1
|
/**
|
2
|
* Copyright (C) 2016 EDIT
|
3
|
* European Distributed Institute of Taxonomy
|
4
|
* http://www.e-taxonomy.eu
|
5
|
*
|
6
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
7
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
8
|
*/
|
9
|
package eu.etaxonomy.cdm.io.mexico;
|
10
|
|
11
|
import java.net.URISyntaxException;
|
12
|
import java.util.Arrays;
|
13
|
import java.util.List;
|
14
|
import java.util.Map;
|
15
|
import java.util.Set;
|
16
|
import java.util.UUID;
|
17
|
|
18
|
import org.apache.logging.log4j.LogManager;
|
19
|
import org.apache.logging.log4j.Logger;
|
20
|
import org.springframework.stereotype.Component;
|
21
|
|
22
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
23
|
import eu.etaxonomy.cdm.common.URI;
|
24
|
import eu.etaxonomy.cdm.common.UTF8;
|
25
|
import eu.etaxonomy.cdm.model.common.Extension;
|
26
|
import eu.etaxonomy.cdm.model.common.ExtensionType;
|
27
|
import eu.etaxonomy.cdm.model.common.Identifier;
|
28
|
import eu.etaxonomy.cdm.model.media.ExternalLinkType;
|
29
|
import eu.etaxonomy.cdm.model.name.IBotanicalName;
|
30
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
31
|
import eu.etaxonomy.cdm.model.name.Rank;
|
32
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
33
|
import eu.etaxonomy.cdm.model.name.TaxonNameFactory;
|
34
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
35
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
36
|
import eu.etaxonomy.cdm.model.taxon.Classification;
|
37
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
38
|
import eu.etaxonomy.cdm.model.taxon.TaxonNode;
|
39
|
import eu.etaxonomy.cdm.model.term.IdentifierType;
|
40
|
import eu.etaxonomy.cdm.model.term.TermVocabulary;
|
41
|
import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
|
42
|
|
43
|
/**
|
44
|
* @author a.mueller
|
45
|
* @since 16.06.2016
|
46
|
*/
|
47
|
@Component
|
48
|
public class MexicoBorhidiExcelImport<CONFIG extends MexicoBorhidiImportConfigurator>
|
49
|
extends SimpleExcelTaxonImport<CONFIG>{
|
50
|
|
51
|
private static final Logger logger = LogManager.getLogger();
|
52
|
private static final long serialVersionUID = -3607776356577606657L;
|
53
|
|
54
|
private static List<String> expectedKeys= Arrays.asList(new String[]{
|
55
|
"FullnameNoAuthors","OutputNameID","OutputFullNameWithAuthors","RefType"
|
56
|
,"OutputAbbreviatedTitle","OutputCollation","OutputVolume",
|
57
|
"OutputIssue","OutputPage","OutputTitlePageYear","OutputYearPublished",
|
58
|
"OutputBHLLink"});
|
59
|
|
60
|
@Override
|
61
|
protected void firstPass(SimpleExcelTaxonImportState<CONFIG> state) {
|
62
|
String line = state.getCurrentLine() + ": ";
|
63
|
Map<String, String> record = state.getOriginalRecord();
|
64
|
|
65
|
Set<String> keys = record.keySet();
|
66
|
checkAllKeysExist(line, keys, expectedKeys);
|
67
|
|
68
|
if (record.get("FullnameNoAuthors") == null ){
|
69
|
logger.warn("No FullnameNoAuthors given: " + line);
|
70
|
return;
|
71
|
}
|
72
|
|
73
|
//Name
|
74
|
IBotanicalName speciesName = makeName(record, state);
|
75
|
|
76
|
//Taxon
|
77
|
Reference sec = state.getConfig().getSecReference();
|
78
|
Taxon taxon = Taxon.NewInstance(speciesName, sec);
|
79
|
TaxonNode rubiaceae = getHighestNode(state);
|
80
|
|
81
|
taxon.addSource(makeOriginalSource(state));
|
82
|
|
83
|
//make genus
|
84
|
makeGenus(state, speciesName, sec, taxon, rubiaceae);
|
85
|
}
|
86
|
|
87
|
|
88
|
private Classification classification;
|
89
|
private TaxonNode rubiaceaeNode;
|
90
|
|
91
|
private TaxonNode getHighestNode(SimpleExcelTaxonImportState<CONFIG> state) {
|
92
|
if (rubiaceaeNode == null){
|
93
|
MexicoBorhidiImportConfigurator config = state.getConfig();
|
94
|
classification = Classification.NewInstance(state.getConfig().getClassificationName());
|
95
|
classification.setUuid(config.getClassificationUuid());
|
96
|
classification.setReference(config.getSecReference());
|
97
|
IBotanicalName nameRubiaceae = TaxonNameFactory.NewBotanicalInstance(Rank.FAMILY());
|
98
|
nameRubiaceae.setGenusOrUninomial("Rubiaceae");
|
99
|
Taxon rubiaceaeTaxon = Taxon.NewInstance(nameRubiaceae, classification.getReference());
|
100
|
rubiaceaeNode = classification.addChildTaxon(rubiaceaeTaxon, null, null);
|
101
|
getClassificationService().save(classification);
|
102
|
}
|
103
|
return rubiaceaeNode;
|
104
|
}
|
105
|
|
106
|
private IBotanicalName makeName(Map<String, String> record, SimpleExcelTaxonImportState<CONFIG> state) {
|
107
|
String line = state.getCurrentLine() + ": ";
|
108
|
|
109
|
String fullNameStr = getValue(record, "OutputFullNameWithAuthors");
|
110
|
// String volume = getValue(record, "OutputVolume");
|
111
|
// String issue = getValue(record, "OutputIssue");
|
112
|
// String page = getValue(record, "OutputPage");
|
113
|
String titleYear = getValue(record, "OutputTitlePageYear");
|
114
|
String publishedYear = getValue(record, "OutputYearPublished");
|
115
|
String refAbbrevTitle = getValue(record, "OutputAbbreviatedTitle");
|
116
|
String outputCollation = getValue(record, "OutputCollation");
|
117
|
String refType = getValue(record, "RefType");
|
118
|
|
119
|
|
120
|
TaxonName name = (TaxonName)nameParser.parseFullName(fullNameStr, NomenclaturalCode.ICNAFP, Rank.SPECIES());
|
121
|
if (name.isProtectedTitleCache()){
|
122
|
//for the 2 ined. names
|
123
|
name = (TaxonName)nameParser.parseReferencedName(fullNameStr, NomenclaturalCode.ICNAFP, Rank.SPECIES());
|
124
|
}
|
125
|
if (name.isProtectedTitleCache()){
|
126
|
logger.warn(line + "Name could not be parsed: " + fullNameStr );
|
127
|
}else{
|
128
|
replaceAuthorNamesAndNomRef(state, name);
|
129
|
}
|
130
|
|
131
|
if (refAbbrevTitle != null){
|
132
|
String[] volumeDetail = makeVolumeDetail(outputCollation);
|
133
|
String detail;
|
134
|
String volume = null;
|
135
|
if (volumeDetail.length > 1){
|
136
|
volume = volumeDetail[0].trim();
|
137
|
detail = volumeDetail[1].trim();
|
138
|
}else{
|
139
|
detail = volumeDetail[0].trim();
|
140
|
}
|
141
|
|
142
|
refAbbrevTitle = refAbbrevTitle.trim();
|
143
|
boolean isArticle = "A".equalsIgnoreCase(refType);
|
144
|
|
145
|
if (isArticle){
|
146
|
if (! "A".equalsIgnoreCase(refType)){
|
147
|
logger.warn(line + "RefType problem with article " + refType);
|
148
|
}
|
149
|
|
150
|
Reference journal = state.getReference(refAbbrevTitle);
|
151
|
if (journal == null){
|
152
|
journal = ReferenceFactory.newJournal();
|
153
|
journal.setAbbrevTitle(refAbbrevTitle);
|
154
|
state.putReference(refAbbrevTitle, journal);
|
155
|
journal.addSource(makeOriginalSource(state));
|
156
|
|
157
|
}
|
158
|
Reference article = ReferenceFactory.newArticle();
|
159
|
|
160
|
|
161
|
// String detail = page;
|
162
|
name.setNomenclaturalMicroReference(detail);
|
163
|
|
164
|
article.setVolume(CdmUtils.Ne(volume));
|
165
|
article.setInReference(journal);
|
166
|
|
167
|
titleYear = (isBlank(publishedYear)? titleYear : UTF8.QUOT_DBL_LOW9 + titleYear + UTF8.QUOT_DBL_HIGH_REV9 + "[" + publishedYear + "]");
|
168
|
article.setDatePublished(TimePeriodParser.parseStringVerbatim(titleYear));
|
169
|
|
170
|
article.setAuthorship(name.getCombinationAuthorship());
|
171
|
|
172
|
Reference existingArticle = state.getReference(article.getTitleCache());
|
173
|
if (existingArticle != null){
|
174
|
name.setNomenclaturalReference(existingArticle);
|
175
|
}else{
|
176
|
name.setNomenclaturalReference(article);
|
177
|
state.putReference(article.getTitleCache(), article);
|
178
|
article.addSource(makeOriginalSource(state));
|
179
|
}
|
180
|
}else{
|
181
|
if (! "B".equalsIgnoreCase(refType)){
|
182
|
logger.warn(line + "RefType problem with book" + refType);
|
183
|
}
|
184
|
|
185
|
Reference book = ReferenceFactory.newBook();
|
186
|
book.setAbbrevTitle(refAbbrevTitle);
|
187
|
|
188
|
//year
|
189
|
titleYear = (isBlank(publishedYear)? titleYear : UTF8.QUOT_DBL_LOW9 + titleYear + UTF8.QUOT_DBL_HIGH_REV9 + "[" + publishedYear + "]");
|
190
|
book.setDatePublished(TimePeriodParser.parseStringVerbatim(titleYear));
|
191
|
|
192
|
book.setAuthorship(name.getCombinationAuthorship());
|
193
|
|
194
|
//deduplicate
|
195
|
Reference existingBook = state.getReference(book.getTitleCache());
|
196
|
if (existingBook != null){
|
197
|
name.setNomenclaturalReference(existingBook);
|
198
|
}else{
|
199
|
name.setNomenclaturalReference(book);
|
200
|
state.putReference(book.getTitleCache(), book);
|
201
|
}
|
202
|
|
203
|
book.setVolume(volume);
|
204
|
|
205
|
//String detail = page;
|
206
|
name.setNomenclaturalMicroReference(detail);
|
207
|
}
|
208
|
}
|
209
|
|
210
|
addNomRefExtension(state, name);
|
211
|
|
212
|
//add protologue
|
213
|
String bhlLink = record.get("OutputBHLLink");
|
214
|
if (isNotBlank(bhlLink)){
|
215
|
try {
|
216
|
URI uri = new URI(bhlLink);
|
217
|
name.addProtologue(uri, null, ExternalLinkType.WebSite);
|
218
|
} catch (URISyntaxException e) {
|
219
|
logger.warn(line + "URI could not be parsed: " + e.getMessage());
|
220
|
}
|
221
|
}
|
222
|
|
223
|
//add tropicos identifier
|
224
|
String tropicosId = record.get("OutputNameID");
|
225
|
if (isNotBlank(tropicosId)){
|
226
|
String tropicosIdTypeLabel = "Tropicos Name Identifier";
|
227
|
UUID uuid = IdentifierType.uuidTropicosNameIdentifier;
|
228
|
TermVocabulary<IdentifierType> voc = null; //for now it goes to user defined voc
|
229
|
IdentifierType identifierType = this.getIdentiferType(state, uuid, tropicosIdTypeLabel, tropicosIdTypeLabel, null, voc);
|
230
|
Identifier identifier = Identifier.NewInstance(tropicosId, identifierType);
|
231
|
name.addIdentifier(identifier);
|
232
|
}
|
233
|
|
234
|
name.addSource(makeOriginalSource(state));
|
235
|
|
236
|
|
237
|
return name;
|
238
|
}
|
239
|
|
240
|
private String[] makeVolumeDetail(String outputCollation) {
|
241
|
if (outputCollation == null){
|
242
|
return new String[0];
|
243
|
}else{
|
244
|
String[] split = outputCollation.split(":");
|
245
|
return split;
|
246
|
}
|
247
|
}
|
248
|
|
249
|
private void addNomRefExtension(SimpleExcelTaxonImportState<CONFIG> state, IBotanicalName name) {
|
250
|
String newExtensionStr = name.getFullTitleCache() + " - BORHIDI";
|
251
|
UUID uuidNomRefExtension = MexicoConabioTransformer.uuidNomRefExtension;
|
252
|
for (Extension extension : name.getExtensions()){
|
253
|
if (extension.getType().getUuid().equals(uuidNomRefExtension)){
|
254
|
extension.setValue(extension.getValue() + "\n" + newExtensionStr);
|
255
|
return;
|
256
|
}
|
257
|
}
|
258
|
String label = "Nomenclatural reference in Sources";
|
259
|
String abbrev = "Nom. ref. src.";
|
260
|
ExtensionType extensionType = getExtensionType(state, uuidNomRefExtension, label, label, abbrev);
|
261
|
Extension.NewInstance((TaxonName)name, newExtensionStr, extensionType);
|
262
|
}
|
263
|
}
|