
// System.out.println(atomisedMap.keySet());
|
2615 |
|
|
fullName = extractAuthorFromNames(rank, fullName, atomisedMap, myname);
|
2616 |
|
|
myname.setOriginalName(fullName);
|
2617 |
|
|
myname.setNewName(newName);
|
2618 |
|
|
myname.setRank(rank);
|
2619 |
|
|
myname.setIdentifier(identifier);
|
2620 |
|
|
myname.setStatus(status);
|
2621 |
|
|
myname.setSource(refMods);
|
2622 |
|
|
|
2623 |
|
|
// boolean higherAdded=false;
|
2624 |
|
|
|
2625 |
|
|
|
2626 |
|
|
boolean parseNameManually=false;
|
2627 |
6e861bce
|
Andreas Müller
|
INonViralNameParser<?> parser = NonViralNameParserImpl.NewInstance();
|
2628 |
6c41cae2
|
p.kelbert
|
TaxonNameBase<?,?> nameToBeFilledTest ;
|
2629 |
|
|
|
2630 |
|
|
//if selected the atomised version
|
2631 |
|
|
if(newName==atomisedNameStr){
|
2632 |
|
|
nameToBeFilledTest = parser.parseFullName(atomisedNameStr, nomenclaturalCode, rank);
|
2633 |
|
|
if (nameToBeFilledTest.hasProblem()){
|
2634 |
|
|
addProblemNameToFile("ato",atomisedNameStr,nomenclaturalCode,rank, nameToBeFilledTest.getParsingProblems().toString());
|
2635 |
|
|
nameToBeFilledTest = parser.parseFullName(fullName, nomenclaturalCode,rank);
|
2636 |
|
|
if (nameToBeFilledTest.hasProblem()){
|
2637 |
|
|
addProblemNameToFile("full",fullName,nomenclaturalCode,rank, nameToBeFilledTest.getParsingProblems().toString());
|
2638 |
|
|
parseNameManually=true;
|
2639 |
|
|
}
|
2640 |
|
|
}
|
2641 |
|
|
}else{
|
2642 |
|
|
nameToBeFilledTest = parser.parseFullName(fullName, nomenclaturalCode, rank);
|
2643 |
6b2c2013
|
p.kelbert
|
if (nameToBeFilledTest.hasProblem()){
|
2644 |
6c41cae2
|
p.kelbert
|
addProblemNameToFile("fullversion",fullName,nomenclaturalCode,rank, nameToBeFilledTest.getParsingProblems().toString());
|
2645 |
|
|
nameToBeFilledTest = parser.parseFullName(fullName, nomenclaturalCode,rank);
|
2646 |
6b2c2013
|
p.kelbert
|
parseNameManually=true;
|
2647 |
6c41cae2
|
p.kelbert
|
if(!originalName.equalsIgnoreCase(atomisedNameStr)) {
|
2648 |
|
|
addNameDifferenceToFile(originalName,atomisedNameStr);
|
2649 |
|
|
}
|
2650 |
6b2c2013
|
p.kelbert
|
}
|
2651 |
|
|
}
|
2652 |
|
|
|
2653 |
|
|
if(parseNameManually){
|
2654 |
6c41cae2
|
p.kelbert
|
//System.out.println("DO IT MANUALLY");
|
2655 |
6b2c2013
|
p.kelbert
|
createSynonym(rank, newName, atomisedMap, myname);
|
2656 |
|
|
}
|
2657 |
|
|
else{
|
2658 |
6c41cae2
|
p.kelbert
|
//System.out.println("AUTOMATIC!");
|
2659 |
6b2c2013
|
p.kelbert
|
// createAtomisedTaxonString(newName, atomisedMap, myname);
|
2660 |
|
|
myname.setParsedName(nameToBeFilledTest);
|
2661 |
|
|
myname.buildTaxon();
|
2662 |
|
|
}
|
2663 |
6c41cae2
|
p.kelbert
|
//System.out.println("RETURN SYNONYM "+myname.getSyno().toString());
|
2664 |
6b2c2013
|
p.kelbert
|
return myname;
|
2665 |
|
|
}
|
2666 |
285d1929
|
p.kelbert
|
/**
|
2667 |
|
|
* @param name
|
2668 |
|
|
* @throws TransformerFactoryConfigurationError
|
2669 |
|
|
* @throws TransformerException
|
2670 |
|
|
* @return a list of possible names
|
2671 |
|
|
*/
|
2672 |
2b5a7979
|
p.kelbert
|
@SuppressWarnings({ "null", "rawtypes" })
|
2673 |
|
|
private MyName extractScientificName(Node name, Reference<?> refMods) throws TransformerFactoryConfigurationError, TransformerException {
|
2674 |
6c41cae2
|
p.kelbert
|
logger.info("extractScientificName");
|
2675 |
1de96f38
|
p.kelbert
|
|
2676 |
2b5a7979
|
p.kelbert
|
String[] rankListToPrint_tmp ={"dwc:genus","dwc:specificepithet","dwc:species","dwc:subspecies", "dwc:infraspecificepithet","dwc:scientificnameauthorship"};
|
2677 |
1de96f38
|
p.kelbert
|
List<String> rankListToPrint = new ArrayList<String>();
|
2678 |
|
|
for (String r : rankListToPrint_tmp) {
|
2679 |
|
|
rankListToPrint.add(r.toLowerCase());
|
2680 |
|
|
}
|
2681 |
|
|
|
2682 |
285d1929
|
p.kelbert
|
Rank rank = Rank.UNKNOWN_RANK();
|
2683 |
|
|
NodeList children = name.getChildNodes();
|
2684 |
6c41cae2
|
p.kelbert
|
String originalName="";
|
2685 |
285d1929
|
p.kelbert
|
String fullName = "";
|
2686 |
|
|
String newName="";
|
2687 |
136e16b1
|
Andreas Müller
|
String identifier="";
|
2688 |
285d1929
|
p.kelbert
|
HashMap<String, String> atomisedMap = new HashMap<String, String>();
|
2689 |
|
|
List<String> atomisedName= new ArrayList<String>();
|
2690 |
|
|
|
2691 |
|
|
String rankStr = "";
|
2692 |
136e16b1
|
Andreas Müller
|
Rank tmpRank ;
|
2693 |
1de96f38
|
p.kelbert
|
|
2694 |
2b5a7979
|
p.kelbert
|
String status= extractStatus(children);
|
2695 |
1de96f38
|
p.kelbert
|
|
2696 |
285d1929
|
p.kelbert
|
for (int i=0;i<children.getLength();i++){
|
2697 |
|
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:xmldata")){
|
2698 |
|
|
NodeList atom = children.item(i).getChildNodes();
|
2699 |
|
|
for (int k=0;k<atom.getLength();k++){
|
2700 |
2b5a7979
|
p.kelbert
|
identifier = extractIdentifier(identifier, atom.item(k));
|
2701 |
136e16b1
|
Andreas Müller
|
tmpRank = null;
|
2702 |
285d1929
|
p.kelbert
|
rankStr = atom.item(k).getNodeName().toLowerCase();
|
2703 |
|
|
// logger.info("RANKSTR:*"+rankStr+"*");
|
2704 |
|
|
if (rankStr.equalsIgnoreCase("dwc:taxonRank")) {
|
2705 |
|
|
rankStr=atom.item(k).getTextContent().trim();
|
2706 |
|
|
tmpRank = getRank(rankStr);
|
2707 |
|
|
}
|
2708 |
136e16b1
|
Andreas Müller
|
// if ((tmpRank != null) && (tmpRank.isLower(rank) || rank.equals(Rank.UNKNOWN_RANK()))) {
|
2709 |
|
|
if (tmpRank != null){
|
2710 |
285d1929
|
p.kelbert
|
rank=tmpRank;
|
2711 |
|
|
}
|
2712 |
136e16b1
|
Andreas Müller
|
atomisedMap.put(rankStr.toLowerCase(),atom.item(k).getTextContent().trim());
|
2713 |
285d1929
|
p.kelbert
|
}
|
2714 |
2b5a7979
|
p.kelbert
|
addAtomisedNamesToMap(rankListToPrint, rank, atomisedName, atom);
|
2715 |
285d1929
|
p.kelbert
|
}
|
2716 |
|
|
if(children.item(i).getNodeName().equalsIgnoreCase("#text") && !StringUtils.isBlank(children.item(i).getTextContent())){
|
2717 |
|
|
// logger.info("name non atomised: "+children.item(i).getTextContent());
|
2718 |
|
|
fullName = children.item(i).getTextContent().trim();
|
2719 |
|
|
// logger.info("fullname: "+fullName);
|
2720 |
|
|
}
|
2721 |
|
|
}
|
2722 |
6c41cae2
|
p.kelbert
|
originalName=fullName;
|
2723 |
2b5a7979
|
p.kelbert
|
fullName = cleanName(fullName, atomisedName);
|
2724 |
285d1929
|
p.kelbert
|
namesMap.put(fullName,atomisedMap);
|
2725 |
2b5a7979
|
p.kelbert
|
|
2726 |
|
|
String atomisedNameStr = getAtomisedNameStr(atomisedName);
|
2727 |
285d1929
|
p.kelbert
|
|
2728 |
|
|
if (fullName != null){
|
2729 |
2b5a7979
|
p.kelbert
|
// System.out.println("fullname: "+fullName);
|
2730 |
|
|
// System.out.println("atomised: "+atomisedNameStr);
|
2731 |
285d1929
|
p.kelbert
|
if (!fullName.equalsIgnoreCase(atomisedNameStr)) {
|
2732 |
6c41cae2
|
p.kelbert
|
//System.out.println("atomisedNameStr vs. fullName:"+atomisedNameStr+"--"+fullName);
|
2733 |
2b5a7979
|
p.kelbert
|
if (skippQuestion){
|
2734 |
|
|
// String defaultN = "";
|
2735 |
|
|
if (atomisedNameStr.length()>fullName.length()) {
|
2736 |
|
|
newName=atomisedNameStr;
|
2737 |
|
|
} else {
|
2738 |
|
|
if (fullName.length()>atomisedNameStr.length() && (rank.isLower(Rank.SPECIES()) && fullName.length()>2 && !fullName.substring(0, 1).equals("."))) {
|
2739 |
6c41cae2
|
p.kelbert
|
//System.out.println("là");
|
2740 |
|
|
newName=askWhichScientificName(fullName,atomisedNameStr,classification.getTitleCache(),name);
|
2741 |
2b5a7979
|
p.kelbert
|
} else {
|
2742 |
6c41cae2
|
p.kelbert
|
//System.out.println("ici");
|
2743 |
2b5a7979
|
p.kelbert
|
newName=fullName;
|
2744 |
|
|
}
|
2745 |
|
|
}
|
2746 |
|
|
} else {
|
2747 |
6c41cae2
|
p.kelbert
|
newName=askWhichScientificName(fullName,atomisedNameStr,classification.getTitleCache(),name);
|
2748 |
2b5a7979
|
p.kelbert
|
}
|
2749 |
285d1929
|
p.kelbert
|
} else {
|
2750 |
|
|
newName=fullName;
|
2751 |
|
|
}
|
2752 |
|
|
}
|
2753 |
2b5a7979
|
p.kelbert
|
//not really needed
|
2754 |
|
|
// rank = askForRank(newName, rank, nomenclaturalCode);
|
2755 |
|
|
// System.out.println("atomised: "+atomisedMap.toString());
|
2756 |
|
|
|
2757 |
|
|
// String[] names = new String[5];
|
2758 |
6b2c2013
|
p.kelbert
|
MyName myname = new MyName(false);
|
2759 |
2b5a7979
|
p.kelbert
|
|
2760 |
6c41cae2
|
p.kelbert
|
//System.out.println("\n\nBUILD "+newName+ "(rank: "+rank+")");
|
2761 |
2b5a7979
|
p.kelbert
|
// System.out.println(atomisedMap.keySet());
|
2762 |
|
|
fullName = extractAuthorFromNames(rank, fullName, atomisedMap, myname);
|
2763 |
1de96f38
|
p.kelbert
|
myname.setOriginalName(fullName);
|
2764 |
|
|
myname.setNewName(newName);
|
2765 |
6c41cae2
|
p.kelbert
|
|
2766 |
1de96f38
|
p.kelbert
|
myname.setRank(rank);
|
2767 |
|
|
myname.setIdentifier(identifier);
|
2768 |
|
|
myname.setStatus(status);
|
2769 |
2b5a7979
|
p.kelbert
|
myname.setSource(refMods);
|
2770 |
|
|
|
2771 |
|
|
// boolean higherAdded=false;
|
2772 |
|
|
|
2773 |
|
|
|
2774 |
|
|
boolean parseNameManually=false;
|
2775 |
|
|
INonViralNameParser parser = NonViralNameParserImpl.NewInstance();
|
2776 |
6c41cae2
|
p.kelbert
|
TaxonNameBase nameToBeFilledTest = null;
|
2777 |
|
|
|
2778 |
|
|
//if selected the atomised version
|
2779 |
|
|
if(newName==atomisedNameStr){
|
2780 |
|
|
nameToBeFilledTest = parser.parseFullName(atomisedNameStr, nomenclaturalCode, rank);
|
2781 |
|
|
if (nameToBeFilledTest.hasProblem()){
|
2782 |
|
|
addProblemNameToFile("ato",atomisedNameStr,nomenclaturalCode,rank, nameToBeFilledTest.getParsingProblems().toString());
|
2783 |
|
|
nameToBeFilledTest = parser.parseFullName(fullName, nomenclaturalCode,rank);
|
2784 |
|
|
if (nameToBeFilledTest.hasProblem()){
|
2785 |
|
|
addProblemNameToFile("full",fullName,nomenclaturalCode,rank, nameToBeFilledTest.getParsingProblems().toString());
|
2786 |
|
|
parseNameManually=true;
|
2787 |
|
|
}
|
2788 |
|
|
}
|
2789 |
|
|
}else{
|
2790 |
|
|
nameToBeFilledTest = parser.parseFullName(fullName, nomenclaturalCode, rank);
|
2791 |
2b5a7979
|
p.kelbert
|
if (nameToBeFilledTest.hasProblem()){
|
2792 |
6c41cae2
|
p.kelbert
|
addProblemNameToFile("fullversion",fullName,nomenclaturalCode,rank, nameToBeFilledTest.getParsingProblems().toString());
|
2793 |
|
|
nameToBeFilledTest = parser.parseFullName(fullName, nomenclaturalCode,rank);
|
2794 |
2b5a7979
|
p.kelbert
|
parseNameManually=true;
|
2795 |
6c41cae2
|
p.kelbert
|
if(!originalName.equalsIgnoreCase(atomisedNameStr)) {
|
2796 |
|
|
addNameDifferenceToFile(originalName,atomisedNameStr);
|
2797 |
|
|
}
|
2798 |
2b5a7979
|
p.kelbert
|
}
|
2799 |
|
|
}
|
2800 |
6c41cae2
|
p.kelbert
|
|
2801 |
|
|
//System.out.println("parseNameManually: "+parseNameManually);
|
2802 |
2b5a7979
|
p.kelbert
|
if(parseNameManually){
|
2803 |
|
|
createAtomisedTaxon(rank, newName, atomisedMap, myname);
|
2804 |
|
|
}
|
2805 |
|
|
else{
|
2806 |
|
|
createAtomisedTaxonString(newName, atomisedMap, myname);
|
2807 |
|
|
myname.setParsedName(nameToBeFilledTest);
|
2808 |
|
|
myname.buildTaxon();
|
2809 |
|
|
}
|
2810 |
1de96f38
|
p.kelbert
|
return myname;
|
2811 |
285d1929
|
p.kelbert
|
|
2812 |
|
|
}
|
2813 |
|
|
|
2814 |
|
|
/**
|
2815 |
2b5a7979
|
p.kelbert
|
* @param atomisedName
|
2816 |
|
|
* @return
|
2817 |
285d1929
|
p.kelbert
|
*/
|
2818 |
2b5a7979
|
p.kelbert
|
private String getAtomisedNameStr(List<String> atomisedName) {
|
2819 |
6c41cae2
|
p.kelbert
|
//logger.info("getAtomisedNameStr");
|
2820 |
2b5a7979
|
p.kelbert
|
String atomisedNameStr = StringUtils.join(atomisedName," ");
|
2821 |
|
|
while(atomisedNameStr.contains(" ")) {
|
2822 |
|
|
atomisedNameStr=atomisedNameStr.replace(" ", " ");
|
2823 |
|
|
}
|
2824 |
|
|
atomisedNameStr=atomisedNameStr.trim();
|
2825 |
|
|
return atomisedNameStr;
|
2826 |
285d1929
|
p.kelbert
|
}
|
2827 |
|
|
|
2828 |
2b5a7979
|
p.kelbert
|
/**
|
2829 |
|
|
* @param children
|
2830 |
|
|
* @param status
|
2831 |
|
|
* @return
|
2832 |
|
|
*/
|
2833 |
|
|
private String extractStatus(NodeList children) {
|
2834 |
6c41cae2
|
p.kelbert
|
logger.info("extractStatus");
|
2835 |
1de96f38
|
p.kelbert
|
String status="";
|
2836 |
2b5a7979
|
p.kelbert
|
for (int i=0;i<children.getLength();i++){
|
2837 |
|
|
if(children.item(i).getNodeName().equalsIgnoreCase("tax:status") ||
|
2838 |
|
|
(children.item(i).getNodeName().equalsIgnoreCase("tax:namePart") &&
|
2839 |
|
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("status"))){
|
2840 |
|
|
status = children.item(i).getTextContent().trim();
|
2841 |
|
|
}
|
2842 |
|
|
}
|
2843 |
|
|
return status;
|
2844 |
|
|
}
|
2845 |
1de96f38
|
p.kelbert
|
|
2846 |
2b5a7979
|
p.kelbert
|
/**
|
2847 |
|
|
* @param identifier
|
2848 |
|
|
* @param atom
|
2849 |
|
|
* @param k
|
2850 |
|
|
* @return
|
2851 |
|
|
*/
|
2852 |
|
|
private String extractIdentifier(String identifier, Node atom) {
|
2853 |
6c41cae2
|
p.kelbert
|
//logger.info("extractIdentifier");
|
2854 |
2b5a7979
|
p.kelbert
|
if (atom.getNodeName().equalsIgnoreCase("tax:xid")){
|
2855 |
|
|
try{
|
2856 |
|
|
identifier = atom.getAttributes().getNamedItem("identifier").getNodeValue();
|
2857 |
|
|
}catch(Exception e){
|
2858 |
|
|
System.out.println("pb with identifier, maybe empty");
|
2859 |
1de96f38
|
p.kelbert
|
}
|
2860 |
2b5a7979
|
p.kelbert
|
try{
|
2861 |
|
|
identifier+="__"+atom.getAttributes().getNamedItem("source").getNodeValue();
|
2862 |
|
|
}catch(Exception e){
|
2863 |
|
|
System.out.println("pb with identifier, maybe empty");
|
2864 |
|
|
}
|
2865 |
|
|
}
|
2866 |
|
|
return identifier;
|
2867 |
|
|
}
|
2868 |
1de96f38
|
p.kelbert
|
|
2869 |
2b5a7979
|
p.kelbert
|
/**
|
2870 |
|
|
* @param rankListToPrint
|
2871 |
|
|
* @param rank
|
2872 |
|
|
* @param atomisedName
|
2873 |
|
|
* @param atom
|
2874 |
|
|
*/
|
2875 |
|
|
private void addAtomisedNamesToMap(List<String> rankListToPrint, Rank rank, List<String> atomisedName, NodeList atom) {
|
2876 |
6c41cae2
|
p.kelbert
|
logger.info("addAtomisedNamesToMap");
|
2877 |
2b5a7979
|
p.kelbert
|
for (int k=0;k<atom.getLength();k++){
|
2878 |
|
|
if (!atom.item(k).getNodeName().equalsIgnoreCase("dwc:taxonRank") ) {
|
2879 |
|
|
if (atom.item(k).getNodeName().equalsIgnoreCase("dwc:subgenus") || atom.item(k).getNodeName().equalsIgnoreCase("dwcranks:subgenus")) {
|
2880 |
|
|
atomisedName.add("("+atom.item(k).getTextContent().trim()+")");
|
2881 |
|
|
} else{
|
2882 |
|
|
if(atom.item(k).getNodeName().equalsIgnoreCase("dwcranks:varietyepithet") || atom.item(k).getNodeName().equalsIgnoreCase("dwc:Subspecies")) {
|
2883 |
|
|
if(atom.item(k).getNodeName().equalsIgnoreCase("dwcranks:varietyepithet")){
|
2884 |
|
|
atomisedName.add("var. "+atom.item(k).getTextContent().trim());
|
2885 |
|
|
}
|
2886 |
|
|
if(atom.item(k).getNodeName().equalsIgnoreCase("dwc:Subspecies") || atom.item(k).getNodeName().equalsIgnoreCase("dwc:infraspecificepithet")) {
|
2887 |
|
|
atomisedName.add("subsp. "+atom.item(k).getTextContent().trim());
|
2888 |
|
|
}
|
2889 |
|
|
}
|
2890 |
|
|
else{
|
2891 |
|
|
if(rankListToPrint.contains(atom.item(k).getNodeName().toLowerCase())) {
|
2892 |
|
|
atomisedName.add(atom.item(k).getTextContent().trim());
|
2893 |
|
|
}
|
2894 |
|
|
else{
|
2895 |
|
|
// System.out.println("rank : "+rank.toString());
|
2896 |
|
|
if (rank.isHigher(Rank.GENUS()) && (atom.item(k).getNodeName().indexOf("dwcranks:")>-1 || atom.item(k).getNodeName().indexOf("dwc:Family")>-1)) {
|
2897 |
|
|
atomisedName.add(atom.item(k).getTextContent().trim());
|
2898 |
|
|
}
|
2899 |
|
|
// else{
|
2900 |
|
|
// System.out.println("on a oublie qqn "+atom.item(k).getNodeName());
|
2901 |
|
|
// }
|
2902 |
|
|
}
|
2903 |
|
|
// else{
|
2904 |
|
|
// System.out.println("on a oublie qqn "+atom.item(k).getNodeName());
|
2905 |
|
|
// }
|
2906 |
|
|
}
|
2907 |
|
|
}
|
2908 |
|
|
}
|
2909 |
1de96f38
|
p.kelbert
|
}
|
2910 |
2b5a7979
|
p.kelbert
|
}
|
2911 |
|
|
|
2912 |
|
|
/**
|
2913 |
|
|
* @param fullName
|
2914 |
|
|
* @param atomisedName
|
2915 |
|
|
* @return
|
2916 |
|
|
*/
|
2917 |
|
|
private String cleanName(String name, List<String> atomisedName) {
|
2918 |
6c41cae2
|
p.kelbert
|
//logger.info("cleanName");
|
2919 |
2b5a7979
|
p.kelbert
|
String fullName =name;
|
2920 |
|
|
if (fullName != null){
|
2921 |
|
|
fullName = fullName.replace("( ", "(");
|
2922 |
|
|
fullName = fullName.replace(" )",")");
|
2923 |
|
|
|
2924 |
|
|
if (fullName.trim().isEmpty()){
|
2925 |
|
|
fullName=StringUtils.join(atomisedName," ");
|
2926 |
|
|
}
|
2927 |
|
|
|
2928 |
|
|
while(fullName.contains(" ")) {
|
2929 |
|
|
fullName=fullName.replace(" ", " ");
|
2930 |
|
|
// logger.info("while");
|
2931 |
|
|
}
|
2932 |
|
|
fullName=fullName.trim();
|
2933 |
1de96f38
|
p.kelbert
|
}
|
2934 |
2b5a7979
|
p.kelbert
|
return fullName;
|
2935 |
|
|
}
|
2936 |
|
|
|
2937 |
|
|
/**
|
2938 |
|
|
* @param rank
|
2939 |
|
|
* @param fullName
|
2940 |
|
|
* @param atomisedMap
|
2941 |
|
|
* @param myname
|
2942 |
|
|
* @return
|
2943 |
|
|
*/
|
2944 |
|
|
private String extractAuthorFromNames(Rank rank, String name, HashMap<String, String> atomisedMap,
|
2945 |
|
|
MyName myname) {
|
2946 |
6c41cae2
|
p.kelbert
|
logger.info("extractAuthorFromNames");
|
2947 |
2b5a7979
|
p.kelbert
|
String fullName=name;
|
2948 |
|
|
if (atomisedMap.get("dwc:scientificnameauthorship") == null && fullName!=null){
|
2949 |
|
|
// System.out.println("rank : "+rank.toString());
|
2950 |
|
|
if(rank.isHigher(Rank.SPECIES())){
|
2951 |
|
|
try{
|
2952 |
|
|
String author=null;
|
2953 |
|
|
if(atomisedMap.get("dwcranks:subgenus") != null) {
|
2954 |
|
|
author = fullName.split(atomisedMap.get("dwcranks:subgenus"))[1].trim();
|
2955 |
|
|
}
|
2956 |
|
|
if(atomisedMap.get("dwc:subgenus") != null) {
|
2957 |
|
|
author = fullName.split(atomisedMap.get("dwc:subgenus"))[1].trim();
|
2958 |
|
|
}
|
2959 |
|
|
if(author == null) {
|
2960 |
|
|
if(atomisedMap.get("dwc:genus") != null) {
|
2961 |
|
|
author = fullName.split(atomisedMap.get("dwc:genus"))[1].trim();
|
2962 |
|
|
}
|
2963 |
|
|
}
|
2964 |
|
|
if(author != null){
|
2965 |
|
|
fullName = fullName.substring(0, fullName.indexOf(author));
|
2966 |
|
|
author=author.replaceAll(",","").trim();
|
2967 |
|
|
myname.setAuthor(author);
|
2968 |
|
|
}
|
2969 |
|
|
}catch(Exception e){
|
2970 |
|
|
//could not extract the author
|
2971 |
|
|
}
|
2972 |
|
|
}
|
2973 |
|
|
if(rank.equals(Rank.SPECIES())){
|
2974 |
|
|
try{
|
2975 |
|
|
String author=null;
|
2976 |
|
|
if(author == null) {
|
2977 |
|
|
if(atomisedMap.get("dwc:species") != null) {
|
2978 |
|
|
String[] t = fullName.split(atomisedMap.get("dwc:species"));
|
2979 |
|
|
// System.out.println("NB ELEMENTS "+t.length +"fullName "+fullName+", "+atomisedMap.get("dwc:species"));
|
2980 |
|
|
author = fullName.split(atomisedMap.get("dwc:species"))[1].trim();
|
2981 |
|
|
// System.out.println("AUTEUR "+author);
|
2982 |
|
|
}
|
2983 |
|
|
}
|
2984 |
|
|
if(author != null){
|
2985 |
|
|
fullName = fullName.substring(0, fullName.indexOf(author));
|
2986 |
|
|
author=author.replaceAll(",","").trim();
|
2987 |
|
|
myname.setAuthor(author);
|
2988 |
|
|
}
|
2989 |
|
|
}catch(Exception e){
|
2990 |
|
|
//could not extract the author
|
2991 |
|
|
}
|
2992 |
|
|
}
|
2993 |
|
|
}else{
|
2994 |
|
|
myname.setAuthor(atomisedMap.get("dwc:scientificnameauthorship"));
|
2995 |
1de96f38
|
p.kelbert
|
}
|
2996 |
2b5a7979
|
p.kelbert
|
return fullName;
|
2997 |
|
|
}
|
2998 |
|
|
|
2999 |
|
|
/**
|
3000 |
|
|
* @param newName
|
3001 |
|
|
* @param atomisedMap
|
3002 |
|
|
* @param myname
|
3003 |
|
|
*/
|
3004 |
|
|
private void createAtomisedTaxonString(String newName, HashMap<String, String> atomisedMap, MyName myname) {
|
3005 |
6c41cae2
|
p.kelbert
|
logger.info("createAtomisedTaxonString "+atomisedMap);
|
3006 |
2b5a7979
|
p.kelbert
|
if(atomisedMap.get("dwc:family") != null && checkRankValidForImport(Rank.FAMILY())){
|
3007 |
|
|
myname.setFamilyStr(atomisedMap.get("dwc:family"));
|
3008 |
|
|
}
|
3009 |
|
|
if(atomisedMap.get("dwcranks:subfamily") != null && checkRankValidForImport(Rank.SUBFAMILY())){
|
3010 |
|
|
myname.setSubfamilyStr(atomisedMap.get("dwcranks:subfamily"));
|
3011 |
|
|
}
|
3012 |
|
|
if(atomisedMap.get("dwcranks:tribe") != null && checkRankValidForImport(Rank.TRIBE())){
|
3013 |
|
|
myname.setTribeStr(atomisedMap.get("dwcranks:tribe"));
|
3014 |
|
|
}
|
3015 |
|
|
if(atomisedMap.get("dwcranks:subtribe") != null && checkRankValidForImport(Rank.SUBTRIBE())){
|
3016 |
|
|
myname.setSubtribeStr(atomisedMap.get("dwcranks:subtribe"));
|
3017 |
|
|
}
|
3018 |
|
|
if(atomisedMap.get("dwc:genus") != null && checkRankValidForImport(Rank.GENUS())){
|
3019 |
|
|
myname.setGenusStr(atomisedMap.get("dwc:genus"));
|
3020 |
|
|
}
|
3021 |
|
|
if(atomisedMap.get("dwcranks:subgenus") != null && checkRankValidForImport(Rank.SUBGENUS())){
|
3022 |
|
|
myname.setSubgenusStr(atomisedMap.get("dwcranks:subgenus"));
|
3023 |
|
|
}
|
3024 |
|
|
if(atomisedMap.get("dwc:subgenus") != null && checkRankValidForImport(Rank.SUBGENUS())){
|
3025 |
|
|
myname.setSubgenusStr(atomisedMap.get("dwc:subgenus"));
|
3026 |
|
|
}
|
3027 |
|
|
if(atomisedMap.get("dwc:species") != null && checkRankValidForImport(Rank.SPECIES())){
|
3028 |
|
|
String n=newName;
|
3029 |
|
|
if(atomisedMap.get("dwc:infraspecificepithet") != null) {
|
3030 |
|
|
n=newName.split(atomisedMap.get("dwc:infraspecificepithet"))[0];
|
3031 |
|
|
n=n.replace("subsp.","");
|
3032 |
|
|
}
|
3033 |
|
|
if(atomisedMap.get("dwc:subspecies") != null) {
|
3034 |
|
|
n=newName.split(atomisedMap.get("dwc:subspecies"))[0];
|
3035 |
|
|
n=n.replace("subsp.","");
|
3036 |
|
|
}
|
3037 |
|
|
if(atomisedMap.get("dwcranks:varietyepithet") != null) {
|
3038 |
|
|
n=newName.split(atomisedMap.get("dwcranks:varietyepithet"))[0];
|
3039 |
|
|
n=n.replace("var.","");
|
3040 |
|
|
n=n.replace("v.","");
|
3041 |
|
|
}
|
3042 |
|
|
if(atomisedMap.get("dwcranks:formepithet") != null) {
|
3043 |
|
|
//TODO
|
3044 |
|
|
System.out.println("TODO FORMA");
|
3045 |
|
|
n=newName.split(atomisedMap.get("dwcranks:formepithet"))[0];
|
3046 |
|
|
n=n.replace("forma","");
|
3047 |
|
|
}
|
3048 |
|
|
n=n.trim();
|
3049 |
|
|
String author = myname.getAuthor();
|
3050 |
|
|
if(n.split(" ").length>2)
|
3051 |
|
|
{
|
3052 |
|
|
String n2=n.split(" ")[0]+" "+n.split(" ")[1];
|
3053 |
|
|
String a= "";
|
3054 |
|
|
try{
|
3055 |
|
|
a=n.split(n2)[1].trim();
|
3056 |
6c41cae2
|
p.kelbert
|
}catch(Exception e){
|
3057 |
|
|
logger.info("no author in "+n+"?");}
|
3058 |
2b5a7979
|
p.kelbert
|
|
3059 |
|
|
myname.setAuthor(a);
|
3060 |
6c41cae2
|
p.kelbert
|
//System.out.println("FINDCREATESPECIES --"+n2+"--"+n+"**"+a+"##");
|
3061 |
2b5a7979
|
p.kelbert
|
n=n2;
|
3062 |
|
|
|
3063 |
|
|
}
|
3064 |
|
|
|
3065 |
|
|
myname.setSpeciesStr(atomisedMap.get("dwc:species"));
|
3066 |
|
|
myname.setAuthor(author);
|
3067 |
|
|
}
|
3068 |
|
|
if(atomisedMap.get("dwc:subspecies") != null && checkRankValidForImport(Rank.SUBSPECIES())){
|
3069 |
|
|
myname.setSubspeciesStr(atomisedMap.get("dwc:subspecies"));
|
3070 |
|
|
}
|
3071 |
|
|
if(atomisedMap.get("dwc:infraspecificepithet") != null && checkRankValidForImport(Rank.SUBSPECIES())){
|
3072 |
|
|
myname.setSubspeciesStr(atomisedMap.get("dwc:infraspecificepithet"));
|
3073 |
|
|
}
|
3074 |
|
|
if(atomisedMap.get("dwcranks:varietyepithet") != null && checkRankValidForImport(Rank.VARIETY())){
|
3075 |
|
|
myname.setVarietyStr(atomisedMap.get("dwcranks:varietyepithet"));
|
3076 |
|
|
}
|
3077 |
|
|
if(atomisedMap.get("dwcranks:formepithet") != null && checkRankValidForImport(Rank.FORM())){
|
3078 |
|
|
myname.setFormStr(atomisedMap.get("dwcranks:formepithet"));
|
3079 |
|
|
}
|
3080 |
|
|
}
|
3081 |
|
|
|
3082 |
6b2c2013
|
p.kelbert
|
private void createSynonym(Rank rank, String newName, HashMap<String, String> atomisedMap, MyName myname) {
|
3083 |
6c41cae2
|
p.kelbert
|
logger.info("createSynonym");
|
3084 |
|
|
//System.out.println("createsynonym");
|
3085 |
6b2c2013
|
p.kelbert
|
if(rank.equals(Rank.UNKNOWN_RANK())){
|
3086 |
|
|
myname.setNotParsableTaxon(newName);
|
3087 |
|
|
}else
|
3088 |
|
|
{if(atomisedMap.get("dwc:family") != null && checkRankValidForImport(Rank.FAMILY()) && rank.equals(Rank.FAMILY())){
|
3089 |
|
|
myname.setFamily(myname.findOrCreateTaxon(atomisedMap.get("dwc:family"),newName, Rank.FAMILY(),rank));
|
3090 |
|
|
}
|
3091 |
|
|
if(atomisedMap.get("dwcranks:subfamily") != null && checkRankValidForImport(Rank.SUBFAMILY()) && rank.equals(Rank.SUBFAMILY())){
|
3092 |
|
|
myname.setSubfamily(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:subfamily"), newName,Rank.SUBFAMILY(),rank));
|
3093 |
|
|
}
|
3094 |
|
|
if(atomisedMap.get("dwcranks:tribe") != null && checkRankValidForImport(Rank.TRIBE()) && rank.equals(Rank.TRIBE())){
|
3095 |
|
|
myname.setTribe(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:tribe"),newName, Rank.TRIBE(),rank));
|
3096 |
|
|
}
|
3097 |
|
|
if(atomisedMap.get("dwcranks:subtribe") != null && checkRankValidForImport(Rank.SUBTRIBE()) && rank.equals(Rank.SUBTRIBE())){
|
3098 |
|
|
myname.setSubtribe(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:subtribe"),newName, Rank.SUBTRIBE(),rank));
|
3099 |
|
|
}
|
3100 |
|
|
if(atomisedMap.get("dwc:genus") != null && checkRankValidForImport(Rank.GENUS()) && rank.equals(Rank.GENUS())){
|
3101 |
|
|
myname.setGenus(myname.findOrCreateTaxon(atomisedMap.get("dwc:genus"),newName, Rank.GENUS(),rank));
|
3102 |
|
|
}
|
3103 |
|
|
if(atomisedMap.get("dwcranks:subgenus") != null && checkRankValidForImport(Rank.SUBGENUS()) && rank.equals(Rank.SUBGENUS())){
|
3104 |
|
|
myname.setSubgenus(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:subgenus"),newName, Rank.SUBGENUS(),rank));
|
3105 |
|
|
}
|
3106 |
|
|
if(atomisedMap.get("dwc:subgenus") != null && checkRankValidForImport(Rank.SUBGENUS()) && rank.equals(Rank.SUBGENUS())){
|
3107 |
|
|
myname.setSubgenus(myname.findOrCreateTaxon(atomisedMap.get("dwc:subgenus"),newName, Rank.SUBGENUS(),rank));
|
3108 |
|
|
}
|
3109 |
|
|
if(atomisedMap.get("dwc:species") != null && checkRankValidForImport(Rank.SPECIES()) && rank.equals(Rank.SPECIES())){
|
3110 |
|
|
String n=newName;
|
3111 |
|
|
if(atomisedMap.get("dwc:infraspecificepithet") != null) {
|
3112 |
|
|
n=newName.split(atomisedMap.get("dwc:infraspecificepithet"))[0];
|
3113 |
|
|
n=n.replace("subsp.","");
|
3114 |
|
|
}
|
3115 |
|
|
if(atomisedMap.get("dwc:subspecies") != null) {
|
3116 |
|
|
n=newName.split(atomisedMap.get("dwc:subspecies"))[0];
|
3117 |
|
|
n=n.replace("subsp.","");
|
3118 |
|
|
}
|
3119 |
|
|
if(atomisedMap.get("dwcranks:varietyepithet") != null) {
|
3120 |
|
|
n=newName.split(atomisedMap.get("dwcranks:varietyepithet"))[0];
|
3121 |
|
|
n=n.replace("var.","");
|
3122 |
|
|
n=n.replace("v.","");
|
3123 |
|
|
}
|
3124 |
|
|
if(atomisedMap.get("dwcranks:formepithet") != null) {
|
3125 |
|
|
//TODO
|
3126 |
6c41cae2
|
p.kelbert
|
//System.out.println("TODO FORMA");
|
3127 |
6b2c2013
|
p.kelbert
|
n=newName.split(atomisedMap.get("dwcranks:formepithet"))[0];
|
3128 |
|
|
n=n.replace("forma","");
|
3129 |
|
|
}
|
3130 |
|
|
n=n.trim();
|
3131 |
|
|
String author = myname.getAuthor();
|
3132 |
|
|
if(n.split(" ").length>2)
|
3133 |
|
|
{
|
3134 |
|
|
String n2=n.split(" ")[0]+" "+n.split(" ")[1];
|
3135 |
|
|
String a="";
|
3136 |
|
|
try{
|
3137 |
|
|
a= n.split(n2)[1].trim();
|
3138 |
6c41cae2
|
p.kelbert
|
}catch(Exception e){logger.info("no author in "+n);}
|
3139 |
6b2c2013
|
p.kelbert
|
myname.setAuthor(a);
|
3140 |
6c41cae2
|
p.kelbert
|
//System.out.println("FINDCREATESPECIES --"+n2+"--"+n+"**"+a+"##");
|
3141 |
6b2c2013
|
p.kelbert
|
n=n2;
|
3142 |
|
|
|
3143 |
|
|
}
|
3144 |
|
|
|
3145 |
|
|
myname.setSpecies(myname.findOrCreateTaxon(atomisedMap.get("dwc:species"),n, Rank.SPECIES(),rank));
|
3146 |
|
|
myname.setAuthor(author);
|
3147 |
|
|
}
|
3148 |
|
|
if(atomisedMap.get("dwc:subspecies") != null && checkRankValidForImport(Rank.SUBSPECIES()) && rank.equals(Rank.SUBSPECIES())){
|
3149 |
|
|
myname.setSubspecies(myname.findOrCreateTaxon(atomisedMap.get("dwc:subspecies"), newName,Rank.SUBSPECIES(),rank));
|
3150 |
|
|
}
|
3151 |
|
|
if(atomisedMap.get("dwc:infraspecificepithet") != null && checkRankValidForImport(Rank.SUBSPECIES()) && rank.equals(Rank.SUBSPECIES())){
|
3152 |
|
|
myname.setSubspecies(myname.findOrCreateTaxon(atomisedMap.get("dwc:infraspecificepithet"),newName, Rank.SUBSPECIES(),rank));
|
3153 |
|
|
}
|
3154 |
|
|
if(atomisedMap.get("dwcranks:varietyepithet") != null && checkRankValidForImport(Rank.VARIETY()) && rank.equals(Rank.VARIETY())){
|
3155 |
|
|
myname.setVariety(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:varietyepithet"),newName, Rank.VARIETY(),rank));
|
3156 |
|
|
}
|
3157 |
|
|
if(atomisedMap.get("dwcranks:formepithet") != null && checkRankValidForImport(Rank.FORM()) && rank.equals(Rank.FORM())){
|
3158 |
|
|
myname.setForm(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:formepithet"), newName,Rank.FORM(),rank));
|
3159 |
|
|
}
|
3160 |
|
|
}
|
3161 |
|
|
|
3162 |
|
|
}
|
3163 |
2b5a7979
|
p.kelbert
|
/**
|
3164 |
|
|
* @param rank
|
3165 |
|
|
* @param newName
|
3166 |
|
|
* @param atomisedMap
|
3167 |
|
|
* @param myname
|
3168 |
|
|
*/
|
3169 |
|
|
private void createAtomisedTaxon(Rank rank, String newName, HashMap<String, String> atomisedMap, MyName myname) {
|
3170 |
6c41cae2
|
p.kelbert
|
logger.info("createAtomisedTaxon "+atomisedMap);
|
3171 |
2b5a7979
|
p.kelbert
|
if(rank.equals(Rank.UNKNOWN_RANK())){
|
3172 |
|
|
myname.setNotParsableTaxon(newName);
|
3173 |
|
|
}
|
3174 |
|
|
else{
|
3175 |
|
|
if(atomisedMap.get("dwc:family") != null && checkRankValidForImport(Rank.FAMILY())){
|
3176 |
|
|
myname.setFamily(myname.findOrCreateTaxon(atomisedMap.get("dwc:family"),newName, Rank.FAMILY(),rank));
|
3177 |
|
|
}
|
3178 |
|
|
if(atomisedMap.get("dwcranks:subfamily") != null && checkRankValidForImport(Rank.SUBFAMILY())){
|
3179 |
|
|
myname.setSubfamily(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:subfamily"), newName,Rank.SUBFAMILY(),rank));
|
3180 |
|
|
}
|
3181 |
|
|
if(atomisedMap.get("dwcranks:tribe") != null && checkRankValidForImport(Rank.TRIBE())){
|
3182 |
|
|
myname.setTribe(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:tribe"),newName, Rank.TRIBE(),rank));
|
3183 |
|
|
}
|
3184 |
|
|
if(atomisedMap.get("dwcranks:subtribe") != null && checkRankValidForImport(Rank.SUBTRIBE())){
|
3185 |
|
|
myname.setSubtribe(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:subtribe"),newName, Rank.SUBTRIBE(),rank));
|
3186 |
|
|
}
|
3187 |
|
|
if(atomisedMap.get("dwc:genus") != null && checkRankValidForImport(Rank.GENUS())){
|
3188 |
|
|
myname.setGenus(myname.findOrCreateTaxon(atomisedMap.get("dwc:genus"),newName, Rank.GENUS(),rank));
|
3189 |
|
|
}
|
3190 |
|
|
if(atomisedMap.get("dwcranks:subgenus") != null && checkRankValidForImport(Rank.SUBGENUS())){
|
3191 |
|
|
myname.setSubgenus(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:subgenus"),newName, Rank.SUBGENUS(),rank));
|
3192 |
|
|
}
|
3193 |
|
|
if(atomisedMap.get("dwc:subgenus") != null && checkRankValidForImport(Rank.SUBGENUS())){
|
3194 |
|
|
myname.setSubgenus(myname.findOrCreateTaxon(atomisedMap.get("dwc:subgenus"),newName, Rank.SUBGENUS(),rank));
|
3195 |
|
|
}
|
3196 |
|
|
if(atomisedMap.get("dwc:species") != null && checkRankValidForImport(Rank.SPECIES())){
|
3197 |
|
|
String n=newName;
|
3198 |
|
|
if(atomisedMap.get("dwc:infraspecificepithet") != null) {
|
3199 |
|
|
n=newName.split(atomisedMap.get("dwc:infraspecificepithet"))[0];
|
3200 |
|
|
n=n.replace("subsp.","");
|
3201 |
|
|
}
|
3202 |
|
|
if(atomisedMap.get("dwc:subspecies") != null) {
|
3203 |
|
|
n=newName.split(atomisedMap.get("dwc:subspecies"))[0];
|
3204 |
|
|
n=n.replace("subsp.","");
|
3205 |
|
|
}
|
3206 |
|
|
if(atomisedMap.get("dwcranks:varietyepithet") != null) {
|
3207 |
|
|
n=newName.split(atomisedMap.get("dwcranks:varietyepithet"))[0];
|
3208 |
|
|
n=n.replace("var.","");
|
3209 |
|
|
n=n.replace("v.","");
|
3210 |
|
|
}
|
3211 |
|
|
if(atomisedMap.get("dwcranks:formepithet") != null) {
|
3212 |
|
|
//TODO
|
3213 |
6c41cae2
|
p.kelbert
|
//System.out.println("TODO FORMA");
|
3214 |
2b5a7979
|
p.kelbert
|
n=newName.split(atomisedMap.get("dwcranks:formepithet"))[0];
|
3215 |
|
|
n=n.replace("forma","");
|
3216 |
|
|
}
|
3217 |
|
|
n=n.trim();
|
3218 |
|
|
String author = myname.getAuthor();
|
3219 |
|
|
if(n.split(" ").length>2)
|
3220 |
|
|
{
|
3221 |
|
|
String n2=n.split(" ")[0]+" "+n.split(" ")[1];
|
3222 |
|
|
String a="";
|
3223 |
|
|
try{
|
3224 |
6b2c2013
|
p.kelbert
|
a= n.split(n2)[1].trim();
|
3225 |
6c41cae2
|
p.kelbert
|
}catch(Exception e){logger.info("no author in "+n);}
|
3226 |
2b5a7979
|
p.kelbert
|
myname.setAuthor(a);
|
3227 |
6c41cae2
|
p.kelbert
|
//System.out.println("FINDCREATESPECIES --"+n2+"--"+n+"**"+a+"##");
|
3228 |
2b5a7979
|
p.kelbert
|
n=n2;
|
3229 |
|
|
|
3230 |
|
|
}
|
3231 |
|
|
|
3232 |
|
|
myname.setSpecies(myname.findOrCreateTaxon(atomisedMap.get("dwc:species"),n, Rank.SPECIES(),rank));
|
3233 |
|
|
myname.setAuthor(author);
|
3234 |
|
|
}
|
3235 |
|
|
if(atomisedMap.get("dwc:subspecies") != null && checkRankValidForImport(Rank.SUBSPECIES())){
|
3236 |
|
|
myname.setSubspecies(myname.findOrCreateTaxon(atomisedMap.get("dwc:subspecies"), newName,Rank.SUBSPECIES(),rank));
|
3237 |
|
|
}
|
3238 |
|
|
if(atomisedMap.get("dwc:infraspecificepithet") != null && checkRankValidForImport(Rank.SUBSPECIES())){
|
3239 |
|
|
myname.setSubspecies(myname.findOrCreateTaxon(atomisedMap.get("dwc:infraspecificepithet"),newName, Rank.SUBSPECIES(),rank));
|
3240 |
|
|
}
|
3241 |
|
|
if(atomisedMap.get("dwcranks:varietyepithet") != null && checkRankValidForImport(Rank.VARIETY())){
|
3242 |
|
|
myname.setVariety(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:varietyepithet"),newName, Rank.VARIETY(),rank));
|
3243 |
|
|
}
|
3244 |
|
|
if(atomisedMap.get("dwcranks:formepithet") != null && checkRankValidForImport(Rank.FORM())){
|
3245 |
|
|
myname.setForm(myname.findOrCreateTaxon(atomisedMap.get("dwcranks:formepithet"), newName,Rank.FORM(),rank));
|
3246 |
|
|
}
|
3247 |
|
|
}
|
3248 |
|
|
}
|
3249 |
|
|
|
3250 |
|
|
/**
|
3251 |
|
|
* @return
|
3252 |
|
|
*/
|
3253 |
|
|
private boolean checkRankValidForImport(Rank currentRank) {
|
3254 |
6c41cae2
|
p.kelbert
|
//logger.info("checkRankValidForImport");
|
3255 |
2b5a7979
|
p.kelbert
|
return currentRank.isLower(configState.getConfig().getMaxRank()) || currentRank.equals(configState.getConfig().getMaxRank());
|
3256 |
|
|
}
|
3257 |
|
|
|
3258 |
|
|
|
3259 |
|
|
|
3260 |
|
|
/**
|
3261 |
|
|
* @param classification2
|
3262 |
|
|
*/
|
3263 |
|
|
public void updateClassification(Classification classification2) {
|
3264 |
6c41cae2
|
p.kelbert
|
//logger.info("updateClassification");
|
3265 |
2b5a7979
|
p.kelbert
|
classification = classification2;
|
3266 |
|
|
}
|
3267 |
|
|
|
3268 |
6c41cae2
|
p.kelbert
|
/**
|
3269 |
|
|
* @param tnb
|
3270 |
|
|
* cast the current taxonnamebase into a botanical name or zoological or bacterial name
|
3271 |
|
|
* if errors, cast into a classis nonviralname
|
3272 |
|
|
* @param taxonnamebase2
|
3273 |
|
|
*/
|
3274 |
|
|
@SuppressWarnings("rawtypes")
|
3275 |
|
|
public NonViralName<?> castTaxonNameBase(TaxonNameBase tnb, NonViralName<?> nvn) {
|
3276 |
|
|
//logger.info("castTaxonNameBase");
|
3277 |
|
|
NonViralName<?> taxonnamebase2 = nvn;
|
3278 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNAFP)) {
|
3279 |
|
|
try{
|
3280 |
|
|
taxonnamebase2=(BotanicalName) tnb;
|
3281 |
|
|
}catch(Exception e){
|
3282 |
|
|
taxonnamebase2= (NonViralName<?>) tnb;
|
3283 |
|
|
}
|
3284 |
|
|
}
|
3285 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICZN)) {
|
3286 |
|
|
try{
|
3287 |
|
|
taxonnamebase2=(ZoologicalName) tnb;
|
3288 |
|
|
}catch(Exception e){
|
3289 |
|
|
taxonnamebase2= (NonViralName<?>) tnb;
|
3290 |
|
|
}
|
3291 |
|
|
}
|
3292 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNB)) {
|
3293 |
|
|
try{
|
3294 |
|
|
taxonnamebase2=(BacterialName) tnb;
|
3295 |
|
|
}catch(Exception e){
|
3296 |
|
|
taxonnamebase2= (NonViralName<?>) tnb;
|
3297 |
|
|
}
|
3298 |
|
|
}
|
3299 |
|
|
return taxonnamebase2;
|
3300 |
|
|
}
|
3301 |
|
|
|
3302 |
|
|
/**
|
3303 |
|
|
* @param tnb
|
3304 |
|
|
* cast the current taxonnamebase into a botanical name or zoological or bacterial name
|
3305 |
|
|
* if errors, cast into a classis nonviralname
|
3306 |
|
|
* @param taxonnamebase2
|
3307 |
|
|
*/
|
3308 |
|
|
@SuppressWarnings("rawtypes")
|
3309 |
|
|
public NonViralName<?> castTaxonNameBase(TaxonNameBase tnb) {
|
3310 |
|
|
//logger.info("castTaxonNameBase2");
|
3311 |
|
|
NonViralName<?> taxonnamebase2 = null;
|
3312 |
|
|
tnb=CdmBase.deproxy(tnb, TaxonNameBase.class);
|
3313 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNAFP)) {
|
3314 |
|
|
try{
|
3315 |
|
|
taxonnamebase2=(BotanicalName) tnb;
|
3316 |
|
|
}catch(Exception e){
|
3317 |
|
|
taxonnamebase2= (NonViralName<?>) tnb;
|
3318 |
|
|
}
|
3319 |
|
|
}
|
3320 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICZN)) {
|
3321 |
|
|
try{
|
3322 |
|
|
taxonnamebase2=(ZoologicalName) tnb;
|
3323 |
|
|
}catch(Exception e){
|
3324 |
|
|
taxonnamebase2= (NonViralName<?>) tnb;
|
3325 |
|
|
}
|
3326 |
|
|
}
|
3327 |
|
|
if (nomenclaturalCode.equals(NomenclaturalCode.ICNB)) {
|
3328 |
|
|
try{
|
3329 |
|
|
taxonnamebase2=(BacterialName) tnb;
|
3330 |
|
|
}catch(Exception e){
|
3331 |
|
|
taxonnamebase2= (NonViralName<?>) tnb;
|
3332 |
|
|
}
|
3333 |
|
|
}
|
3334 |
|
|
return taxonnamebase2;
|
3335 |
|
|
}
|
3336 |
|
|
|
3337 |
2b5a7979
|
p.kelbert
|
public class MyName {
|
3338 |
6b2c2013
|
p.kelbert
|
/**
|
3339 |
|
|
* @param isSynonym
|
3340 |
|
|
*/
|
3341 |
|
|
public MyName(boolean isSynonym) {
|
3342 |
|
|
super();
|
3343 |
|
|
this.isSynonym = isSynonym;
|
3344 |
|
|
}
|
3345 |
|
|
|
3346 |
2b5a7979
|
p.kelbert
|
String originalName="";
|
3347 |
|
|
String newName="";
|
3348 |
|
|
Rank rank=Rank.UNKNOWN_RANK();
|
3349 |
|
|
String identifier="";
|
3350 |
|
|
String status="";
|
3351 |
|
|
String author=null;
|
3352 |
|
|
|
3353 |
|
|
NonViralName<?> taxonnamebase;
|
3354 |
|
|
|
3355 |
|
|
Reference<?> refMods ;
|
3356 |
|
|
|
3357 |
|
|
Taxon family,subfamily,tribe,subtribe,genus,subgenus,species,subspecies, variety,form;
|
3358 |
|
|
NonViralName<?> familyName, subfamilyName, tribeName,subtribeName,genusName,subgenusName,speciesName,subspeciesName;
|
3359 |
|
|
String familyStr, subfamilyStr, tribeStr,subtribeStr,genusStr,subgenusStr,speciesStr,subspeciesStr,formStr,varietyStr;
|
3360 |
|
|
Taxon higherTaxa;
|
3361 |
|
|
Rank higherRank;
|
3362 |
|
|
private Taxon taxon;
|
3363 |
6b2c2013
|
p.kelbert
|
private Synonym syno;
|
3364 |
|
|
|
3365 |
|
|
/**
|
3366 |
|
|
* @return the syno
|
3367 |
|
|
*/
|
3368 |
|
|
public Synonym getSyno() {
|
3369 |
|
|
return syno;
|
3370 |
|
|
}
|
3371 |
|
|
|
3372 |
6c41cae2
|
p.kelbert
|
@Override
|
3373 |
|
|
public String toString(){
|
3374 |
|
|
List<String> tot=new ArrayList<String>();
|
3375 |
|
|
String[] n= {familyStr, subfamilyStr, tribeStr,subtribeStr,genusStr,subgenusStr,speciesStr,subspeciesStr,formStr,varietyStr};
|
3376 |
|
|
for (String elt:n){
|
3377 |
|
|
if (!StringUtils.isEmpty(elt)) {
|
3378 |
|
|
tot.add(elt);
|
3379 |
|
|
} else {
|
3380 |
|
|
tot.add("*");
|
3381 |
|
|
}
|
3382 |
|
|
}
|
3383 |
|
|
return StringUtils.join(tot," ");
|
3384 |
|
|
}
|
3385 |
6b2c2013
|
p.kelbert
|
/**
|
3386 |
|
|
* @param syno the syno to set
|
3387 |
|
|
*/
|
3388 |
|
|
public void setSyno(Synonym syno) {
|
3389 |
|
|
this.syno = syno;
|
3390 |
|
|
}
|
3391 |
|
|
|
3392 |
|
|
boolean isSynonym=false;
|
3393 |
2b5a7979
|
p.kelbert
|
|
3394 |
6b2c2013
|
p.kelbert
|
/**
|
3395 |
|
|
* @return the isSynonym
|
3396 |
|
|
*/
|
3397 |
|
|
public boolean isSynonym() {
|
3398 |
|
|
return isSynonym;
|
3399 |
|
|
}
|
3400 |
|
|
|
3401 |
|
|
/**
|
3402 |
|
|
* @param isSynonym the isSynonym to set
|
3403 |
|
|
*/
|
3404 |
|
|
public void setSynonym(boolean isSynonym) {
|
3405 |
|
|
this.isSynonym = isSynonym;
|
3406 |
|
|
}
|
3407 |
2b5a7979
|
p.kelbert
|
|
3408 |
|
|
public void setSource(Reference<?> re){
|
3409 |
|
|
refMods=re;
|
3410 |
1de96f38
|
p.kelbert
|
}
|
3411 |
2b5a7979
|
p.kelbert
|
|
3412 |
1de96f38
|
p.kelbert
|
/**
|
3413 |
2b5a7979
|
p.kelbert
|
* @param string
|
3414 |
1de96f38
|
p.kelbert
|
*/
|
3415 |
2b5a7979
|
p.kelbert
|
public void setFormStr(String string) {
|
3416 |
|
|
this.formStr=string;
|
3417 |
|
|
|
3418 |
1de96f38
|
p.kelbert
|
}
|
3419 |
|
|
/**
|
3420 |
2b5a7979
|
p.kelbert
|
* @param string
|
3421 |
1de96f38
|
p.kelbert
|
*/
|
3422 |
2b5a7979
|
p.kelbert
|
public void setVarietyStr(String string) {
|
3423 |
|
|
this.varietyStr=string;
|
3424 |
|
|
|
3425 |
1de96f38
|
p.kelbert
|
}
|
3426 |
|
|
/**
|
3427 |
2b5a7979
|
p.kelbert
|
* @param string
|
3428 |
1de96f38
|
p.kelbert
|
*/
|
3429 |
2b5a7979
|
p.kelbert
|
public void setSubspeciesStr(String string) {
|
3430 |
|
|
this.subspeciesStr=string;
|
3431 |
|
|
|
3432 |
1de96f38
|
p.kelbert
|
}
|
3433 |
|
|
/**
|
3434 |
2b5a7979
|
p.kelbert
|
* @param string
|
3435 |
1de96f38
|
p.kelbert
|
*/
|
3436 |
2b5a7979
|
p.kelbert
|
public void setSpeciesStr(String string) {
|
3437 |
|
|
this.speciesStr=string;
|
3438 |
|
|
|
3439 |
1de96f38
|
p.kelbert
|
}
|
3440 |
|
|
/**
|
3441 |
2b5a7979
|
p.kelbert
|
* @param string
|
3442 |
1de96f38
|
p.kelbert
|
*/
|
3443 |
2b5a7979
|
p.kelbert
|
public void setSubgenusStr(String string) {
|
3444 |
|
|
this.subgenusStr=string;
|
3445 |
|
|
|
3446 |
1de96f38
|
p.kelbert
|
}
|
3447 |
|
|
/**
|
3448 |
2b5a7979
|
p.kelbert
|
* @param string
|
3449 |
1de96f38
|
p.kelbert
|
*/
|
3450 |
2b5a7979
|
p.kelbert
|
public void setGenusStr(String string) {
|
3451 |
|
|
this.genusStr=string;
|
3452 |
|
|
|
3453 |
1de96f38
|
p.kelbert
|
}
|
3454 |
|
|
/**
|
3455 |
2b5a7979
|
p.kelbert
|
* @param string
|
3456 |
1de96f38
|
p.kelbert
|
*/
|
3457 |
2b5a7979
|
p.kelbert
|
public void setSubtribeStr(String string) {
|
3458 |
|
|
this.subtribeStr=string;
|
3459 |
|
|
|
3460 |
|
|
}
|
3461 |
|
|
/**
|
3462 |
|
|
* @param string
|
3463 |
|
|
*/
|
3464 |
|
|
public void setTribeStr(String string) {
|
3465 |
|
|
this.tribeStr=string;
|
3466 |
|
|
|
3467 |
|
|
}
|
3468 |
|
|
/**
|
3469 |
|
|
* @param string
|
3470 |
|
|
*/
|
3471 |
|
|
public void setSubfamilyStr(String string) {
|
3472 |
|
|
this.subfamilyStr=string;
|
3473 |
|
|
|
3474 |
|
|
}
|
3475 |
|
|
/**
|
3476 |
|
|
* @param string
|
3477 |
|
|
*/
|
3478 |
|
|
public void setFamilyStr(String string) {
|
3479 |
|
|
this.familyStr=string;
|
3480 |
|
|
|
3481 |
|
|
}
|
3482 |
|
|
/**
|
3483 |
|
|
* @return the familyStr
|
3484 |
|
|
*/
|
3485 |
|
|
public String getFamilyStr() {
|
3486 |
|
|
return familyStr;
|
3487 |
|
|
}
|
3488 |
|
|
/**
|
3489 |
|
|
* @return the subfamilyStr
|
3490 |
|
|
*/
|
3491 |
|
|
public String getSubfamilyStr() {
|
3492 |
|
|
return subfamilyStr;
|
3493 |
|
|
}
|
3494 |
|
|
/**
|
3495 |
|
|
* @return the tribeStr
|
3496 |
|
|
*/
|
3497 |
|
|
public String getTribeStr() {
|
3498 |
|
|
return tribeStr;
|
3499 |
|
|
}
|
3500 |
|
|
/**
|
3501 |
|
|
* @return the subtribeStr
|
3502 |
|
|
*/
|
3503 |
|
|
public String getSubtribeStr() {
|
3504 |
|
|
return subtribeStr;
|
3505 |
|
|
}
|
3506 |
|
|
/**
|
3507 |
|
|
* @return the genusStr
|
3508 |
|
|
*/
|
3509 |
|
|
public String getGenusStr() {
|
3510 |
|
|
return genusStr;
|
3511 |
|
|
}
|
3512 |
|
|
/**
|
3513 |
|
|
* @return the subgenusStr
|
3514 |
|
|
*/
|
3515 |
|
|
public String getSubgenusStr() {
|
3516 |
|
|
return subgenusStr;
|
3517 |
|
|
}
|
3518 |
|
|
/**
|
3519 |
|
|
* @return the speciesStr
|
3520 |
|
|
*/
|
3521 |
|
|
public String getSpeciesStr() {
|
3522 |
|
|
return speciesStr;
|
3523 |
|
|
}
|
3524 |
|
|
/**
|
3525 |
|
|
* @return the subspeciesStr
|
3526 |
|
|
*/
|
3527 |
|
|
public String getSubspeciesStr() {
|
3528 |
|
|
return subspeciesStr;
|
3529 |
|
|
}
|
3530 |
|
|
/**
|
3531 |
|
|
* @return the formStr
|
3532 |
|
|
*/
|
3533 |
|
|
public String getFormStr() {
|
3534 |
|
|
return formStr;
|
3535 |
|
|
}
|
3536 |
|
|
/**
|
3537 |
|
|
* @return the varietyStr
|
3538 |
|
|
*/
|
3539 |
|
|
public String getVarietyStr() {
|
3540 |
|
|
return varietyStr;
|
3541 |
|
|
}
|
3542 |
|
|
|
3543 |
|
|
/**
|
3544 |
|
|
* @param newName2
|
3545 |
|
|
*/
|
3546 |
|
|
public void setNotParsableTaxon(String newName2) {
|
3547 |
6c41cae2
|
p.kelbert
|
//takes too much time
|
3548 |
|
|
// List<TaxonBase> tmpList = importer.getTaxonService().list(Taxon.class, 0, 0, null, null);
|
3549 |
2b5a7979
|
p.kelbert
|
|
3550 |
|
|
NomenclaturalStatusType statusType = null;
|
3551 |
|
|
if (!getStatus().isEmpty()){
|
3552 |
|
|
try {
|
3553 |
|
|
statusType = nomStatusString2NomStatus(getStatus());
|
3554 |
|
|
} catch (UnknownCdmTypeException e) {
|
3555 |
|
|
addProblematicStatusToFile(getStatus());
|
3556 |
|
|
logger.warn("Problem with status");
|
3557 |
|
|
}
|
3558 |
|
|
}
|
3559 |
6c41cae2
|
p.kelbert
|
List<TaxonBase> tmpList = new ArrayList<TaxonBase>();
|
3560 |
|
|
|
3561 |
|
|
Pager<TaxonBase> taxontest = importer.getTaxonService().findByTitle(TaxonBase.class, newName2, MatchMode.BEGINNING, null, null, null, null, null);
|
3562 |
|
|
tmpList.addAll(taxontest.getRecords());
|
3563 |
|
|
|
3564 |
|
|
//logger.info("tmpList returned: "+tmpList.size());
|
3565 |
|
|
|
3566 |
2b5a7979
|
p.kelbert
|
|
3567 |
|
|
boolean foundIdentic=false;
|
3568 |
6e861bce
|
Andreas Müller
|
TaxonBase<?> tmptaxonbase=null;
|
3569 |
2b5a7979
|
p.kelbert
|
// Taxon tmpPartial=null;
|
3570 |
6e861bce
|
Andreas Müller
|
for (TaxonBase<?> tmpb:tmpList){
|
3571 |
2b5a7979
|
p.kelbert
|
if(tmpb !=null){
|
3572 |
6e861bce
|
Andreas Müller
|
TaxonNameBase<?,?> tnb = tmpb.getName();
|
3573 |
2b5a7979
|
p.kelbert
|
Rank crank=null;
|
3574 |
|
|
if (tnb != null){
|
3575 |
6c41cae2
|
p.kelbert
|
if (tnb.getTitleCache().split("sec.")[0].trim().equalsIgnoreCase(newName2) ){
|
3576 |
2b5a7979
|
p.kelbert
|
crank =tnb.getRank();
|
3577 |
|
|
if (crank !=null && rank !=null){
|
3578 |
|
|
if (crank.equals(rank)){
|
3579 |
|
|
foundIdentic=true;
|
3580 |
|
|
try{
|
3581 |
6b2c2013
|
p.kelbert
|
if(!isSynonym) {
|
3582 |
|
|
tmptaxonbase=tmpb;
|
3583 |
|
|
} else {
|
3584 |
|
|
tmptaxonbase=tmpb;
|
3585 |
|
|
}
|
3586 |
6c41cae2
|
p.kelbert
|
break;
|
3587 |
2b5a7979
|
p.kelbert
|
}catch(Exception e){
|
3588 |
|
|
e.printStackTrace();
|
3589 |
|
|
}
|
3590 |
|
|
}
|
3591 |
|
|
}
|
3592 |
|
|
}
|
3593 |
|
|
}
|
3594 |
|
|
}
|
3595 |
|
|
}
|
3596 |
|
|
boolean statusMatch=false;
|
3597 |
|
|
boolean appendedMatch=false;
|
3598 |
6b2c2013
|
p.kelbert
|
if(tmptaxonbase !=null && foundIdentic){
|
3599 |
|
|
statusMatch=compareStatus(tmptaxonbase, statusType);
|
3600 |
|
|
if (!getStatus().isEmpty() && ! (tmptaxonbase.getAppendedPhrase() == null)) {
|
3601 |
|
|
appendedMatch=tmptaxonbase.getAppendedPhrase().equals(getStatus());
|
3602 |
2b5a7979
|
p.kelbert
|
}
|
3603 |
6b2c2013
|
p.kelbert
|
if (getStatus().isEmpty() && tmptaxonbase.getAppendedPhrase() == null) {
|
3604 |
2b5a7979
|
p.kelbert
|
appendedMatch=true;
|
3605 |
|
|
}
|
3606 |
|
|
|
3607 |
|
|
}
|
3608 |
6b2c2013
|
p.kelbert
|
if ((tmptaxonbase == null || !foundIdentic) || (tmptaxonbase != null && !statusMatch) || (tmptaxonbase != null && !appendedMatch && !statusMatch)){
|
3609 |
2b5a7979
|
p.kelbert
|
|
3610 |
|
|
NonViralName<?> tnb = getNonViralNameAccNomenclature();
|
3611 |
|
|
tnb.setRank(rank);
|
3612 |
|
|
|
3613 |
|
|
if(statusType != null) {
|
3614 |
|
|
tnb.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
3615 |
|
|
}
|
3616 |
|
|
if(getStatus()!=null) {
|
3617 |
|
|
tnb.setAppendedPhrase(getStatus());
|
3618 |
|
|
}
|
3619 |
|
|
|
3620 |
|
|
tnb.setTitleCache(newName2,true);
|
3621 |
6b2c2013
|
p.kelbert
|
tmptaxonbase = findMatchingTaxon(tnb,refMods);
|
3622 |
|
|
if(tmptaxonbase==null){
|
3623 |
|
|
tmptaxonbase=Taxon.NewInstance(tnb, refMods);
|
3624 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
3625 |
|
|
tmptaxonbase.setSec(configState.getConfig().getSecundum());
|
3626 |
|
|
}
|
3627 |
|
|
// tmptaxonbase.setSec(refMods);
|
3628 |
6b2c2013
|
p.kelbert
|
if(!isSynonym) {
|
3629 |
|
|
classification.addChildTaxon((Taxon)tmptaxonbase, null, null);
|
3630 |
|
|
sourceHandler.addSource(refMods, (Taxon)tmptaxonbase);
|
3631 |
|
|
}
|
3632 |
2b5a7979
|
p.kelbert
|
}
|
3633 |
|
|
}
|
3634 |
6b2c2013
|
p.kelbert
|
if(!isSynonym) {
|
3635 |
|
|
tmptaxonbase = CdmBase.deproxy(tmptaxonbase, Taxon.class);
|
3636 |
|
|
} else {
|
3637 |
|
|
tmptaxonbase = CdmBase.deproxy(tmptaxonbase, Synonym.class);
|
3638 |
|
|
}
|
3639 |
2b5a7979
|
p.kelbert
|
if (author != null) {
|
3640 |
|
|
if (!getIdentifier().isEmpty() && (getIdentifier().length()>2)){
|
3641 |
6b2c2013
|
p.kelbert
|
setLSID(getIdentifier(), tmptaxonbase);
|
3642 |
|
|
importer.getTaxonService().saveOrUpdate(tmptaxonbase);
|
3643 |
|
|
if(!isSynonym) {
|
3644 |
|
|
tmptaxonbase = CdmBase.deproxy(tmptaxonbase, Taxon.class);
|
3645 |
|
|
} else {
|
3646 |
|
|
tmptaxonbase = CdmBase.deproxy(tmptaxonbase, Synonym.class);
|
3647 |
|
|
}
|
3648 |
2b5a7979
|
p.kelbert
|
}
|
3649 |
|
|
}
|
3650 |
6e861bce
|
Andreas Müller
|
TaxonNameBase<?,?> tnb = CdmBase.deproxy(tmptaxonbase.getName(), TaxonNameBase.class);
|
3651 |
2b5a7979
|
p.kelbert
|
|
3652 |
6b2c2013
|
p.kelbert
|
if(!isSynonym) {
|
3653 |
|
|
this.taxon=(Taxon)tmptaxonbase;
|
3654 |
|
|
} else {
|
3655 |
|
|
this.syno=(Synonym)tmptaxonbase;
|
3656 |
|
|
}
|
3657 |
2b5a7979
|
p.kelbert
|
castTaxonNameBase(tnb, taxonnamebase);
|
3658 |
|
|
|
3659 |
|
|
}
|
3660 |
|
|
|
3661 |
|
|
/**
|
3662 |
|
|
*
|
3663 |
|
|
*/
|
3664 |
|
|
public void buildTaxon() {
|
3665 |
6c41cae2
|
p.kelbert
|
//System.out.println("BUILD TAXON");
|
3666 |
|
|
logger.info("buildTaxon");
|
3667 |
2b5a7979
|
p.kelbert
|
NomenclaturalStatusType statusType = null;
|
3668 |
|
|
if (!getStatus().isEmpty()){
|
3669 |
|
|
try {
|
3670 |
|
|
statusType = nomStatusString2NomStatus(getStatus());
|
3671 |
|
|
taxonnamebase.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
3672 |
|
|
} catch (UnknownCdmTypeException e) {
|
3673 |
|
|
addProblematicStatusToFile(getStatus());
|
3674 |
|
|
logger.warn("Problem with status");
|
3675 |
|
|
}
|
3676 |
|
|
}
|
3677 |
|
|
importer.getNameService().save(taxonnamebase);
|
3678 |
6b2c2013
|
p.kelbert
|
|
3679 |
6e861bce
|
Andreas Müller
|
TaxonBase<?> tmptaxonbase;
|
3680 |
6b2c2013
|
p.kelbert
|
if (!isSynonym) {
|
3681 |
|
|
tmptaxonbase =Taxon.NewInstance(taxonnamebase, refMods); //sec set null
|
3682 |
|
|
}
|
3683 |
|
|
else {
|
3684 |
|
|
tmptaxonbase =Synonym.NewInstance(taxonnamebase, refMods); //sec set null
|
3685 |
|
|
}
|
3686 |
2b5a7979
|
p.kelbert
|
boolean exist = false;
|
3687 |
|
|
for (TaxonNode p : classification.getAllNodes()){
|
3688 |
|
|
try{
|
3689 |
6b2c2013
|
p.kelbert
|
if(p.getTaxon().getTitleCache().equalsIgnoreCase(tmptaxonbase.getTitleCache())) {
|
3690 |
2b5a7979
|
p.kelbert
|
if(compareStatus(p.getTaxon(), statusType)){
|
3691 |
6b2c2013
|
p.kelbert
|
try{
|
3692 |
|
|
if (!isSynonym) {
|
3693 |
|
|
tmptaxonbase=CdmBase.deproxy(p.getTaxon(), Taxon.class);
|
3694 |
|
|
} else {
|
3695 |
|
|
tmptaxonbase=CdmBase.deproxy(p.getTaxon(), Synonym.class);
|
3696 |
|
|
}
|
3697 |
|
|
exist =true;
|
3698 |
|
|
}catch(Exception e){
|
3699 |
|
|
logger.warn("Found the same name but from another type (taxon/synonym)");
|
3700 |
6e861bce
|
Andreas Müller
|
TaxonNameBase<?,?> existingTnb = getTaxon().getName();
|
3701 |
6b2c2013
|
p.kelbert
|
if (isSynonym){
|
3702 |
|
|
tmptaxonbase = new Synonym(existingTnb, refMods);
|
3703 |
|
|
importer.getTaxonService().saveOrUpdate(tmptaxonbase);
|
3704 |
|
|
tmptaxonbase=CdmBase.deproxy(tmptaxonbase, Synonym.class);
|
3705 |
|
|
exist =true;
|
3706 |
|
|
}
|
3707 |
|
|
else{
|
3708 |
|
|
tmptaxonbase = new Taxon(existingTnb, refMods);
|
3709 |
|
|
}
|
3710 |
|
|
}
|
3711 |
2b5a7979
|
p.kelbert
|
}
|
3712 |
|
|
}
|
3713 |
|
|
}catch(NullPointerException n){logger.warn(" A taxon is either null or its titlecache is null - ignore it?");}
|
3714 |
|
|
}
|
3715 |
|
|
if (!exist){
|
3716 |
|
|
|
3717 |
|
|
boolean insertAsExisting =false;
|
3718 |
6c41cae2
|
p.kelbert
|
List<Taxon> existingTaxons=new ArrayList<Taxon>();
|
3719 |
|
|
try {
|
3720 |
|
|
existingTaxons = getMatchingTaxon(taxonnamebase);
|
3721 |
|
|
} catch (Exception e1) {
|
3722 |
|
|
// TODO Auto-generated catch block
|
3723 |
|
|
e1.printStackTrace();
|
3724 |
|
|
}
|
3725 |
2b5a7979
|
p.kelbert
|
double similarityScore=0.0;
|
3726 |
6c41cae2
|
p.kelbert
|
double similarityAuthor=-1;
|
3727 |
|
|
String author1="";
|
3728 |
|
|
String author2="";
|
3729 |
|
|
String t1="";
|
3730 |
|
|
String t2="";
|
3731 |
2b5a7979
|
p.kelbert
|
for (Taxon bestMatchingTaxon:existingTaxons){
|
3732 |
6c41cae2
|
p.kelbert
|
//System.out.println("tnbase "+taxonnamebase.getTitleCache());
|
3733 |
|
|
//System.out.println("bestex "+bestMatchingTaxon.getTitleCache());
|
3734 |
|
|
try {
|
3735 |
|
|
if(taxonnamebase.getAuthorshipCache()!=null) {
|
3736 |
|
|
author1=taxonnamebase.getAuthorshipCache();
|
3737 |
|
|
}
|
3738 |
|
|
} catch (Exception e) {
|
3739 |
|
|
// TODO Auto-generated catch block
|
3740 |
|
|
e.printStackTrace();
|
3741 |
|
|
}
|
3742 |
|
|
try {
|
3743 |
|
|
if(castTaxonNameBase(bestMatchingTaxon.getName()).getAuthorshipCache()!=null) {
|
3744 |
|
|
author2=castTaxonNameBase(bestMatchingTaxon.getName()).getAuthorshipCache();
|
3745 |
|
|
}
|
3746 |
|
|
} catch (Exception e) {
|
3747 |
|
|
// TODO Auto-generated catch block
|
3748 |
|
|
e.printStackTrace();
|
3749 |
|
|
}
|
3750 |
|
|
try {
|
3751 |
|
|
t1=taxonnamebase.getTitleCache().split("sec.")[0].trim();
|
3752 |
|
|
if (author1!=null && !StringUtils.isEmpty(author1)) {
|
3753 |
|
|
t1=t1.split(Pattern.quote(author1))[0];
|
3754 |
|
|
}
|
3755 |
|
|
} catch (Exception e) {
|
3756 |
|
|
// TODO Auto-generated catch block
|
3757 |
|
|
e.printStackTrace();
|
3758 |
|
|
}
|
3759 |
|
|
try {
|
3760 |
|
|
t2=bestMatchingTaxon.getTitleCache().split("sec.")[0].trim();
|
3761 |
|
|
if (author2!=null && !StringUtils.isEmpty(author2)) {
|
3762 |
|
|
t2=t2.split(Pattern.quote(author2))[0];
|
3763 |
|
|
}
|
3764 |
|
|
} catch (Exception e) {
|
3765 |
|
|
// TODO Auto-generated catch block
|
3766 |
|
|
e.printStackTrace();
|
3767 |
|
|
}
|
3768 |
|
|
|
3769 |
|
|
similarityScore=similarity(t1.trim(), t2.trim());
|
3770 |
|
|
//System.out.println("taxonscore "+similarityScore);
|
3771 |
|
|
similarityAuthor=similarity(author1.trim(), author2.trim());
|
3772 |
|
|
//System.out.println("authorscore "+similarityAuthor);
|
3773 |
|
|
insertAsExisting = compareAndCheckTaxon(taxonnamebase, refMods, similarityScore, bestMatchingTaxon,similarityAuthor);
|
3774 |
2b5a7979
|
p.kelbert
|
if(insertAsExisting) {
|
3775 |
6b2c2013
|
p.kelbert
|
tmptaxonbase=bestMatchingTaxon;
|
3776 |
2b5a7979
|
p.kelbert
|
break;
|
3777 |
|
|
}
|
3778 |
|
|
}
|
3779 |
|
|
if (!insertAsExisting){
|
3780 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
3781 |
|
|
tmptaxonbase.setSec(configState.getConfig().getSecundum());
|
3782 |
|
|
}
|
3783 |
|
|
|
3784 |
|
|
// tmptaxonbase.setSec(refMods);
|
3785 |
2b5a7979
|
p.kelbert
|
if (taxonnamebase.getRank().equals(configState.getConfig().getMaxRank())) {
|
3786 |
6c41cae2
|
p.kelbert
|
//System.out.println("****************************"+tmptaxonbase);
|
3787 |
6b2c2013
|
p.kelbert
|
if (!isSynonym) {
|
3788 |
|
|
classification.addChildTaxon((Taxon)tmptaxonbase, refMods, null);
|
3789 |
|
|
}
|
3790 |
2b5a7979
|
p.kelbert
|
} else{
|
3791 |
|
|
hierarchy = new HashMap<Rank, Taxon>();
|
3792 |
6c41cae2
|
p.kelbert
|
//System.out.println("LOOK FOR PARENT "+taxonnamebase.toString()+", "+tmptaxonbase.toString());
|
3793 |
6b2c2013
|
p.kelbert
|
if (!isSynonym){
|
3794 |
|
|
lookForParentNode(taxonnamebase,(Taxon)tmptaxonbase, refMods,this);
|
3795 |
6c41cae2
|
p.kelbert
|
//System.out.println("HIERARCHY "+hierarchy);
|
3796 |
6b2c2013
|
p.kelbert
|
Taxon parent = buildHierarchy();
|
3797 |
|
|
if(!taxonExistsInClassification(parent,(Taxon)tmptaxonbase)){
|
3798 |
6c41cae2
|
p.kelbert
|
if(parent !=null) {
|
3799 |
|
|
classification.addParentChild(parent, (Taxon)tmptaxonbase, refMods, null);
|
3800 |
|
|
} else {
|
3801 |
|
|
classification.addChildTaxon((Taxon)tmptaxonbase, refMods, null);
|
3802 |
|
|
}
|
3803 |
6b2c2013
|
p.kelbert
|
importer.getClassificationService().saveOrUpdate(classification);
|
3804 |
|
|
}
|
3805 |
2b5a7979
|
p.kelbert
|
}
|
3806 |
|
|
// Set<TaxonNode> nodeList = classification.getAllNodes();
|
3807 |
|
|
// for(TaxonNode tn:nodeList) {
|
3808 |
|
|
// System.out.println(tn.getTaxon());
|
3809 |
|
|
// }
|
3810 |
|
|
}
|
3811 |
|
|
}
|
3812 |
|
|
importer.getClassificationService().saveOrUpdate(classification);
|
3813 |
|
|
// refreshTransaction();
|
3814 |
6b2c2013
|
p.kelbert
|
if(isSynonym) {
|
3815 |
|
|
try{
|
3816 |
|
|
Synonym castTest=CdmBase.deproxy(tmptaxonbase, Synonym.class);
|
3817 |
|
|
}catch(Exception e){
|
3818 |
6e861bce
|
Andreas Müller
|
TaxonNameBase<?,?> existingTnb = tmptaxonbase.getName();
|
3819 |
6b2c2013
|
p.kelbert
|
Synonym castTest = new Synonym(existingTnb, refMods);
|
3820 |
|
|
importer.getTaxonService().saveOrUpdate(castTest);
|
3821 |
|
|
tmptaxonbase=CdmBase.deproxy(castTest, Synonym.class);
|
3822 |
|
|
}
|
3823 |
|
|
}
|
3824 |
|
|
}
|
3825 |
|
|
if(!isSynonym) {
|
3826 |
|
|
taxon=CdmBase.deproxy(tmptaxonbase, Taxon.class);
|
3827 |
|
|
} else {
|
3828 |
|
|
syno=CdmBase.deproxy(tmptaxonbase, Synonym.class);
|
3829 |
2b5a7979
|
p.kelbert
|
}
|
3830 |
6b2c2013
|
p.kelbert
|
|
3831 |
|
|
|
3832 |
2b5a7979
|
p.kelbert
|
|
3833 |
|
|
}
|
3834 |
|
|
|
3835 |
|
|
|
3836 |
|
|
/**
|
3837 |
|
|
*
|
3838 |
|
|
*/
|
3839 |
|
|
private Taxon buildHierarchy() {
|
3840 |
6c41cae2
|
p.kelbert
|
logger.info("buildHierarchy");
|
3841 |
2b5a7979
|
p.kelbert
|
Taxon higherTaxon = null;
|
3842 |
6c41cae2
|
p.kelbert
|
//add the maxRank as a root
|
3843 |
2b5a7979
|
p.kelbert
|
if(hierarchy.containsKey(configState.getConfig().getMaxRank())){
|
3844 |
6c41cae2
|
p.kelbert
|
Taxon ct=hierarchy.get(configState.getConfig().getMaxRank());
|
3845 |
|
|
if(!taxonExistsInClassification(higherTaxon, ct)) {
|
3846 |
|
|
//System.out.println("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"+hierarchy.get(configState.getConfig().getMaxRank()));
|
3847 |
|
|
classification.addChildTaxon(ct, refMods, null);
|
3848 |
2b5a7979
|
p.kelbert
|
}
|
3849 |
|
|
higherTaxon = hierarchy.get(configState.getConfig().getMaxRank());
|
3850 |
6c41cae2
|
p.kelbert
|
// return higherTaxon;
|
3851 |
2b5a7979
|
p.kelbert
|
}
|
3852 |
6c41cae2
|
p.kelbert
|
//add the relation to the highertaxon, except if the current rank to add IS the maxRank
|
3853 |
|
|
if(hierarchy.containsKey(Rank.SUBFAMILY()) && !configState.getConfig().getMaxRank().equals(Rank.SUBFAMILY())){
|
3854 |
|
|
higherTaxon=saveAndGetHigherTaxon(Rank.SUBFAMILY(),higherTaxon);
|
3855 |
2b5a7979
|
p.kelbert
|
}
|
3856 |
6c41cae2
|
p.kelbert
|
if(hierarchy.containsKey(Rank.TRIBE())&& !configState.getConfig().getMaxRank().equals(Rank.TRIBE())){
|
3857 |
|
|
higherTaxon=saveAndGetHigherTaxon(Rank.TRIBE(),higherTaxon);
|
3858 |
2b5a7979
|
p.kelbert
|
}
|
3859 |
6c41cae2
|
p.kelbert
|
if(hierarchy.containsKey(Rank.SUBTRIBE())&& !configState.getConfig().getMaxRank().equals(Rank.SUBTRIBE())){
|
3860 |
|
|
higherTaxon=saveAndGetHigherTaxon(Rank.SUBTRIBE(),higherTaxon);
|
3861 |
2b5a7979
|
p.kelbert
|
}
|
3862 |
6c41cae2
|
p.kelbert
|
if(hierarchy.containsKey(Rank.GENUS())&& !configState.getConfig().getMaxRank().equals(Rank.SUBGENUS())){
|
3863 |
|
|
higherTaxon=saveAndGetHigherTaxon(Rank.GENUS(),higherTaxon);
|
3864 |
2b5a7979
|
p.kelbert
|
}
|
3865 |
6c41cae2
|
p.kelbert
|
if(hierarchy.containsKey(Rank.SUBGENUS())&& !configState.getConfig().getMaxRank().equals(Rank.SUBGENUS())){
|
3866 |
|
|
higherTaxon=saveAndGetHigherTaxon(Rank.SUBGENUS(),higherTaxon);
|
3867 |
2b5a7979
|
p.kelbert
|
}
|
3868 |
|
|
importer.getClassificationService().saveOrUpdate(classification);
|
3869 |
|
|
return higherTaxon;
|
3870 |
|
|
}
|
3871 |
|
|
|
3872 |
6c41cae2
|
p.kelbert
|
private Taxon saveAndGetHigherTaxon(Rank r, Taxon higherTaxon){
|
3873 |
|
|
Taxon ct=hierarchy.get(r);
|
3874 |
|
|
if(!taxonExistsInClassification(higherTaxon,ct )) {
|
3875 |
|
|
if(higherTaxon != null && ct!=null) {
|
3876 |
|
|
classification.addParentChild(higherTaxon, ct, refMods, null);
|
3877 |
|
|
} else
|
3878 |
|
|
if(higherTaxon == null && ct !=null) {
|
3879 |
|
|
classification.addChildTaxon(ct, refMods, null);
|
3880 |
|
|
}
|
3881 |
|
|
}
|
3882 |
|
|
return ct;
|
3883 |
|
|
}
|
3884 |
|
|
|
3885 |
2b5a7979
|
p.kelbert
|
private boolean taxonExistsInClassification(Taxon parent, Taxon child){
|
3886 |
6c41cae2
|
p.kelbert
|
logger.info("taxonExistsInClassification");
|
3887 |
2b5a7979
|
p.kelbert
|
// System.out.println("LOOK IF TAXA EXIST "+parent+", "+child);
|
3888 |
|
|
boolean found=false;
|
3889 |
|
|
if(parent !=null){
|
3890 |
|
|
for (TaxonNode p : classification.getAllNodes()){
|
3891 |
|
|
if(p.getTaxon().getTitleCache().equalsIgnoreCase(parent.getTitleCache())) {
|
3892 |
|
|
for (TaxonNode c : p.getChildNodes()) {
|
3893 |
|
|
if (c.getTaxon().getTitleCache().equalsIgnoreCase(child.getTitleCache())) {
|
3894 |
|
|
found=true;
|
3895 |
|
|
break;
|
3896 |
|
|
}
|
3897 |
|
|
}
|
3898 |
|
|
}
|
3899 |
|
|
}
|
3900 |
|
|
}
|
3901 |
|
|
else{
|
3902 |
|
|
for (TaxonNode p : classification.getAllNodes()){
|
3903 |
|
|
if(p.getTaxon().getTitleCache().equalsIgnoreCase(child.getTitleCache())) {
|
3904 |
|
|
found=true;
|
3905 |
|
|
break;
|
3906 |
|
|
}
|
3907 |
|
|
}
|
3908 |
|
|
}
|
3909 |
|
|
// System.out.println("LOOK IF TAXA EXIST? "+found);
|
3910 |
|
|
return found;
|
3911 |
|
|
}
|
3912 |
|
|
/**
|
3913 |
|
|
* @param nameToBeFilledTest
|
3914 |
|
|
*/
|
3915 |
|
|
@SuppressWarnings("rawtypes")
|
3916 |
|
|
public void setParsedName(TaxonNameBase nameToBeFilledTest) {
|
3917 |
|
|
this.taxonnamebase = (NonViralName<?>) nameToBeFilledTest;
|
3918 |
|
|
|
3919 |
|
|
}
|
3920 |
|
|
//variety dwcranks:varietyEpithet
|
3921 |
|
|
/**
|
3922 |
|
|
* @return the author
|
3923 |
|
|
*/
|
3924 |
|
|
public String getAuthor() {
|
3925 |
|
|
return author;
|
3926 |
|
|
}
|
3927 |
|
|
/**
|
3928 |
|
|
* @return
|
3929 |
|
|
*/
|
3930 |
|
|
public Taxon getTaxon() {
|
3931 |
|
|
return taxon;
|
3932 |
1de96f38
|
p.kelbert
|
}
|
3933 |
2b5a7979
|
p.kelbert
|
/**
|
3934 |
|
|
* @return
|
3935 |
|
|
*/
|
3936 |
|
|
public NonViralName<?> getTaxonNameBase() {
|
3937 |
|
|
return taxonnamebase;
|
3938 |
|
|
}
|
3939 |
|
|
|
3940 |
|
|
/**
|
3941 |
|
|
* @param findOrCreateTaxon
|
3942 |
|
|
*/
|
3943 |
|
|
public void setForm(Taxon form) {
|
3944 |
|
|
this.form=form;
|
3945 |
1de96f38
|
p.kelbert
|
|
3946 |
2b5a7979
|
p.kelbert
|
}
|
3947 |
|
|
/**
|
3948 |
|
|
* @param findOrCreateTaxon
|
3949 |
|
|
*/
|
3950 |
|
|
public void setVariety(Taxon variety) {
|
3951 |
|
|
this.variety=variety;
|
3952 |
1de96f38
|
p.kelbert
|
|
3953 |
2b5a7979
|
p.kelbert
|
}
|
3954 |
|
|
/**
|
3955 |
|
|
* @param string
|
3956 |
|
|
* @return
|
3957 |
|
|
*/
|
3958 |
|
|
@SuppressWarnings("rawtypes")
|
3959 |
|
|
public Taxon findOrCreateTaxon(String partialname,String fullname, Rank rank, Rank globalrank) {
|
3960 |
6c41cae2
|
p.kelbert
|
logger.info("findOrCreateTaxon");
|
3961 |
6b2c2013
|
p.kelbert
|
sourceUrlRef=CdmBase.deproxy(sourceUrlRef, Reference.class);
|
3962 |
6c41cae2
|
p.kelbert
|
//takes too much time
|
3963 |
|
|
// List<TaxonBase> tmpList = importer.getTaxonService().list(Taxon.class, 0, 0, null, null);
|
3964 |
|
|
// logger.info("tmpList returned: "+tmpList.size());
|
3965 |
2b5a7979
|
p.kelbert
|
|
3966 |
|
|
NomenclaturalStatusType statusType = null;
|
3967 |
|
|
if (!getStatus().isEmpty()){
|
3968 |
|
|
try {
|
3969 |
|
|
statusType = nomStatusString2NomStatus(getStatus());
|
3970 |
|
|
} catch (UnknownCdmTypeException e) {
|
3971 |
|
|
addProblematicStatusToFile(getStatus());
|
3972 |
|
|
logger.warn("Problem with status");
|
3973 |
|
|
}
|
3974 |
|
|
}
|
3975 |
|
|
|
3976 |
6c41cae2
|
p.kelbert
|
List<TaxonBase> tmpListFiltered = new ArrayList<TaxonBase>();
|
3977 |
|
|
|
3978 |
|
|
Pager<TaxonBase> taxontest = importer.getTaxonService().findByTitle(TaxonBase.class, fullname, MatchMode.BEGINNING, null, null, null, null, null);
|
3979 |
|
|
|
3980 |
|
|
tmpListFiltered.addAll(taxontest.getRecords());
|
3981 |
|
|
taxontest = importer.getTaxonService().findByTitle(TaxonBase.class, partialname, MatchMode.BEGINNING, null, null, null, null, null);
|
3982 |
|
|
tmpListFiltered.addAll(taxontest.getRecords());
|
3983 |
|
|
|
3984 |
|
|
//logger.info("tmpListFiltered returned: "+tmpListFiltered.size());
|
3985 |
|
|
|
3986 |
|
|
boolean nameCorrected=false;
|
3987 |
|
|
if (fullname.indexOf(partialname)<0) {
|
3988 |
|
|
nameCorrected=true;
|
3989 |
|
|
}
|
3990 |
|
|
|
3991 |
2b5a7979
|
p.kelbert
|
boolean foundIdentic=false;
|
3992 |
|
|
Taxon tmp=null;
|
3993 |
|
|
// Taxon tmpPartial=null;
|
3994 |
6c41cae2
|
p.kelbert
|
for (TaxonBase tmpb:tmpListFiltered){
|
3995 |
2b5a7979
|
p.kelbert
|
if(tmpb !=null){
|
3996 |
|
|
TaxonNameBase tnb = tmpb.getName();
|
3997 |
|
|
Rank crank=null;
|
3998 |
|
|
if (tnb != null){
|
3999 |
6c41cae2
|
p.kelbert
|
// //System.out.println(tnb.getTitleCache());
|
4000 |
2b5a7979
|
p.kelbert
|
// if (tnb.getTitleCache().split("sec.")[0].equals(partialname) ||tnb.getTitleCache().split("sec.")[0].equals(fullname) ){
|
4001 |
|
|
if(globalrank.equals(rank) || (globalrank.isLower(Rank.SPECIES()) && rank.equals(Rank.SPECIES()))){
|
4002 |
6c41cae2
|
p.kelbert
|
if (tnb.getTitleCache().split("sec.")[0].trim().equalsIgnoreCase(fullname) ){
|
4003 |
2b5a7979
|
p.kelbert
|
crank =tnb.getRank();
|
4004 |
|
|
if (crank !=null && rank !=null){
|
4005 |
|
|
if (crank.equals(rank)){
|
4006 |
|
|
foundIdentic=true;
|
4007 |
|
|
try{
|
4008 |
|
|
tmp=(Taxon)tmpb;
|
4009 |
6c41cae2
|
p.kelbert
|
break;
|
4010 |
2b5a7979
|
p.kelbert
|
}catch(Exception e){
|
4011 |
|
|
e.printStackTrace();
|
4012 |
|
|
}
|
4013 |
|
|
}
|
4014 |
|
|
}
|
4015 |
|
|
}
|
4016 |
6c41cae2
|
p.kelbert
|
if(nameCorrected){ //for corrected names such as Anochetus -- A. blf-pat
|
4017 |
|
|
if (tnb.getTitleCache().split("sec.")[0].trim().equalsIgnoreCase(partialname) ){
|
4018 |
2b5a7979
|
p.kelbert
|
crank =tnb.getRank();
|
4019 |
|
|
if (crank !=null && rank !=null){
|
4020 |
|
|
if (crank.equals(rank)){
|
4021 |
|
|
foundIdentic=true;
|
4022 |
|
|
try{
|
4023 |
|
|
tmp=(Taxon)tmpb;
|
4024 |
6c41cae2
|
p.kelbert
|
break;
|
4025 |
2b5a7979
|
p.kelbert
|
}catch(Exception e){
|
4026 |
|
|
e.printStackTrace();
|
4027 |
|
|
}
|
4028 |
|
|
}
|
4029 |
|
|
}
|
4030 |
|
|
}
|
4031 |
|
|
}
|
4032 |
|
|
}
|
4033 |
|
|
else{
|
4034 |
6c41cae2
|
p.kelbert
|
if (tnb.getTitleCache().split("sec.")[0].trim().equalsIgnoreCase(partialname) ){
|
4035 |
2b5a7979
|
p.kelbert
|
crank =tnb.getRank();
|
4036 |
|
|
if (crank !=null && rank !=null){
|
4037 |
|
|
if (crank.equals(rank)){
|
4038 |
|
|
foundIdentic=true;
|
4039 |
|
|
try{
|
4040 |
|
|
tmp=(Taxon)tmpb;
|
4041 |
6c41cae2
|
p.kelbert
|
break;
|
4042 |
2b5a7979
|
p.kelbert
|
}catch(Exception e){
|
4043 |
|
|
e.printStackTrace();
|
4044 |
|
|
}
|
4045 |
|
|
}
|
4046 |
|
|
}
|
4047 |
|
|
}
|
4048 |
|
|
}
|
4049 |
|
|
}
|
4050 |
|
|
}
|
4051 |
|
|
}
|
4052 |
|
|
boolean statusMatch=false;
|
4053 |
|
|
boolean appendedMatch=false;
|
4054 |
|
|
if(tmp !=null && foundIdentic){
|
4055 |
|
|
statusMatch=compareStatus(tmp, statusType);
|
4056 |
|
|
if (!getStatus().isEmpty() && ! (tmp.getAppendedPhrase() == null)) {
|
4057 |
|
|
appendedMatch=tmp.getAppendedPhrase().equals(getStatus());
|
4058 |
|
|
}
|
4059 |
|
|
if (getStatus().isEmpty() && tmp.getAppendedPhrase() == null) {
|
4060 |
|
|
appendedMatch=true;
|
4061 |
|
|
}
|
4062 |
|
|
|
4063 |
|
|
}
|
4064 |
|
|
if ((tmp == null || !foundIdentic) || (tmp != null && !statusMatch) || (tmp != null && !appendedMatch && !statusMatch)){
|
4065 |
|
|
|
4066 |
|
|
NonViralName<?> tnb = getNonViralNameAccNomenclature();
|
4067 |
|
|
tnb.setRank(rank);
|
4068 |
|
|
|
4069 |
|
|
if(statusType != null) {
|
4070 |
|
|
tnb.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
4071 |
|
|
}
|
4072 |
|
|
if(getStatus()!=null) {
|
4073 |
|
|
tnb.setAppendedPhrase(getStatus());
|
4074 |
|
|
}
|
4075 |
|
|
|
4076 |
6c41cae2
|
p.kelbert
|
if(rank.equals(Rank.UNKNOWN_RANK())){
|
4077 |
d03aed8f
|
Andreas Müller
|
tnb.setTitleCache(fullname, true);
|
4078 |
6c41cae2
|
p.kelbert
|
// tnb.setGenusOrUninomial(fullname);
|
4079 |
|
|
}
|
4080 |
|
|
if(rank.isHigher(Rank.GENUS())) {
|
4081 |
|
|
tnb.setGenusOrUninomial(partialname);
|
4082 |
2b5a7979
|
p.kelbert
|
}
|
4083 |
|
|
|
4084 |
|
|
if(rank.isHigher(Rank.SPECIES())) {
|
4085 |
d03aed8f
|
Andreas Müller
|
tnb.setTitleCache(partialname, true);
|
4086 |
2b5a7979
|
p.kelbert
|
}
|
4087 |
|
|
|
4088 |
|
|
if (rank.equals(globalrank) && author != null) {
|
4089 |
|
|
if(fullname.indexOf("opulifolium")>-1) {
|
4090 |
6c41cae2
|
p.kelbert
|
//System.out.println("AUTOR: "+author);
|
4091 |
2b5a7979
|
p.kelbert
|
}
|
4092 |
|
|
tnb.setCombinationAuthorTeam(findOrCreateAuthor(author));
|
4093 |
|
|
if (getIdentifier() !=null && !getIdentifier().isEmpty()){
|
4094 |
|
|
Taxon taxonLSID = getTaxonByLSID(getIdentifier());
|
4095 |
|
|
if (taxonLSID !=null) {
|
4096 |
|
|
tmp=taxonLSID;
|
4097 |
|
|
}
|
4098 |
|
|
}
|
4099 |
|
|
}
|
4100 |
|
|
|
4101 |
|
|
if(tmp == null){
|
4102 |
|
|
if (rank.equals(Rank.FAMILY())) {
|
4103 |
|
|
tmp = buildFamily(tnb);
|
4104 |
|
|
}
|
4105 |
|
|
if (rank.equals(Rank.SUBFAMILY())) {
|
4106 |
|
|
tmp = buildSubfamily(tnb);
|
4107 |
|
|
}
|
4108 |
|
|
if (rank.equals(Rank.TRIBE())) {
|
4109 |
|
|
tmp = buildTribe(tnb);
|
4110 |
|
|
}
|
4111 |
|
|
if (rank.equals(Rank.SUBTRIBE())) {
|
4112 |
|
|
tmp = buildSubtribe(tnb);
|
4113 |
|
|
}
|
4114 |
|
|
if (rank.equals(Rank.GENUS())) {
|
4115 |
|
|
tmp = buildGenus(partialname, tnb);
|
4116 |
|
|
}
|
4117 |
|
|
|
4118 |
|
|
if (rank.equals(Rank.SUBGENUS())) {
|
4119 |
|
|
tmp = buildSubgenus(partialname, tnb);
|
4120 |
|
|
}
|
4121 |
|
|
if (rank.equals(Rank.SPECIES())) {
|
4122 |
|
|
tmp = buildSpecies(partialname, tnb);
|
4123 |
|
|
}
|
4124 |
|
|
|
4125 |
|
|
if (rank.equals(Rank.SUBSPECIES())) {
|
4126 |
|
|
tmp = buildSubspecies(partialname, tnb);
|
4127 |
|
|
}
|
4128 |
|
|
|
4129 |
|
|
if (rank.equals(Rank.VARIETY())) {
|
4130 |
|
|
tmp = buildVariety(fullname, partialname, tnb);
|
4131 |
|
|
}
|
4132 |
|
|
|
4133 |
|
|
if (rank.equals(Rank.FORM())) {
|
4134 |
|
|
tmp = buildForm(fullname, partialname, tnb);
|
4135 |
|
|
}
|
4136 |
|
|
|
4137 |
|
|
importer.getClassificationService().saveOrUpdate(classification);
|
4138 |
|
|
}
|
4139 |
|
|
}
|
4140 |
|
|
|
4141 |
|
|
tmp = CdmBase.deproxy(tmp, Taxon.class);
|
4142 |
|
|
if (rank.equals(globalrank) && author != null) {
|
4143 |
|
|
if (!getIdentifier().isEmpty() && (getIdentifier().length()>2)){
|
4144 |
|
|
setLSID(getIdentifier(), tmp);
|
4145 |
|
|
importer.getTaxonService().saveOrUpdate(tmp);
|
4146 |
|
|
tmp = CdmBase.deproxy(tmp, Taxon.class);
|
4147 |
|
|
}
|
4148 |
|
|
}
|
4149 |
|
|
TaxonNameBase tnb = CdmBase.deproxy(tmp.getName(), TaxonNameBase.class);
|
4150 |
|
|
|
4151 |
|
|
this.taxon=tmp;
|
4152 |
|
|
castTaxonNameBase(tnb, taxonnamebase);
|
4153 |
|
|
return tmp;
|
4154 |
|
|
}
|
4155 |
|
|
/**
|
4156 |
|
|
* @param tnb
|
4157 |
|
|
* @return
|
4158 |
|
|
*/
|
4159 |
|
|
private Taxon buildSubfamily(NonViralName<?> tnb) {
|
4160 |
|
|
Taxon tmp;
|
4161 |
6c41cae2
|
p.kelbert
|
// tnb.generateTitle();
|
4162 |
2b5a7979
|
p.kelbert
|
tmp = findMatchingTaxon(tnb,refMods);
|
4163 |
|
|
if(tmp ==null){
|
4164 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4165 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
4166 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4167 |
|
|
}
|
4168 |
|
|
// tmp.setSec(refMods);
|
4169 |
|
|
// sourceHandler.addSource(refMods, tmp);
|
4170 |
2b5a7979
|
p.kelbert
|
if(family != null) {
|
4171 |
|
|
classification.addParentChild(family, tmp, null, null);
|
4172 |
|
|
higherRank=Rank.FAMILY();
|
4173 |
|
|
higherTaxa=family;
|
4174 |
|
|
} else {
|
4175 |
6c41cae2
|
p.kelbert
|
//System.out.println("ADDCHILDTAXON SUBFAMILY "+tmp);
|
4176 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4177 |
|
|
}
|
4178 |
|
|
}
|
4179 |
|
|
return tmp;
|
4180 |
|
|
}
|
4181 |
|
|
/**
|
4182 |
|
|
* @param tnb
|
4183 |
|
|
* @return
|
4184 |
|
|
*/
|
4185 |
|
|
private Taxon buildFamily(NonViralName<?> tnb) {
|
4186 |
|
|
Taxon tmp;
|
4187 |
6c41cae2
|
p.kelbert
|
// tnb.generateTitle();
|
4188 |
2b5a7979
|
p.kelbert
|
tmp = findMatchingTaxon(tnb,refMods);
|
4189 |
|
|
if(tmp ==null){
|
4190 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4191 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
4192 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4193 |
|
|
}
|
4194 |
|
|
// tmp.setSec(refMods);
|
4195 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4196 |
|
|
//System.out.println("ADDCHILDTAXON FAMILY "+tmp);
|
4197 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4198 |
|
|
}
|
4199 |
|
|
return tmp;
|
4200 |
|
|
}
|
4201 |
|
|
/**
|
4202 |
|
|
* @param fullname
|
4203 |
|
|
* @param tnb
|
4204 |
|
|
* @return
|
4205 |
|
|
*/
|
4206 |
|
|
private Taxon buildForm(String fullname, String partialname, NonViralName<?> tnb) {
|
4207 |
|
|
Taxon tmp;
|
4208 |
|
|
if (genusName !=null) {
|
4209 |
|
|
tnb.setGenusOrUninomial(genusName.getGenusOrUninomial());
|
4210 |
|
|
}
|
4211 |
|
|
if (subgenusName !=null) {
|
4212 |
|
|
tnb.setInfraGenericEpithet(subgenusName.getInfraGenericEpithet());
|
4213 |
|
|
}
|
4214 |
|
|
if(speciesName !=null) {
|
4215 |
|
|
tnb.setSpecificEpithet(speciesName.getSpecificEpithet());
|
4216 |
|
|
}
|
4217 |
|
|
if(subspeciesName != null) {
|
4218 |
|
|
tnb.setInfraSpecificEpithet(subspeciesName.getInfraSpecificEpithet());
|
4219 |
|
|
}
|
4220 |
|
|
if(partialname!= null) {
|
4221 |
|
|
tnb.setInfraSpecificEpithet(partialname);
|
4222 |
|
|
}
|
4223 |
|
|
tnb.generateTitle();
|
4224 |
|
|
//TODO how to save form??
|
4225 |
|
|
tnb.setTitleCache(fullname, true);
|
4226 |
|
|
tmp = findMatchingTaxon(tnb,refMods);
|
4227 |
|
|
if(tmp ==null){
|
4228 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4229 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
4230 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4231 |
|
|
}
|
4232 |
|
|
// tmp.setSec(refMods);
|
4233 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4234 |
2b5a7979
|
p.kelbert
|
if (subspecies !=null) {
|
4235 |
|
|
classification.addParentChild(subspecies, tmp, null, null);
|
4236 |
|
|
higherRank=Rank.SUBSPECIES();
|
4237 |
|
|
higherTaxa=subspecies;
|
4238 |
|
|
} else {
|
4239 |
|
|
if (species !=null) {
|
4240 |
|
|
classification.addParentChild(species, tmp, null, null);
|
4241 |
|
|
higherRank=Rank.SPECIES();
|
4242 |
|
|
higherTaxa=species;
|
4243 |
|
|
}
|
4244 |
|
|
else{
|
4245 |
6c41cae2
|
p.kelbert
|
// System.out.println("ADDCHILDTAXON FORM "+tmp);
|
4246 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4247 |
|
|
}
|
4248 |
|
|
}
|
4249 |
|
|
}
|
4250 |
|
|
return tmp;
|
4251 |
|
|
}
|
4252 |
|
|
/**
|
4253 |
|
|
* @param fullname
|
4254 |
|
|
* @param tnb
|
4255 |
|
|
* @return
|
4256 |
|
|
*/
|
4257 |
|
|
private Taxon buildVariety(String fullname, String partialname, NonViralName<?> tnb) {
|
4258 |
|
|
Taxon tmp;
|
4259 |
|
|
if (genusName !=null) {
|
4260 |
|
|
tnb.setGenusOrUninomial(genusName.getGenusOrUninomial());
|
4261 |
|
|
}
|
4262 |
|
|
if (subgenusName !=null) {
|
4263 |
|
|
tnb.setInfraGenericEpithet(subgenusName.getInfraGenericEpithet());
|
4264 |
|
|
}
|
4265 |
|
|
if(speciesName !=null) {
|
4266 |
|
|
tnb.setSpecificEpithet(speciesName.getSpecificEpithet());
|
4267 |
|
|
}
|
4268 |
|
|
if(subspeciesName != null) {
|
4269 |
|
|
tnb.setInfraSpecificEpithet(subspeciesName.getSpecificEpithet());
|
4270 |
|
|
}
|
4271 |
|
|
if(partialname != null) {
|
4272 |
|
|
tnb.setInfraSpecificEpithet(partialname);
|
4273 |
|
|
}
|
4274 |
|
|
//TODO how to save variety?
|
4275 |
|
|
tnb.setTitleCache(fullname, true);
|
4276 |
|
|
tmp = findMatchingTaxon(tnb,refMods);
|
4277 |
|
|
if(tmp ==null){
|
4278 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4279 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
4280 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4281 |
|
|
}
|
4282 |
|
|
// tmp.setSec(refMods);
|
4283 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4284 |
2b5a7979
|
p.kelbert
|
if (subspecies !=null) {
|
4285 |
|
|
classification.addParentChild(subspecies, tmp, null, null);
|
4286 |
|
|
higherRank=Rank.SUBSPECIES();
|
4287 |
|
|
higherTaxa=subspecies;
|
4288 |
|
|
} else {
|
4289 |
|
|
if(species !=null) {
|
4290 |
|
|
classification.addParentChild(species, tmp, null, null);
|
4291 |
|
|
higherRank=Rank.SPECIES();
|
4292 |
|
|
higherTaxa=species;
|
4293 |
|
|
}
|
4294 |
|
|
else{
|
4295 |
6c41cae2
|
p.kelbert
|
//System.out.println("ADDCHILDTAXON VARIETY "+tmp);
|
4296 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4297 |
|
|
}
|
4298 |
|
|
}
|
4299 |
|
|
}
|
4300 |
|
|
return tmp;
|
4301 |
|
|
}
|
4302 |
|
|
/**
|
4303 |
|
|
* @param partialname
|
4304 |
|
|
* @param tnb
|
4305 |
|
|
* @return
|
4306 |
|
|
*/
|
4307 |
|
|
private Taxon buildSubspecies(String partialname, NonViralName<?> tnb) {
|
4308 |
|
|
Taxon tmp;
|
4309 |
|
|
if (genusName !=null) {
|
4310 |
|
|
tnb.setGenusOrUninomial(genusName.getGenusOrUninomial());
|
4311 |
|
|
}
|
4312 |
|
|
if (subgenusName !=null) {
|
4313 |
|
|
// System.out.println("SUB:"+subgenusName.getInfraGenericEpithet());
|
4314 |
|
|
tnb.setInfraGenericEpithet(subgenusName.getInfraGenericEpithet());
|
4315 |
|
|
}
|
4316 |
|
|
if(speciesName !=null) {
|
4317 |
|
|
// System.out.println("SPE:"+speciesName.getSpecificEpithet());
|
4318 |
|
|
tnb.setSpecificEpithet(speciesName.getSpecificEpithet());
|
4319 |
|
|
}
|
4320 |
|
|
tnb.setInfraSpecificEpithet(partialname);
|
4321 |
|
|
tnb.generateTitle();
|
4322 |
|
|
tmp = findMatchingTaxon(tnb,refMods);
|
4323 |
|
|
if(tmp ==null){
|
4324 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4325 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum())
|
4326 |
|
|
{
|
4327 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4328 |
|
|
// tmp.setSec(refMods);
|
4329 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4330 |
|
|
}
|
4331 |
2b5a7979
|
p.kelbert
|
|
4332 |
|
|
if(species != null) {
|
4333 |
|
|
classification.addParentChild(species, tmp, null, null);
|
4334 |
|
|
higherRank=Rank.SPECIES();
|
4335 |
|
|
higherTaxa=species;
|
4336 |
|
|
}
|
4337 |
|
|
else{
|
4338 |
6c41cae2
|
p.kelbert
|
//System.out.println("ADDCHILDTAXON SUBSPECIES "+tmp);
|
4339 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4340 |
|
|
}
|
4341 |
|
|
}
|
4342 |
|
|
return tmp;
|
4343 |
|
|
}
|
4344 |
|
|
/**
|
4345 |
|
|
* @param partialname
|
4346 |
|
|
* @param tnb
|
4347 |
|
|
* @return
|
4348 |
|
|
*/
|
4349 |
|
|
private Taxon buildSpecies(String partialname, NonViralName<?> tnb) {
|
4350 |
|
|
Taxon tmp;
|
4351 |
|
|
if (genusName !=null) {
|
4352 |
|
|
tnb.setGenusOrUninomial(genusName.getGenusOrUninomial());
|
4353 |
|
|
}
|
4354 |
|
|
if (subgenusName !=null) {
|
4355 |
|
|
tnb.setInfraGenericEpithet(subgenusName.getInfraGenericEpithet());
|
4356 |
|
|
}
|
4357 |
|
|
tnb.setSpecificEpithet(partialname.toLowerCase());
|
4358 |
|
|
tnb.generateTitle();
|
4359 |
|
|
tmp = findMatchingTaxon(tnb,refMods);
|
4360 |
|
|
if(tmp ==null){
|
4361 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4362 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
4363 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4364 |
|
|
}
|
4365 |
|
|
// tmp.setSec(refMods);
|
4366 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4367 |
2b5a7979
|
p.kelbert
|
if (subgenus !=null) {
|
4368 |
|
|
classification.addParentChild(subgenus, tmp, null, null);
|
4369 |
|
|
higherRank=Rank.SUBGENUS();
|
4370 |
|
|
higherTaxa=subgenus;
|
4371 |
|
|
} else {
|
4372 |
|
|
if (genus !=null) {
|
4373 |
|
|
classification.addParentChild(genus, tmp, null, null);
|
4374 |
|
|
higherRank=Rank.GENUS();
|
4375 |
|
|
higherTaxa=genus;
|
4376 |
|
|
}
|
4377 |
|
|
else{
|
4378 |
6c41cae2
|
p.kelbert
|
//System.out.println("ADDCHILDTAXON SPECIES "+tmp);
|
4379 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4380 |
|
|
}
|
4381 |
|
|
}
|
4382 |
|
|
}
|
4383 |
|
|
return tmp;
|
4384 |
|
|
}
|
4385 |
|
|
/**
|
4386 |
|
|
* @param partialname
|
4387 |
|
|
* @param tnb
|
4388 |
|
|
* @return
|
4389 |
|
|
*/
|
4390 |
|
|
private Taxon buildSubgenus(String partialname, NonViralName<?> tnb) {
|
4391 |
|
|
Taxon tmp;
|
4392 |
|
|
tnb.setInfraGenericEpithet(partialname);
|
4393 |
|
|
if (genusName !=null) {
|
4394 |
|
|
tnb.setGenusOrUninomial(genusName.getGenusOrUninomial());
|
4395 |
|
|
}
|
4396 |
|
|
tnb.generateTitle();
|
4397 |
|
|
tmp = findMatchingTaxon(tnb,refMods);
|
4398 |
|
|
if(tmp ==null){
|
4399 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4400 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
4401 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4402 |
|
|
}
|
4403 |
|
|
// tmp.setSec(refMods);
|
4404 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4405 |
2b5a7979
|
p.kelbert
|
if(genus != null) {
|
4406 |
|
|
classification.addParentChild(genus, tmp, null, null);
|
4407 |
|
|
higherRank=Rank.GENUS();
|
4408 |
|
|
higherTaxa=genus;
|
4409 |
|
|
} else{
|
4410 |
6c41cae2
|
p.kelbert
|
//System.out.println("ADDCHILDTAXON SUBGENUS "+tmp);
|
4411 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4412 |
|
|
}
|
4413 |
|
|
}
|
4414 |
|
|
return tmp;
|
4415 |
|
|
}
|
4416 |
|
|
/**
|
4417 |
|
|
* @param partialname
|
4418 |
|
|
* @param tnb
|
4419 |
|
|
* @return
|
4420 |
|
|
*/
|
4421 |
|
|
private Taxon buildGenus(String partialname, NonViralName<?> tnb) {
|
4422 |
|
|
Taxon tmp;
|
4423 |
|
|
tnb.setGenusOrUninomial(partialname);
|
4424 |
|
|
tnb.generateTitle();
|
4425 |
|
|
|
4426 |
|
|
tmp = findMatchingTaxon(tnb,refMods);
|
4427 |
|
|
if(tmp ==null){
|
4428 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4429 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum())
|
4430 |
|
|
{
|
4431 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4432 |
|
|
// tmp.setSec(refMods);
|
4433 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4434 |
|
|
}
|
4435 |
2b5a7979
|
p.kelbert
|
|
4436 |
|
|
if(subtribe != null) {
|
4437 |
|
|
classification.addParentChild(subtribe, tmp, null, null);
|
4438 |
|
|
higherRank=Rank.SUBTRIBE();
|
4439 |
|
|
higherTaxa=subtribe;
|
4440 |
|
|
} else{
|
4441 |
|
|
if(tribe !=null) {
|
4442 |
|
|
classification.addParentChild(tribe, tmp, null, null);
|
4443 |
|
|
higherRank=Rank.TRIBE();
|
4444 |
|
|
higherTaxa=tribe;
|
4445 |
|
|
} else{
|
4446 |
|
|
if(subfamily !=null) {
|
4447 |
|
|
classification.addParentChild(subfamily, tmp, null, null);
|
4448 |
|
|
higherRank=Rank.SUBFAMILY();
|
4449 |
|
|
higherTaxa=subfamily;
|
4450 |
|
|
} else
|
4451 |
|
|
if(family !=null) {
|
4452 |
|
|
classification.addParentChild(family, tmp, null, null);
|
4453 |
|
|
higherRank=Rank.FAMILY();
|
4454 |
|
|
higherTaxa=family;
|
4455 |
|
|
}
|
4456 |
|
|
else{
|
4457 |
6c41cae2
|
p.kelbert
|
//System.out.println("ADDCHILDTAXON GENUS "+tmp);
|
4458 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4459 |
|
|
}
|
4460 |
|
|
}
|
4461 |
|
|
}
|
4462 |
|
|
}
|
4463 |
|
|
return tmp;
|
4464 |
|
|
}
|
4465 |
|
|
|
4466 |
|
|
/**
|
4467 |
|
|
* @param tnb
|
4468 |
|
|
* @return
|
4469 |
|
|
*/
|
4470 |
|
|
private Taxon buildSubtribe(NonViralName<?> tnb) {
|
4471 |
|
|
Taxon tmp;
|
4472 |
|
|
tnb.generateTitle();
|
4473 |
|
|
tmp = findMatchingTaxon(tnb,refMods);
|
4474 |
|
|
if(tmp==null){
|
4475 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4476 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
4477 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4478 |
|
|
}
|
4479 |
|
|
// tmp.setSec(refMods);
|
4480 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4481 |
2b5a7979
|
p.kelbert
|
if(tribe != null) {
|
4482 |
|
|
classification.addParentChild(tribe, tmp, null, null);
|
4483 |
|
|
higherRank=Rank.TRIBE();
|
4484 |
|
|
higherTaxa=tribe;
|
4485 |
|
|
} else{
|
4486 |
6c41cae2
|
p.kelbert
|
//System.out.println("ADDCHILDTAXON SUBTRIBE "+tmp);
|
4487 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4488 |
|
|
}
|
4489 |
|
|
}
|
4490 |
|
|
return tmp;
|
4491 |
|
|
}
|
4492 |
|
|
/**
|
4493 |
|
|
* @param tnb
|
4494 |
|
|
* @return
|
4495 |
|
|
*/
|
4496 |
|
|
private Taxon buildTribe(NonViralName<?> tnb) {
|
4497 |
|
|
Taxon tmp;
|
4498 |
|
|
tnb.generateTitle();
|
4499 |
|
|
tmp = findMatchingTaxon(tnb,refMods);
|
4500 |
|
|
if(tmp==null){
|
4501 |
|
|
tmp = Taxon.NewInstance(tnb, sourceUrlRef);
|
4502 |
6c41cae2
|
p.kelbert
|
if(!configState.getConfig().doKeepOriginalSecundum()) {
|
4503 |
|
|
tmp.setSec(configState.getConfig().getSecundum());
|
4504 |
|
|
}
|
4505 |
|
|
// tmp.setSec(refMods);
|
4506 |
|
|
//sourceHandler.addSource(refMods, tmp);
|
4507 |
2b5a7979
|
p.kelbert
|
if (subfamily !=null) {
|
4508 |
|
|
classification.addParentChild(subfamily, tmp, null, null);
|
4509 |
|
|
higherRank=Rank.SUBFAMILY();
|
4510 |
|
|
higherTaxa=subfamily;
|
4511 |
|
|
} else {
|
4512 |
|
|
if(family != null) {
|
4513 |
|
|
classification.addParentChild(family, tmp, null, null);
|
4514 |
|
|
higherRank=Rank.FAMILY();
|
4515 |
|
|
higherTaxa=family;
|
4516 |
|
|
}
|
4517 |
|
|
else{
|
4518 |
6c41cae2
|
p.kelbert
|
//System.out.println("ADDCHILDTAXON TRIBE "+tmp);
|
4519 |
2b5a7979
|
p.kelbert
|
classification.addChildTaxon(tmp, null, null);
|
4520 |
|
|
}
|
4521 |
|
|
}
|
4522 |
|
|
}
|
4523 |
|
|
return tmp;
|
4524 |
|
|
}
|
4525 |
6c41cae2
|
p.kelbert
|
|
4526 |
2b5a7979
|
p.kelbert
|
/**
|
4527 |
|
|
* @param identifier2
|
4528 |
|
|
* @return
|
4529 |
|
|
*/
|
4530 |
|
|
@SuppressWarnings("rawtypes")
|
4531 |
|
|
private Taxon getTaxonByLSID(String identifier) {
|
4532 |
6c41cae2
|
p.kelbert
|
//logger.info("getTaxonByLSID");
|
4533 |
2b5a7979
|
p.kelbert
|
// boolean lsidok=false;
|
4534 |
|
|
String id = identifier.split("__")[0];
|
4535 |
|
|
// String source = identifier.split("__")[1];
|
4536 |
|
|
LSID lsid = null;
|
4537 |
|
|
if (id.indexOf("lsid")>-1){
|
4538 |
|
|
try {
|
4539 |
|
|
lsid = new LSID(id);
|
4540 |
|
|
// lsidok=true;
|
4541 |
|
|
} catch (MalformedLSIDException e) {
|
4542 |
|
|
logger.warn("Malformed LSID");
|
4543 |
|
|
}
|
4544 |
|
|
}
|
4545 |
|
|
if (lsid !=null){
|
4546 |
fdf9eec2
|
Andreas Müller
|
List<Taxon> taxa = importer.getTaxonService().list(Taxon.class, 0, 0, null, null);
|
4547 |
2b5a7979
|
p.kelbert
|
LSID currentlsid=null;
|
4548 |
fdf9eec2
|
Andreas Müller
|
for (Taxon t:taxa){
|
4549 |
2b5a7979
|
p.kelbert
|
currentlsid = t.getLsid();
|
4550 |
|
|
if (currentlsid !=null){
|
4551 |
|
|
if (currentlsid.getLsid().equals(lsid.getLsid())){
|
4552 |
|
|
try{
|
4553 |
|
|
return (Taxon) t;
|
4554 |
|
|
}
|
4555 |
|
|
catch(Exception e){logger.warn("Exception occurred while comparing LSIDs "+e );}
|
4556 |
|
|
}
|
4557 |
|
|
}
|
4558 |
|
|
}
|
4559 |
|
|
}
|
4560 |
|
|
return null;
|
4561 |
|
|
}
|
4562 |
|
|
/**
|
4563 |
|
|
* @param author2
|
4564 |
|
|
* @return
|
4565 |
|
|
*/
|
4566 |
|
|
@SuppressWarnings("rawtypes")
|
4567 |
|
|
private Person findOrCreateAuthor(String author2) {
|
4568 |
6c41cae2
|
p.kelbert
|
//logger.info("findOrCreateAuthor");
|
4569 |
2b5a7979
|
p.kelbert
|
List<UuidAndTitleCache<Person>> hiberPersons = importer.getAgentService().getPersonUuidAndTitleCache();
|
4570 |
|
|
for (UuidAndTitleCache<Person> hibernateP:hiberPersons){
|
4571 |
|
|
if(hibernateP.getTitleCache().equals(author2)) {
|
4572 |
|
|
AgentBase existing = importer.getAgentService().find(hibernateP.getUuid());
|
4573 |
|
|
return CdmBase.deproxy(existing, Person.class);
|
4574 |
|
|
}
|
4575 |
|
|
}
|
4576 |
|
|
Person p = Person.NewInstance();
|
4577 |
|
|
p.setTitleCache(author2,true);
|
4578 |
|
|
importer.getAgentService().saveOrUpdate(p);
|
4579 |
|
|
return CdmBase.deproxy(p, Person.class);
|
4580 |
|
|
}
|
4581 |
|
|
/**
|
4582 |
|
|
* @param author the author to set
|
4583 |
|
|
*/
|
4584 |
|
|
public void setAuthor(String author) {
|
4585 |
|
|
this.author = author;
|
4586 |
|
|
}
|
4587 |
|
|
|
4588 |
|
|
/**
|
4589 |
|
|
* @return the higherTaxa
|
4590 |
|
|
*/
|
4591 |
|
|
public Taxon getHigherTaxa() {
|
4592 |
|
|
return higherTaxa;
|
4593 |
|
|
}
|
4594 |
|
|
/**
|
4595 |
|
|
* @param higherTaxa the higherTaxa to set
|
4596 |
|
|
*/
|
4597 |
|
|
public void setHigherTaxa(Taxon higherTaxa) {
|
4598 |
|
|
this.higherTaxa = higherTaxa;
|
4599 |
|
|
}
|
4600 |
|
|
/**
|
4601 |
|
|
* @return the higherRank
|
4602 |
|
|
*/
|
4603 |
|
|
public Rank getHigherRank() {
|
4604 |
|
|
return higherRank;
|
4605 |
|
|
}
|
4606 |
|
|
/**
|
4607 |
|
|
* @param higherRank the higherRank to set
|
4608 |
|
|
*/
|
4609 |
|
|
public void setHigherRank(Rank higherRank) {
|
4610 |
|
|
this.higherRank = higherRank;
|
4611 |
|
|
}
|
4612 |
|
|
public String getName(){
|
4613 |
|
|
if (newName.isEmpty()) {
|
4614 |
|
|
return originalName;
|
4615 |
|
|
} else {
|
4616 |
|
|
return newName;
|
4617 |
|
|
}
|
4618 |
|
|
|
4619 |
|
|
}
|
4620 |
|
|
/**
|
4621 |
|
|
* @return the fullName
|
4622 |
|
|
*/
|
4623 |
|
|
public String getOriginalName() {
|
4624 |
|
|
return originalName;
|
4625 |
|
|
}
|
4626 |
|
|
/**
|
4627 |
|
|
* @param fullName the fullName to set
|
4628 |
|
|
*/
|
4629 |
|
|
public void setOriginalName(String fullName) {
|
4630 |
|
|
this.originalName = fullName;
|
4631 |
|
|
}
|
4632 |
|
|
/**
|
4633 |
|
|
* @return the newName
|
4634 |
|
|
*/
|
4635 |
|
|
public String getNewName() {
|
4636 |
|
|
return newName;
|
4637 |
|
|
}
|
4638 |
|
|
/**
|
4639 |
|
|
* @param newName the newName to set
|
4640 |
|
|
*/
|
4641 |
|
|
public void setNewName(String newName) {
|
4642 |
|
|
this.newName = newName;
|
4643 |
|
|
}
|
4644 |
|
|
/**
|
4645 |
|
|
* @return the rank
|
4646 |
|
|
*/
|
4647 |
|
|
public Rank getRank() {
|
4648 |
|
|
return rank;
|
4649 |
|
|
}
|
4650 |
|
|
/**
|
4651 |
|
|
* @param rank the rank to set
|
4652 |
|
|
*/
|
4653 |
|
|
public void setRank(Rank rank) {
|
4654 |
|
|
this.rank = rank;
|
4655 |
|
|
}
|
4656 |
|
|
/**
|
4657 |
|
|
* @return the idenfitiger
|
4658 |
|
|
*/
|
4659 |
|
|
public String getIdentifier() {
|
4660 |
|
|
return identifier;
|
4661 |
|
|
}
|
4662 |
|
|
/**
|
4663 |
|
|
* @param idenfitiger the idenfitiger to set
|
4664 |
|
|
*/
|
4665 |
|
|
public void setIdentifier(String identifier) {
|
4666 |
|
|
this.identifier = identifier;
|
4667 |
|
|
}
|
4668 |
|
|
/**
|
4669 |
|
|
* @return the status
|
4670 |
|
|
*/
|
4671 |
|
|
public String getStatus() {
|
4672 |
|
|
if (status == null) {
|
4673 |
|
|
return "";
|
4674 |
|
|
}
|
4675 |
|
|
return status;
|
4676 |
|
|
}
|
4677 |
|
|
/**
|
4678 |
|
|
* @param status the status to set
|
4679 |
|
|
*/
|
4680 |
|
|
public void setStatus(String status) {
|
4681 |
|
|
this.status = status;
|
4682 |
|
|
}
|
4683 |
|
|
/**
|
4684 |
|
|
* @return the family
|
4685 |
|
|
*/
|
4686 |
|
|
public Taxon getFamily() {
|
4687 |
|
|
return family;
|
4688 |
|
|
}
|
4689 |
|
|
/**
|
4690 |
|
|
* @param family the family to set
|
4691 |
|
|
*/
|
4692 |
|
|
@SuppressWarnings("rawtypes")
|
4693 |
|
|
public void setFamily(Taxon family) {
|
4694 |
|
|
this.family = family;
|
4695 |
|
|
TaxonNameBase taxonNameBase = CdmBase.deproxy(family.getName(), TaxonNameBase.class);
|
4696 |
|
|
familyName = castTaxonNameBase(taxonNameBase,familyName);
|
4697 |
|
|
}
|
4698 |
|
|
/**
|
4699 |
|
|
* @return the subfamily
|
4700 |
|
|
*/
|
4701 |
|
|
public Taxon getSubfamily() {
|
4702 |
|
|
return subfamily;
|
4703 |
|
|
}
|
4704 |
|
|
/**
|
4705 |
|
|
* @param subfamily the subfamily to set
|
4706 |
|
|
*/
|
4707 |
|
|
@SuppressWarnings("rawtypes")
|
4708 |
|
|
public void setSubfamily(Taxon subfamily) {
|
4709 |
|
|
this.subfamily = subfamily;
|
4710 |
|
|
TaxonNameBase taxonNameBase = CdmBase.deproxy(subfamily.getName(), TaxonNameBase.class);
|
4711 |
|
|
subfamilyName = castTaxonNameBase(taxonNameBase,subfamilyName);
|
4712 |
|
|
}
|
4713 |
|
|
/**
|
4714 |
|
|
* @return the tribe
|
4715 |
|
|
*/
|
4716 |
|
|
public Taxon getTribe() {
|
4717 |
|
|
return tribe;
|
4718 |
|
|
}
|
4719 |
|
|
/**
|
4720 |
|
|
* @param tribe the tribe to set
|
4721 |
|
|
*/
|
4722 |
|
|
@SuppressWarnings("rawtypes")
|
4723 |
|
|
public void setTribe(Taxon tribe) {
|
4724 |
|
|
this.tribe = tribe;
|
4725 |
|
|
TaxonNameBase taxonNameBase = CdmBase.deproxy(tribe.getName(), TaxonNameBase.class);
|
4726 |
|
|
tribeName = castTaxonNameBase(taxonNameBase,tribeName);
|
4727 |
|
|
}
|
4728 |
|
|
/**
|
4729 |
|
|
* @return the subtribe
|
4730 |
|
|
*/
|
4731 |
|
|
public Taxon getSubtribe() {
|
4732 |
|
|
return subtribe;
|
4733 |
|
|
}
|
4734 |
|
|
/**
|
4735 |
|
|
* @param subtribe the subtribe to set
|
4736 |
|
|
*/
|
4737 |
|
|
@SuppressWarnings("rawtypes")
|
4738 |
|
|
public void setSubtribe(Taxon subtribe) {
|
4739 |
|
|
this.subtribe = subtribe;
|
4740 |
|
|
TaxonNameBase taxonNameBase = CdmBase.deproxy(subtribe.getName(), TaxonNameBase.class);
|
4741 |
|
|
subtribeName =castTaxonNameBase(taxonNameBase,subtribeName);
|
4742 |
|
|
}
|
4743 |
|
|
/**
|
4744 |
|
|
* @return the genus
|
4745 |
|
|
*/
|
4746 |
|
|
public Taxon getGenus() {
|
4747 |
|
|
return genus;
|
4748 |
|
|
}
|
4749 |
|
|
/**
|
4750 |
|
|
* @param genus the genus to set
|
4751 |
|
|
*/
|
4752 |
|
|
@SuppressWarnings("rawtypes")
|
4753 |
|
|
public void setGenus(Taxon genus) {
|
4754 |
|
|
this.genus = genus;
|
4755 |
|
|
TaxonNameBase taxonNameBase = CdmBase.deproxy(genus.getName(), TaxonNameBase.class);
|
4756 |
|
|
genusName = castTaxonNameBase(taxonNameBase,genusName);
|
4757 |
6c41cae2
|
p.kelbert
|
//System.out.println("GENUSNAME: "+genusName.toString());
|
4758 |
2b5a7979
|
p.kelbert
|
}
|
4759 |
|
|
/**
|
4760 |
|
|
* @return the subgenus
|
4761 |
|
|
*/
|
4762 |
|
|
public Taxon getSubgenus() {
|
4763 |
|
|
return subgenus;
|
4764 |
|
|
}
|
4765 |
|
|
/**
|
4766 |
|
|
* @param subgenus the subgenus to set
|
4767 |
|
|
*/
|
4768 |
|
|
@SuppressWarnings("rawtypes")
|
4769 |
|
|
public void setSubgenus(Taxon subgenus) {
|
4770 |
|
|
this.subgenus = subgenus;
|
4771 |
|
|
TaxonNameBase taxonNameBase = CdmBase.deproxy(subgenus.getName(), TaxonNameBase.class);
|
4772 |
|
|
subgenusName = castTaxonNameBase(taxonNameBase,subgenusName);
|
4773 |
|
|
}
|
4774 |
|
|
/**
|
4775 |
|
|
* @return the species
|
4776 |
|
|
*/
|
4777 |
|
|
public Taxon getSpecies() {
|
4778 |
|
|
return species;
|
4779 |
|
|
}
|
4780 |
|
|
/**
|
4781 |
|
|
* @param species the species to set
|
4782 |
|
|
*/
|
4783 |
|
|
public void setSpecies(Taxon species) {
|
4784 |
|
|
this.species = species;
|
4785 |
|
|
@SuppressWarnings("rawtypes")
|
4786 |
|
|
TaxonNameBase taxonNameBase = CdmBase.deproxy(species.getName(), TaxonNameBase.class);
|
4787 |
|
|
speciesName = castTaxonNameBase(taxonNameBase,speciesName);
|
4788 |
|
|
|
4789 |
|
|
}
|
4790 |
|
|
/**
|
4791 |
|
|
* @return the subspecies
|
4792 |
|
|
*/
|
4793 |
|
|
public Taxon getSubspecies() {
|
4794 |
|
|
return subspecies;
|
4795 |
|
|
}
|
4796 |
|
|
/**
|
4797 |
|
|
* @param subspecies the subspecies to set
|
4798 |
|
|
*/
|
4799 |
|
|
@SuppressWarnings("rawtypes")
|
4800 |
|
|
public void setSubspecies(Taxon subspecies) {
|
4801 |
|
|
this.subspecies = subspecies;
|
4802 |
|
|
TaxonNameBase taxonNameBase = CdmBase.deproxy(subspecies.getName(), TaxonNameBase.class);
|
4803 |
|
|
subspeciesName = castTaxonNameBase(taxonNameBase,subspeciesName);
|
4804 |
|
|
|
4805 |
|
|
}
|
4806 |
|
|
|
4807 |
|
|
|
4808 |
|
|
|
4809 |
|
|
}
|
4810 |
|
|
|
4811 |
|
|
|
4812 |
|
|
/**
|
4813 |
|
|
* @param status
|
4814 |
|
|
*/
|
4815 |
|
|
private void addProblematicStatusToFile(String status) {
|
4816 |
|
|
try{
|
4817 |
|
|
FileWriter fstream = new FileWriter("/home/pkelbert/Bureau/StatusUnknown_"+classification.getTitleCache()+".txt",true);
|
4818 |
|
|
BufferedWriter out = new BufferedWriter(fstream);
|
4819 |
|
|
out.write(status+"\n");
|
4820 |
|
|
//Close the output stream
|
4821 |
|
|
out.close();
|
4822 |
|
|
}catch (Exception e){//Catch exception if any
|
4823 |
|
|
System.err.println("Error: " + e.getMessage());
|
4824 |
|
|
}
|
4825 |
|
|
|
4826 |
|
|
}
|
4827 |
|
|
|
4828 |
|
|
|
4829 |
|
|
|
4830 |
|
|
/**
|
4831 |
|
|
* @param tnb
|
4832 |
|
|
* @return
|
4833 |
|
|
*/
|
4834 |
|
|
private Taxon findMatchingTaxon(NonViralName<?> tnb, Reference refMods) {
|
4835 |
6c41cae2
|
p.kelbert
|
logger.info("findMatchingTaxon");
|
4836 |
2b5a7979
|
p.kelbert
|
Taxon tmp=null;
|
4837 |
|
|
|
4838 |
6b2c2013
|
p.kelbert
|
refMods=CdmBase.deproxy(refMods, Reference.class);
|
4839 |
2b5a7979
|
p.kelbert
|
boolean insertAsExisting =false;
|
4840 |
6c41cae2
|
p.kelbert
|
List<Taxon> existingTaxons = new ArrayList<Taxon>();
|
4841 |
|
|
try {
|
4842 |
|
|
existingTaxons = getMatchingTaxon(tnb);
|
4843 |
|
|
} catch (Exception e1) {
|
4844 |
|
|
// TODO Auto-generated catch block
|
4845 |
|
|
e1.printStackTrace();
|
4846 |
|
|
}
|
4847 |
2b5a7979
|
p.kelbert
|
double similarityScore=0.0;
|
4848 |
6c41cae2
|
p.kelbert
|
double similarityAuthor=-1;
|
4849 |
|
|
String author1="";
|
4850 |
|
|
String author2="";
|
4851 |
|
|
String t1="";
|
4852 |
|
|
String t2="";
|
4853 |
2b5a7979
|
p.kelbert
|
for (Taxon bestMatchingTaxon:existingTaxons){
|
4854 |
|
|
if (!existingTaxons.isEmpty() && configState.getConfig().isInteractWithUser() && !insertAsExisting) {
|
4855 |
6c41cae2
|
p.kelbert
|
// System.out.println("tnb "+tnb.getTitleCache());
|
4856 |
|
|
// System.out.println("ext "+bestMatchingTaxon.getTitleCache());
|
4857 |
|
|
try {
|
4858 |
|
|
if(tnb.getAuthorshipCache()!=null) {
|
4859 |
|
|
author1=tnb.getAuthorshipCache();
|
4860 |
|
|
}
|
4861 |
|
|
} catch (Exception e) {
|
4862 |
|
|
// TODO Auto-generated catch block
|
4863 |
|
|
e.printStackTrace();
|
4864 |
|
|
}
|
4865 |
|
|
try {
|
4866 |
|
|
if(castTaxonNameBase(bestMatchingTaxon.getName()).getAuthorshipCache()!=null) {
|
4867 |
|
|
author2=castTaxonNameBase(bestMatchingTaxon.getName()).getAuthorshipCache();
|
4868 |
|
|
}
|
4869 |
|
|
} catch (Exception e) {
|
4870 |
|
|
// TODO Auto-generated catch block
|
4871 |
|
|
e.printStackTrace();
|
4872 |
|
|
}
|
4873 |
|
|
try {
|
4874 |
|
|
t1=tnb.getTitleCache().split("sec.")[0].trim();
|
4875 |
|
|
if (author1!=null && !StringUtils.isEmpty(author1)) {
|
4876 |
|
|
t1=t1.split(Pattern.quote(author1))[0];
|
4877 |
|
|
}
|
4878 |
|
|
} catch (Exception e) {
|
4879 |
|
|
// TODO Auto-generated catch block
|
4880 |
|
|
e.printStackTrace();
|
4881 |
|
|
}
|
4882 |
|
|
try {
|
4883 |
|
|
t2=bestMatchingTaxon.getTitleCache().split("sec.")[0].trim();
|
4884 |
|
|
if (author2!=null && !StringUtils.isEmpty(author2)) {
|
4885 |
|
|
t2=t2.split(Pattern.quote(author2))[0];
|
4886 |
|
|
}
|
4887 |
|
|
} catch (Exception e) {
|
4888 |
|
|
// TODO Auto-generated catch block
|
4889 |
|
|
e.printStackTrace();
|
4890 |
|
|
}
|
4891 |
|
|
similarityScore=similarity(t1.trim(), t2.trim());
|
4892 |
|
|
// System.out.println("taxascore: "+similarityScore);
|
4893 |
|
|
similarityAuthor=similarity(author1.trim(), author2.trim());
|
4894 |
|
|
// System.out.println("authorscore: "+similarityAuthor);
|
4895 |
|
|
insertAsExisting = compareAndCheckTaxon(tnb, refMods, similarityScore, bestMatchingTaxon,similarityAuthor);
|
4896 |
2b5a7979
|
p.kelbert
|
}
|
4897 |
|
|
if(insertAsExisting) {
|
4898 |
6c41cae2
|
p.kelbert
|
//System.out.println("KEEP "+bestMatchingTaxon.toString());
|
4899 |
2b5a7979
|
p.kelbert
|
tmp=bestMatchingTaxon;
|
4900 |
|
|
sourceHandler.addSource(refMods, tmp);
|
4901 |
|
|
return tmp;
|
4902 |
|
|
}
|
4903 |
|
|
}
|
4904 |
|
|
return tmp;
|
4905 |
|
|
}
|
4906 |
|
|
|
4907 |
6c41cae2
|
p.kelbert
|
|
4908 |
2b5a7979
|
p.kelbert
|
/**
|
4909 |
|
|
* @param tnb
|
4910 |
|
|
* @param refMods
|
4911 |
|
|
* @param similarityScore
|
4912 |
|
|
* @param bestMatchingTaxon
|
4913 |
6c41cae2
|
p.kelbert
|
* @param similarityAuthor
|
4914 |
2b5a7979
|
p.kelbert
|
* @return
|
4915 |
|
|
*/
|
4916 |
6e861bce
|
Andreas Müller
|
private boolean compareAndCheckTaxon(NonViralName<?> tnb, Reference<?> refMods, double similarityScore,
|
4917 |
6c41cae2
|
p.kelbert
|
Taxon bestMatchingTaxon, double similarityAuthor) {
|
4918 |
|
|
//logger.info("compareAndCheckTaxon");
|
4919 |
2b5a7979
|
p.kelbert
|
boolean insertAsExisting;
|
4920 |
6c41cae2
|
p.kelbert
|
// if (tnb.getTitleCache().split("sec.")[0].equalsIgnoreCase("Chenopodium") && bestMatchingTaxon.getTitleCache().split("sec.")[0].indexOf("Chenopodium album")>-1) {
|
4921 |
|
|
// insertAsExisting=false;
|
4922 |
|
|
// } else{
|
4923 |
|
|
//a small hack/automatisation for Chenopodium only
|
4924 |
|
|
if (tnb.getTitleCache().split("sec.")[0].trim().equalsIgnoreCase("Chenopodium") &&
|
4925 |
|
|
bestMatchingTaxon.getTitleCache().split("sec.")[0].indexOf("Chenopodium L.")>-1) {
|
4926 |
|
|
insertAsExisting=true;
|
4927 |
|
|
} else {
|
4928 |
|
|
insertAsExisting=askIfReuseBestMatchingTaxon(tnb, bestMatchingTaxon, refMods, similarityScore,similarityAuthor);
|
4929 |
2b5a7979
|
p.kelbert
|
}
|
4930 |
6c41cae2
|
p.kelbert
|
// }
|
4931 |
|
|
|
4932 |
|
|
logDecision(tnb,bestMatchingTaxon,insertAsExisting, refMods);
|
4933 |
2b5a7979
|
p.kelbert
|
return insertAsExisting;
|
4934 |
|
|
}
|
4935 |
|
|
|
4936 |
|
|
/**
|
4937 |
|
|
* @return
|
4938 |
|
|
*/
|
4939 |
|
|
@SuppressWarnings("rawtypes")
|
4940 |
|
|
private List<Taxon> getMatchingTaxon(TaxonNameBase tnb) {
|
4941 |
6c41cae2
|
p.kelbert
|
//logger.info("getMatchingTaxon");
|
4942 |
|
|
Pager<TaxonBase> pager=importer.getTaxonService().findByTitle(TaxonBase.class, tnb.getTitleCache().split("sec.")[0].trim(), MatchMode.BEGINNING, null, null, null, null, null);
|
4943 |
2b5a7979
|
p.kelbert
|
List<TaxonBase>records = pager.getRecords();
|
4944 |
|
|
|
4945 |
|
|
List<Taxon> existingTaxons = new ArrayList<Taxon>();
|
4946 |
|
|
for (TaxonBase r:records){
|
4947 |
|
|
try{
|
4948 |
|
|
Taxon bestMatchingTaxon = (Taxon)r;
|
4949 |
|
|
// System.out.println("best: "+bestMatchingTaxon.getTitleCache());
|
4950 |
|
|
if(compareTaxonNameLength(bestMatchingTaxon.getTitleCache().split(".sec")[0],tnb.getTitleCache().split(".sec")[0])) {
|
4951 |
|
|
existingTaxons.add(bestMatchingTaxon);
|
4952 |
|
|
}
|
4953 |
|
|
}catch(ClassCastException e){logger.warn("classcast exception, might be a synonym, ignore it");}
|
4954 |
|
|
}
|
4955 |
|
|
Taxon bmt = importer.getTaxonService().findBestMatchingTaxon(tnb.getTitleCache());
|
4956 |
|
|
if (!existingTaxons.contains(bmt) && bmt!=null) {
|
4957 |
|
|
if(compareTaxonNameLength(bmt.getTitleCache().split(".sec")[0],tnb.getTitleCache().split(".sec")[0])) {
|
4958 |
|
|
existingTaxons.add(bmt);
|
4959 |
|
|
}
|
4960 |
|
|
}
|
4961 |
|
|
return existingTaxons;
|
4962 |
|
|
}
|
4963 |
|
|
|
4964 |
|
|
/**
|
4965 |
|
|
* Check if the found Taxon can reasonnably be the same
|
4966 |
|
|
* example: with and without author should match, but the subspecies should not be suggested for a genus
|
4967 |
|
|
* */
|
4968 |
|
|
private boolean compareTaxonNameLength(String f, String o){
|
4969 |
6c41cae2
|
p.kelbert
|
//logger.info("compareTaxonNameLength");
|
4970 |
2b5a7979
|
p.kelbert
|
boolean lengthOk=false;
|
4971 |
|
|
int sizeF = f.length();
|
4972 |
|
|
int sizeO = o.length();
|
4973 |
|
|
if (sizeO>=sizeF) {
|
4974 |
|
|
lengthOk=true;
|
4975 |
|
|
}
|
4976 |
|
|
if(sizeF>sizeO) {
|
4977 |
|
|
if (sizeF-sizeO>10) {
|
4978 |
|
|
lengthOk=false;
|
4979 |
|
|
} else {
|
4980 |
|
|
lengthOk=true;
|
4981 |
|
|
}
|
4982 |
|
|
}
|
4983 |
|
|
|
4984 |
|
|
// System.out.println(lengthOk+": compare "+f+" ("+f.length()+") and "+o+" ("+o.length()+")");
|
4985 |
|
|
return lengthOk;
|
4986 |
|
|
}
|
4987 |
|
|
|
4988 |
|
|
private double similarity(String s1, String s2) {
|
4989 |
6c41cae2
|
p.kelbert
|
//logger.info("similarity");
|
4990 |
|
|
//System.out.println("similarity *"+s1+"* vs. *"+s2+"*");
|
4991 |
|
|
if(!StringUtils.isEmpty(s1) && !StringUtils.isEmpty(s2)){
|
4992 |
|
|
String l1=s1.toLowerCase().trim();
|
4993 |
|
|
String l2=s2.toLowerCase().trim();
|
4994 |
|
|
if (l1.length() < l2.length()) { // s1 should always be bigger
|
4995 |
|
|
String swap = l1; l1 = l2; l2 = swap;
|
4996 |
|
|
}
|
4997 |
|
|
int bigLen = l1.length();
|
4998 |
|
|
if (bigLen == 0) { return 1.0; /* both strings are zero length */ }
|
4999 |
|
|
return (bigLen - computeEditDistance(l1, l2)) / (double) bigLen;
|
5000 |
|
|
}
|
5001 |
|
|
else{
|
5002 |
|
|
if(s1!=null && s2!=null){
|
5003 |
|
|
if (s1.equalsIgnoreCase(s2)) {
|
5004 |
|
|
return 1;
|
5005 |
|
|
}
|
5006 |
|
|
}
|
5007 |
|
|
return -1;
|
5008 |
2b5a7979
|
p.kelbert
|
}
|
5009 |
|
|
}
|
5010 |
|
|
|
5011 |
|
|
private int computeEditDistance(String s1, String s2) {
|
5012 |
6c41cae2
|
p.kelbert
|
//logger.info("computeEditDistance");
|
5013 |
2b5a7979
|
p.kelbert
|
int[] costs = new int[s2.length() + 1];
|
5014 |
|
|
for (int i = 0; i <= s1.length(); i++) {
|
5015 |
|
|
int lastValue = i;
|
5016 |
|
|
for (int j = 0; j <= s2.length(); j++) {
|
5017 |
|
|
if (i == 0) {
|
5018 |
|
|
costs[j] = j;
|
5019 |
|
|
} else {
|
5020 |
|
|
if (j > 0) {
|
5021 |
|
|
int newValue = costs[j - 1];
|
5022 |
|
|
if (s1.charAt(i - 1) != s2.charAt(j - 1)) {
|
5023 |
|
|
newValue = Math.min(Math.min(newValue, lastValue),
|
5024 |
|
|
costs[j]) + 1;
|
5025 |
|
|
}
|
5026 |
|
|
costs[j - 1] = lastValue;
|
5027 |
|
|
lastValue = newValue;
|
5028 |
|
|
}
|
5029 |
|
|
}
|
5030 |
|
|
}
|
5031 |
|
|
if (i > 0) {
|
5032 |
|
|
costs[s2.length()] = lastValue;
|
5033 |
|
|
}
|
5034 |
|
|
}
|
5035 |
|
|
return costs[s2.length()];
|
5036 |
|
|
}
|
5037 |
|
|
|
5038 |
|
|
Map<Rank, Taxon> hierarchy = new HashMap<Rank, Taxon>();
|
5039 |
|
|
/**
|
5040 |
|
|
* @param taxonnamebase
|
5041 |
|
|
*/
|
5042 |
|
|
@SuppressWarnings("rawtypes")
|
5043 |
|
|
public void lookForParentNode(NonViralName<?> taxonnamebase, Taxon tax, Reference<?> ref, MyName myName) {
|
5044 |
6c41cae2
|
p.kelbert
|
logger.info("lookForParentNode "+taxonnamebase.getTitleCache()+" for "+myName.toString());
|
5045 |
|
|
//System.out.println("LOOK FOR PARENT NODE "+taxonnamebase.toString()+"; "+tax.toString()+"; "+taxonnamebase.getRank());
|
5046 |
2b5a7979
|
p.kelbert
|
INonViralNameParser parser = NonViralNameParserImpl.NewInstance();
|
5047 |
|
|
if (taxonnamebase.getRank().equals(Rank.FORM())){
|
5048 |
|
|
handleFormHierarchy(ref, myName, parser);
|
5049 |
|
|
}
|
5050 |
|
|
if (taxonnamebase.getRank().equals(Rank.VARIETY())){
|
5051 |
|
|
handleVarietyHierarchy(ref, myName, parser);
|
5052 |
|
|
}
|
5053 |
|
|
if (taxonnamebase.getRank().equals(Rank.SUBSPECIES())){
|
5054 |
|
|
handleSubSpeciesHierarchy(ref, myName, parser);
|
5055 |
|
|
}
|
5056 |
|
|
if (taxonnamebase.getRank().equals(Rank.SPECIES())){
|
5057 |
|
|
handleSpeciesHierarchy(ref, myName, parser);
|
5058 |
|
|
}
|
5059 |
|
|
if (taxonnamebase.getRank().equals(Rank.SUBGENUS())){
|
5060 |
|
|
handleSubgenusHierarchy(ref, myName, parser);
|
5061 |
|
|
}
|
5062 |
|
|
|
5063 |
|
|
if (taxonnamebase.getRank().equals(Rank.GENUS())){
|
5064 |
|
|
handleGenusHierarchy(ref, myName, parser);
|
5065 |
|
|
}
|
5066 |
|
|
if (taxonnamebase.getRank().equals(Rank.SUBTRIBE())){
|
5067 |
|
|
handleSubtribeHierarchy(ref, myName, parser);
|
5068 |
|
|
}
|
5069 |
|
|
if (taxonnamebase.getRank().equals(Rank.TRIBE())){
|
5070 |
|
|
handleTribeHierarchy(ref, myName, parser);
|
5071 |
|
|
}
|
5072 |
|
|
|
5073 |
|
|
if (taxonnamebase.getRank().equals(Rank.SUBFAMILY())){
|
5074 |
|
|
handleSubfamilyHierarchy(ref, myName, parser);
|
5075 |
|
|
}
|
5076 |
|
|
}
|
5077 |
|
|
|
5078 |
|
|
/**
|
5079 |
|
|
* @param ref
|
5080 |
|
|
* @param myName
|
5081 |
|
|
* @param parser
|
5082 |
|
|
*/
|
5083 |
6e861bce
|
Andreas Müller
|
private void handleSubfamilyHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5084 |
6c41cae2
|
p.kelbert
|
System.out.println("handleSubfamilyHierarchy");
|
5085 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getFamilyStr();
|
5086 |
|
|
Rank r = Rank.FAMILY();
|
5087 |
|
|
if(parentStr!=null){
|
5088 |
|
|
|
5089 |
6c41cae2
|
p.kelbert
|
Taxon parent = null;
|
5090 |
|
|
Pager<TaxonBase> taxontest = importer.getTaxonService().findByTitle(TaxonBase.class, parentStr, MatchMode.BEGINNING, null, null, null, null, null);
|
5091 |
|
|
for(TaxonBase tb:taxontest.getRecords()){
|
5092 |
|
|
try {
|
5093 |
|
|
if (tb.getName().getRank().equals(r)) {
|
5094 |
|
|
parent=CdmBase.deproxy(tb, Taxon.class);
|
5095 |
|
|
}
|
5096 |
2b5a7979
|
p.kelbert
|
break;
|
5097 |
6c41cae2
|
p.kelbert
|
} catch (Exception e) {
|
5098 |
|
|
// TODO Auto-generated catch block
|
5099 |
|
|
e.printStackTrace();
|
5100 |
2b5a7979
|
p.kelbert
|
}
|
5101 |
|
|
}
|
5102 |
6c41cae2
|
p.kelbert
|
if(parent == null) {
|
5103 |
|
|
NonViralName<?> parentNameName = (NonViralName<?>) parser.parseFullName(parentStr, nomenclaturalCode, r);
|
5104 |
2b5a7979
|
p.kelbert
|
Taxon tmp = findMatchingTaxon(parentNameName,ref);
|
5105 |
|
|
if(tmp ==null)
|
5106 |
|
|
{
|
5107 |
|
|
parent=Taxon.NewInstance(parentNameName, ref);
|
5108 |
|
|
importer.getTaxonService().save(parent);
|
5109 |
|
|
parent = CdmBase.deproxy(parent, Taxon.class);
|
5110 |
|
|
} else {
|
5111 |
|
|
parent=tmp;
|
5112 |
|
|
}
|
5113 |
|
|
lookForParentNode(parentNameName, parent, ref,myName);
|
5114 |
|
|
|
5115 |
|
|
}
|
5116 |
|
|
hierarchy.put(r,parent);
|
5117 |
|
|
}
|
5118 |
|
|
}
|
5119 |
|
|
|
5120 |
|
|
/**
|
5121 |
|
|
* @param ref
|
5122 |
|
|
* @param myName
|
5123 |
|
|
* @param parser
|
5124 |
|
|
*/
|
5125 |
6e861bce
|
Andreas Müller
|
private void handleTribeHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5126 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getSubfamilyStr();
|
5127 |
|
|
Rank r = Rank.SUBFAMILY();
|
5128 |
|
|
if (parentStr == null){
|
5129 |
|
|
parentStr = myName.getFamilyStr();
|
5130 |
|
|
r = Rank.FAMILY();
|
5131 |
|
|
}
|
5132 |
|
|
if(parentStr!=null){
|
5133 |
|
|
NonViralName<?> parentNameName = (NonViralName<?>) parser.parseFullName(parentStr, nomenclaturalCode, r);
|
5134 |
|
|
Taxon parent = Taxon.NewInstance(parentNameName, ref); //sec set null
|
5135 |
|
|
// importer.getTaxonService().save(parent);
|
5136 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5137 |
|
|
|
5138 |
|
|
boolean parentDoesNotExists = true;
|
5139 |
|
|
for (TaxonNode p : classification.getAllNodes()){
|
5140 |
|
|
if(p.getTaxon().getTitleCache().equalsIgnoreCase(parent.getTitleCache())) {
|
5141 |
|
|
parentDoesNotExists = false;
|
5142 |
|
|
parent=CdmBase.deproxy(p.getTaxon(), Taxon.class);
|
5143 |
|
|
break;
|
5144 |
|
|
}
|
5145 |
|
|
}
|
5146 |
|
|
// if(parentDoesNotExists) {
|
5147 |
|
|
// importer.getTaxonService().save(parent);
|
5148 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5149 |
|
|
// lookForParentNode(parentNameName, parent, ref,myName);
|
5150 |
|
|
// }
|
5151 |
|
|
if(parentDoesNotExists) {
|
5152 |
|
|
Taxon tmp = findMatchingTaxon(parentNameName,ref);
|
5153 |
|
|
if(tmp ==null)
|
5154 |
|
|
{
|
5155 |
|
|
parent=Taxon.NewInstance(parentNameName, ref);
|
5156 |
|
|
importer.getTaxonService().save(parent);
|
5157 |
|
|
parent = CdmBase.deproxy(parent, Taxon.class);
|
5158 |
|
|
} else {
|
5159 |
|
|
parent=tmp;
|
5160 |
|
|
}
|
5161 |
|
|
lookForParentNode(parentNameName, parent, ref,myName);
|
5162 |
|
|
|
5163 |
|
|
}
|
5164 |
|
|
hierarchy.put(r,parent);
|
5165 |
|
|
}
|
5166 |
|
|
}
|
5167 |
|
|
|
5168 |
|
|
/**
|
5169 |
|
|
* @param ref
|
5170 |
|
|
* @param myName
|
5171 |
|
|
* @param parser
|
5172 |
|
|
*/
|
5173 |
6e861bce
|
Andreas Müller
|
private void handleSubtribeHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5174 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getTribeStr();
|
5175 |
|
|
Rank r = Rank.TRIBE();
|
5176 |
|
|
if (parentStr == null){
|
5177 |
|
|
parentStr = myName.getSubfamilyStr();
|
5178 |
|
|
r = Rank.SUBFAMILY();
|
5179 |
|
|
}
|
5180 |
|
|
if (parentStr == null){
|
5181 |
|
|
parentStr = myName.getFamilyStr();
|
5182 |
|
|
r = Rank.FAMILY();
|
5183 |
|
|
}
|
5184 |
|
|
if(parentStr!=null){
|
5185 |
|
|
NonViralName<?> parentNameName = (NonViralName<?>) parser.parseFullName(parentStr, nomenclaturalCode, r);
|
5186 |
|
|
Taxon parent = Taxon.NewInstance(parentNameName, ref); //sec set null
|
5187 |
|
|
// importer.getTaxonService().save(parent);
|
5188 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5189 |
|
|
|
5190 |
|
|
boolean parentDoesNotExists = true;
|
5191 |
|
|
for (TaxonNode p : classification.getAllNodes()){
|
5192 |
|
|
if(p.getTaxon().getTitleCache().equalsIgnoreCase(parent.getTitleCache())) {
|
5193 |
|
|
parentDoesNotExists = false;
|
5194 |
|
|
parent=CdmBase.deproxy(p.getTaxon(), Taxon.class);
|
5195 |
|
|
|
5196 |
|
|
break;
|
5197 |
|
|
}
|
5198 |
|
|
}
|
5199 |
|
|
// if(parentDoesNotExists) {
|
5200 |
|
|
// importer.getTaxonService().save(parent);
|
5201 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5202 |
|
|
// lookForParentNode(parentNameName, parent, ref,myName);
|
5203 |
|
|
// }
|
5204 |
|
|
if(parentDoesNotExists) {
|
5205 |
|
|
Taxon tmp = findMatchingTaxon(parentNameName,ref);
|
5206 |
|
|
if(tmp ==null)
|
5207 |
|
|
{
|
5208 |
|
|
parent=Taxon.NewInstance(parentNameName, ref);
|
5209 |
|
|
importer.getTaxonService().save(parent);
|
5210 |
|
|
parent = CdmBase.deproxy(parent, Taxon.class);
|
5211 |
|
|
} else {
|
5212 |
|
|
parent=tmp;
|
5213 |
|
|
}
|
5214 |
|
|
lookForParentNode(parentNameName, parent, ref,myName);
|
5215 |
|
|
|
5216 |
|
|
}
|
5217 |
|
|
hierarchy.put(r,parent);
|
5218 |
|
|
}
|
5219 |
|
|
}
|
5220 |
|
|
|
5221 |
|
|
/**
|
5222 |
|
|
* @param ref
|
5223 |
|
|
* @param myName
|
5224 |
|
|
* @param parser
|
5225 |
|
|
*/
|
5226 |
6e861bce
|
Andreas Müller
|
private void handleGenusHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5227 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getSubtribeStr();
|
5228 |
|
|
Rank r = Rank.SUBTRIBE();
|
5229 |
|
|
if (parentStr == null){
|
5230 |
|
|
parentStr = myName.getTribeStr();
|
5231 |
|
|
r = Rank.TRIBE();
|
5232 |
|
|
}
|
5233 |
|
|
if (parentStr == null){
|
5234 |
|
|
parentStr = myName.getSubfamilyStr();
|
5235 |
|
|
r = Rank.SUBFAMILY();
|
5236 |
|
|
}
|
5237 |
|
|
if (parentStr == null){
|
5238 |
|
|
parentStr = myName.getFamilyStr();
|
5239 |
|
|
r = Rank.FAMILY();
|
5240 |
|
|
}
|
5241 |
|
|
if(parentStr!=null){
|
5242 |
|
|
NonViralName<?> parentNameName = (NonViralName<?>) parser.parseFullName(parentStr, nomenclaturalCode, r);
|
5243 |
|
|
Taxon parent = Taxon.NewInstance(parentNameName, ref); //sec set null
|
5244 |
|
|
// importer.getTaxonService().save(parent);
|
5245 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5246 |
|
|
|
5247 |
|
|
boolean parentDoesNotExists = true;
|
5248 |
|
|
for (TaxonNode p : classification.getAllNodes()){
|
5249 |
|
|
if(p.getTaxon().getTitleCache().equalsIgnoreCase(parent.getTitleCache())) {
|
5250 |
|
|
// System.out.println(p.getTaxon().getUuid());
|
5251 |
|
|
// System.out.println(parent.getUuid());
|
5252 |
|
|
parentDoesNotExists = false;
|
5253 |
|
|
parent=CdmBase.deproxy(p.getTaxon(), Taxon.class);
|
5254 |
|
|
break;
|
5255 |
|
|
}
|
5256 |
|
|
}
|
5257 |
|
|
// if(parentDoesNotExists) {
|
5258 |
|
|
// importer.getTaxonService().save(parent);
|
5259 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5260 |
|
|
// lookForParentNode(parentNameName, parent, ref,myName);
|
5261 |
|
|
// }
|
5262 |
|
|
if(parentDoesNotExists) {
|
5263 |
|
|
Taxon tmp = findMatchingTaxon(parentNameName,ref);
|
5264 |
|
|
if(tmp ==null)
|
5265 |
|
|
{
|
5266 |
|
|
parent=Taxon.NewInstance(parentNameName, ref);
|
5267 |
|
|
importer.getTaxonService().save(parent);
|
5268 |
|
|
parent = CdmBase.deproxy(parent, Taxon.class);
|
5269 |
|
|
} else {
|
5270 |
|
|
parent=tmp;
|
5271 |
|
|
}
|
5272 |
|
|
lookForParentNode(parentNameName, parent, ref,myName);
|
5273 |
|
|
|
5274 |
|
|
}
|
5275 |
|
|
hierarchy.put(r,parent);
|
5276 |
|
|
}
|
5277 |
|
|
}
|
5278 |
|
|
|
5279 |
|
|
/**
|
5280 |
|
|
* @param ref
|
5281 |
|
|
* @param myName
|
5282 |
|
|
* @param parser
|
5283 |
|
|
*/
|
5284 |
6e861bce
|
Andreas Müller
|
private void handleSubgenusHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5285 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getGenusStr();
|
5286 |
|
|
Rank r = Rank.GENUS();
|
5287 |
|
|
|
5288 |
|
|
if(parentStr==null){
|
5289 |
|
|
parentStr = myName.getSubtribeStr();
|
5290 |
|
|
r = Rank.SUBTRIBE();
|
5291 |
|
|
}
|
5292 |
|
|
if (parentStr == null){
|
5293 |
|
|
parentStr = myName.getTribeStr();
|
5294 |
|
|
r = Rank.TRIBE();
|
5295 |
|
|
}
|
5296 |
|
|
if (parentStr == null){
|
5297 |
|
|
parentStr = myName.getSubfamilyStr();
|
5298 |
|
|
r = Rank.SUBFAMILY();
|
5299 |
|
|
}
|
5300 |
|
|
if (parentStr == null){
|
5301 |
|
|
parentStr = myName.getFamilyStr();
|
5302 |
|
|
r = Rank.FAMILY();
|
5303 |
|
|
}
|
5304 |
|
|
if(parentStr!=null){
|
5305 |
|
|
NonViralName<?> parentNameName = (NonViralName<?>) parser.parseFullName(parentStr, nomenclaturalCode, r);
|
5306 |
|
|
Taxon parent = Taxon.NewInstance(parentNameName, ref); //sec set null
|
5307 |
|
|
// importer.getTaxonService().save(parent);
|
5308 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5309 |
|
|
|
5310 |
|
|
boolean parentDoesNotExists = true;
|
5311 |
|
|
for (TaxonNode p : classification.getAllNodes()){
|
5312 |
|
|
if(p.getTaxon().getTitleCache().equalsIgnoreCase(parent.getTitleCache())) {
|
5313 |
|
|
// System.out.println(p.getTaxon().getUuid());
|
5314 |
|
|
// System.out.println(parent.getUuid());
|
5315 |
|
|
parentDoesNotExists = false;
|
5316 |
|
|
parent=CdmBase.deproxy(p.getTaxon(), Taxon.class);
|
5317 |
|
|
break;
|
5318 |
|
|
}
|
5319 |
|
|
}
|
5320 |
|
|
// if(parentDoesNotExists) {
|
5321 |
|
|
// importer.getTaxonService().save(parent);
|
5322 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5323 |
|
|
// lookForParentNode(parentNameName, parent, ref,myName);
|
5324 |
|
|
// }
|
5325 |
|
|
if(parentDoesNotExists) {
|
5326 |
|
|
Taxon tmp = findMatchingTaxon(parentNameName,ref);
|
5327 |
|
|
if(tmp ==null)
|
5328 |
|
|
{
|
5329 |
|
|
parent=Taxon.NewInstance(parentNameName, ref);
|
5330 |
|
|
importer.getTaxonService().save(parent);
|
5331 |
|
|
parent = CdmBase.deproxy(parent, Taxon.class);
|
5332 |
|
|
} else {
|
5333 |
|
|
parent=tmp;
|
5334 |
|
|
}
|
5335 |
|
|
lookForParentNode(parentNameName, parent, ref,myName);
|
5336 |
|
|
|
5337 |
|
|
}
|
5338 |
|
|
hierarchy.put(r,parent);
|
5339 |
|
|
}
|
5340 |
|
|
}
|
5341 |
|
|
|
5342 |
|
|
/**
|
5343 |
|
|
* @param ref
|
5344 |
|
|
* @param myName
|
5345 |
|
|
* @param parser
|
5346 |
|
|
*/
|
5347 |
6e861bce
|
Andreas Müller
|
private void handleSpeciesHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5348 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getSubgenusStr();
|
5349 |
|
|
Rank r = Rank.SUBGENUS();
|
5350 |
|
|
|
5351 |
|
|
if(parentStr==null){
|
5352 |
|
|
parentStr = myName.getGenusStr();
|
5353 |
|
|
r = Rank.GENUS();
|
5354 |
|
|
}
|
5355 |
|
|
|
5356 |
|
|
if(parentStr==null){
|
5357 |
|
|
parentStr = myName.getSubtribeStr();
|
5358 |
|
|
r = Rank.SUBTRIBE();
|
5359 |
|
|
}
|
5360 |
|
|
if (parentStr == null){
|
5361 |
|
|
parentStr = myName.getTribeStr();
|
5362 |
|
|
r = Rank.TRIBE();
|
5363 |
|
|
}
|
5364 |
|
|
if (parentStr == null){
|
5365 |
|
|
parentStr = myName.getSubfamilyStr();
|
5366 |
|
|
r = Rank.SUBFAMILY();
|
5367 |
|
|
}
|
5368 |
|
|
if (parentStr == null){
|
5369 |
|
|
parentStr = myName.getFamilyStr();
|
5370 |
|
|
r = Rank.FAMILY();
|
5371 |
|
|
}
|
5372 |
|
|
if(parentStr!=null){
|
5373 |
|
|
Taxon parent = handleParentName(ref, myName, parser, parentStr, r);
|
5374 |
6c41cae2
|
p.kelbert
|
//System.out.println("PUT IN HIERARCHY "+r+", "+parent);
|
5375 |
2b5a7979
|
p.kelbert
|
hierarchy.put(r,parent);
|
5376 |
|
|
}
|
5377 |
|
|
}
|
5378 |
|
|
|
5379 |
|
|
/**
|
5380 |
|
|
* @param ref
|
5381 |
|
|
* @param myName
|
5382 |
|
|
* @param parser
|
5383 |
|
|
*/
|
5384 |
6e861bce
|
Andreas Müller
|
private void handleSubSpeciesHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5385 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getSpeciesStr();
|
5386 |
|
|
Rank r = Rank.SPECIES();
|
5387 |
|
|
|
5388 |
|
|
|
5389 |
|
|
if(parentStr==null){
|
5390 |
|
|
parentStr = myName.getSubgenusStr();
|
5391 |
|
|
r = Rank.SUBGENUS();
|
5392 |
|
|
}
|
5393 |
|
|
|
5394 |
|
|
if(parentStr==null){
|
5395 |
|
|
parentStr = myName.getGenusStr();
|
5396 |
|
|
r = Rank.GENUS();
|
5397 |
|
|
}
|
5398 |
|
|
|
5399 |
|
|
if(parentStr==null){
|
5400 |
|
|
parentStr = myName.getSubtribeStr();
|
5401 |
|
|
r = Rank.SUBTRIBE();
|
5402 |
|
|
}
|
5403 |
|
|
if (parentStr == null){
|
5404 |
|
|
parentStr = myName.getTribeStr();
|
5405 |
|
|
r = Rank.TRIBE();
|
5406 |
|
|
}
|
5407 |
|
|
if (parentStr == null){
|
5408 |
|
|
parentStr = myName.getSubfamilyStr();
|
5409 |
|
|
r = Rank.SUBFAMILY();
|
5410 |
|
|
}
|
5411 |
|
|
if (parentStr == null){
|
5412 |
|
|
parentStr = myName.getFamilyStr();
|
5413 |
|
|
r = Rank.FAMILY();
|
5414 |
|
|
}
|
5415 |
|
|
if(parentStr!=null){
|
5416 |
|
|
Taxon parent = handleParentName(ref, myName, parser, parentStr, r);
|
5417 |
6c41cae2
|
p.kelbert
|
//System.out.println("PUT IN HIERARCHY "+r+", "+parent);
|
5418 |
2b5a7979
|
p.kelbert
|
hierarchy.put(r,parent);
|
5419 |
|
|
}
|
5420 |
|
|
}
|
5421 |
|
|
|
5422 |
|
|
|
5423 |
|
|
/**
|
5424 |
|
|
* @param ref
|
5425 |
|
|
* @param myName
|
5426 |
|
|
* @param parser
|
5427 |
|
|
*/
|
5428 |
6e861bce
|
Andreas Müller
|
private void handleFormHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5429 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getSubspeciesStr();
|
5430 |
|
|
Rank r = Rank.SUBSPECIES();
|
5431 |
|
|
|
5432 |
|
|
|
5433 |
|
|
if(parentStr==null){
|
5434 |
|
|
parentStr = myName.getSpeciesStr();
|
5435 |
|
|
r = Rank.SPECIES();
|
5436 |
|
|
}
|
5437 |
|
|
|
5438 |
|
|
if(parentStr==null){
|
5439 |
|
|
parentStr = myName.getSubgenusStr();
|
5440 |
|
|
r = Rank.SUBGENUS();
|
5441 |
|
|
}
|
5442 |
|
|
|
5443 |
|
|
if(parentStr==null){
|
5444 |
|
|
parentStr = myName.getGenusStr();
|
5445 |
|
|
r = Rank.GENUS();
|
5446 |
|
|
}
|
5447 |
|
|
|
5448 |
|
|
if(parentStr==null){
|
5449 |
|
|
parentStr = myName.getSubtribeStr();
|
5450 |
|
|
r = Rank.SUBTRIBE();
|
5451 |
|
|
}
|
5452 |
|
|
if (parentStr == null){
|
5453 |
|
|
parentStr = myName.getTribeStr();
|
5454 |
|
|
r = Rank.TRIBE();
|
5455 |
|
|
}
|
5456 |
|
|
if (parentStr == null){
|
5457 |
|
|
parentStr = myName.getSubfamilyStr();
|
5458 |
|
|
r = Rank.SUBFAMILY();
|
5459 |
|
|
}
|
5460 |
|
|
if (parentStr == null){
|
5461 |
|
|
parentStr = myName.getFamilyStr();
|
5462 |
|
|
r = Rank.FAMILY();
|
5463 |
|
|
}
|
5464 |
|
|
if(parentStr!=null){
|
5465 |
|
|
Taxon parent = handleParentName(ref, myName, parser, parentStr, r);
|
5466 |
6c41cae2
|
p.kelbert
|
//System.out.println("PUT IN HIERARCHY "+r+", "+parent);
|
5467 |
2b5a7979
|
p.kelbert
|
hierarchy.put(r,parent);
|
5468 |
|
|
}
|
5469 |
|
|
}
|
5470 |
|
|
|
5471 |
|
|
/**
|
5472 |
|
|
* @param ref
|
5473 |
|
|
* @param myName
|
5474 |
|
|
* @param parser
|
5475 |
|
|
*/
|
5476 |
6e861bce
|
Andreas Müller
|
private void handleVarietyHierarchy(Reference<?> ref, MyName myName, INonViralNameParser<?> parser) {
|
5477 |
2b5a7979
|
p.kelbert
|
String parentStr = myName.getSubspeciesStr();
|
5478 |
|
|
Rank r = Rank.SUBSPECIES();
|
5479 |
|
|
|
5480 |
|
|
if(parentStr==null){
|
5481 |
|
|
parentStr = myName.getSpeciesStr();
|
5482 |
|
|
r = Rank.SPECIES();
|
5483 |
|
|
}
|
5484 |
|
|
|
5485 |
|
|
if(parentStr==null){
|
5486 |
|
|
parentStr = myName.getSubgenusStr();
|
5487 |
|
|
r = Rank.SUBGENUS();
|
5488 |
|
|
}
|
5489 |
|
|
|
5490 |
|
|
if(parentStr==null){
|
5491 |
|
|
parentStr = myName.getGenusStr();
|
5492 |
|
|
r = Rank.GENUS();
|
5493 |
|
|
}
|
5494 |
|
|
|
5495 |
|
|
if(parentStr==null){
|
5496 |
|
|
parentStr = myName.getSubtribeStr();
|
5497 |
|
|
r = Rank.SUBTRIBE();
|
5498 |
|
|
}
|
5499 |
|
|
if (parentStr == null){
|
5500 |
|
|
parentStr = myName.getTribeStr();
|
5501 |
|
|
r = Rank.TRIBE();
|
5502 |
|
|
}
|
5503 |
|
|
if (parentStr == null){
|
5504 |
|
|
parentStr = myName.getSubfamilyStr();
|
5505 |
|
|
r = Rank.SUBFAMILY();
|
5506 |
|
|
}
|
5507 |
|
|
if (parentStr == null){
|
5508 |
|
|
parentStr = myName.getFamilyStr();
|
5509 |
|
|
r = Rank.FAMILY();
|
5510 |
|
|
}
|
5511 |
|
|
if(parentStr!=null){
|
5512 |
|
|
Taxon parent = handleParentName(ref, myName, parser, parentStr, r);
|
5513 |
6c41cae2
|
p.kelbert
|
//System.out.println("PUT IN HIERARCHY "+r+", "+parent);
|
5514 |
2b5a7979
|
p.kelbert
|
hierarchy.put(r,parent);
|
5515 |
|
|
}
|
5516 |
|
|
}
|
5517 |
|
|
|
5518 |
|
|
/**
|
5519 |
|
|
* @param ref
|
5520 |
|
|
* @param myName
|
5521 |
|
|
* @param parser
|
5522 |
|
|
* @param parentStr
|
5523 |
|
|
* @param r
|
5524 |
|
|
* @return
|
5525 |
|
|
*/
|
5526 |
6e861bce
|
Andreas Müller
|
private Taxon handleParentName(Reference<?> ref, MyName myName, INonViralNameParser<?> parser, String parentStr, Rank r) {
|
5527 |
2b5a7979
|
p.kelbert
|
NonViralName<?> parentNameName = (NonViralName<?>) parser.parseFullName(parentStr, nomenclaturalCode, r);
|
5528 |
|
|
Taxon parent = Taxon.NewInstance(parentNameName, ref); //sec set null
|
5529 |
|
|
// importer.getTaxonService().save(parent);
|
5530 |
|
|
// parent = CdmBase.deproxy(parent, Taxon.class);
|
5531 |
|
|
|
5532 |
|
|
boolean parentDoesNotExists = true;
|
5533 |
|
|
for (TaxonNode p : classification.getAllNodes()){
|
5534 |
|
|
if(p.getTaxon().getTitleCache().split("sec.")[0].trim().equalsIgnoreCase(parent.getTitleCache().split("sec.")[0].trim())) {
|
5535 |
|
|
// System.out.println(p.getTaxon().getUuid());
|
5536 |
|
|
// System.out.println(parent.getUuid());
|
5537 |
|
|
parentDoesNotExists = false;
|
5538 |
|
|
parent=CdmBase.deproxy(p.getTaxon(), Taxon.class);
|
5539 |
|
|
break;
|
5540 |
|
|
}
|
5541 |
|
|
}
|
5542 |
|
|
if(parentDoesNotExists) {
|
5543 |
|
|
Taxon tmp = findMatchingTaxon(parentNameName,ref);
|
5544 |
|
|
// System.out.println("FOUND PARENT "+tmp.toString()+" for "+parentNameName.toString());
|
5545 |
|
|
if(tmp ==null)
|
5546 |
|
|
{
|
5547 |
|
|
parent=Taxon.NewInstance(parentNameName, ref);
|
5548 |
|
|
importer.getTaxonService().save(parent);
|
5549 |
|
|
parent = CdmBase.deproxy(parent, Taxon.class);
|
5550 |
|
|
} else {
|
5551 |
|
|
parent=tmp;
|
5552 |
|
|
}
|
5553 |
|
|
lookForParentNode(parentNameName, parent, ref,myName);
|
5554 |
|
|
|
5555 |
|
|
}
|
5556 |
|
|
return parent;
|
5557 |
|
|
}
|
5558 |
|
|
|
5559 |
6c41cae2
|
p.kelbert
|
private void addNameDifferenceToFile(String originalname, String atomisedname){
|
5560 |
|
|
try{
|
5561 |
|
|
FileWriter fstream = new FileWriter("/home/pkelbert/Bureau/NamesDifferent_"+classification.getTitleCache()+".txt",true);
|
5562 |
|
|
BufferedWriter out = new BufferedWriter(fstream);
|
5563 |
|
|
out.write(originalname+" (original) versus "+replaceNull(atomisedname)+" (atomised) \n");
|
5564 |
|
|
//Close the output stream
|
5565 |
|
|
out.close();
|
5566 |
|
|
}catch (Exception e){//Catch exception if any
|
5567 |
|
|
System.err.println("Error: " + e.getMessage());
|
5568 |
|
|
}
|
5569 |
|
|
}
|
5570 |
2b5a7979
|
p.kelbert
|
/**
|
5571 |
|
|
* @param name
|
5572 |
|
|
* @param author
|
5573 |
|
|
* @param nomenclaturalCode2
|
5574 |
|
|
* @param rank
|
5575 |
|
|
*/
|
5576 |
|
|
private void addProblemNameToFile(String name, String author, NomenclaturalCode nomenclaturalCode2, Rank rank) {
|
5577 |
|
|
try{
|
5578 |
|
|
FileWriter fstream = new FileWriter("/home/pkelbert/Bureau/NameNotParsed.txt",true);
|
5579 |
|
|
BufferedWriter out = new BufferedWriter(fstream);
|
5580 |
|
|
out.write(name+"\t"+replaceNull(author)+"\t"+replaceNull(nomenclaturalCode2)+"\t"+replaceNull(rank)+"\n");
|
5581 |
|
|
//Close the output stream
|
5582 |
|
|
out.close();
|
5583 |
|
|
}catch (Exception e){//Catch exception if any
|
5584 |
|
|
System.err.println("Error: " + e.getMessage());
|
5585 |
|
|
}
|
5586 |
|
|
}
|
5587 |
|
|
|
5588 |
6c41cae2
|
p.kelbert
|
|
5589 |
|
|
/**
|
5590 |
|
|
* @param tnb
|
5591 |
|
|
* @param bestMatchingTaxon
|
5592 |
|
|
* @param insertAsExisting
|
5593 |
|
|
* @param refMods
|
5594 |
|
|
*/
|
5595 |
|
|
private void logDecision(NonViralName<?> tnb, Taxon bestMatchingTaxon, boolean insertAsExisting, Reference refMods) {
|
5596 |
|
|
try{
|
5597 |
|
|
FileWriter fstream = new FileWriter("/home/pkelbert/Bureau/Decisions_"+classification.toString()+".txt",true);
|
5598 |
|
|
BufferedWriter out = new BufferedWriter(fstream);
|
5599 |
|
|
out.write(tnb.getTitleCache()+" sec. "+refMods+"\t"+bestMatchingTaxon.getTitleCache()+"\t"+insertAsExisting+"\n");
|
5600 |
|
|
//Close the output stream
|
5601 |
|
|
out.close();
|
5602 |
|
|
}catch (Exception e){//Catch exception if any
|
5603 |
|
|
System.err.println("Error: " + e.getMessage());
|
5604 |
|
|
}
|
5605 |
|
|
}
|
5606 |
|
|
|
5607 |
|
|
|
5608 |
2b5a7979
|
p.kelbert
|
@SuppressWarnings("unused")
|
5609 |
|
|
private String replaceNull(Object in){
|
5610 |
|
|
if (in == null) {
|
5611 |
|
|
return "";
|
5612 |
|
|
}
|
5613 |
|
|
if (in.getClass().equals(NomenclaturalCode.class)) {
|
5614 |
|
|
return ((NomenclaturalCode)in).getTitleCache();
|
5615 |
|
|
}
|
5616 |
|
|
return in.toString();
|
5617 |
|
|
}
|
5618 |
|
|
|
5619 |
|
|
/**
|
5620 |
|
|
* @param fullName
|
5621 |
|
|
* @param nomenclaturalCode2
|
5622 |
|
|
* @param rank
|
5623 |
|
|
*/
|
5624 |
|
|
private void addProblemNameToFile(String type, String name, NomenclaturalCode nomenclaturalCode2, Rank rank, String problems) {
|
5625 |
|
|
try{
|
5626 |
|
|
FileWriter fstream = new FileWriter("/home/pkelbert/Bureau/NameNotParsed_"+classification.getTitleCache()+".txt",true);
|
5627 |
|
|
BufferedWriter out = new BufferedWriter(fstream);
|
5628 |
|
|
out.write(type+"\t"+name+"\t"+replaceNull(nomenclaturalCode2)+"\t"+replaceNull(rank)+"\t"+problems+"\n");
|
5629 |
|
|
//Close the output stream
|
5630 |
|
|
out.close();
|
5631 |
|
|
}catch (Exception e){//Catch exception if any
|
5632 |
|
|
System.err.println("Error: " + e.getMessage());
|
5633 |
|
|
}
|
5634 |
1de96f38
|
p.kelbert
|
|
5635 |
|
|
}
|
5636 |
285d1929
|
p.kelbert
|
|
5637 |
|
|
}
|
5638 |
1de96f38
|
p.kelbert
|
|
5639 |
|
|
|