1 |
cce11612
|
Andreas Müller
|
/**
|
2 |
|
|
* Copyright (C) 2007 EDIT
|
3 |
8422c0cd
|
Andreas Müller
|
* European Distributed Institute of Taxonomy
|
4 |
cce11612
|
Andreas Müller
|
* http://www.e-taxonomy.eu
|
5 |
8422c0cd
|
Andreas Müller
|
*
|
6 |
cce11612
|
Andreas Müller
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
7 |
|
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
8 |
|
|
*/
|
9 |
|
|
|
10 |
|
|
package eu.etaxonomy.cdm.app.cyprus;
|
11 |
|
|
|
12 |
|
|
import java.io.FileNotFoundException;
|
13 |
|
|
import java.net.URI;
|
14 |
|
|
import java.net.URISyntaxException;
|
15 |
|
|
import java.util.HashSet;
|
16 |
2a6e6815
|
Andreas Müller
|
import java.util.List;
|
17 |
539b7820
|
Andreas Müller
|
import java.util.Map;
|
18 |
cce11612
|
Andreas Müller
|
import java.util.Set;
|
19 |
|
|
import java.util.UUID;
|
20 |
|
|
|
21 |
|
|
import org.apache.commons.lang.StringUtils;
|
22 |
|
|
import org.apache.log4j.Logger;
|
23 |
|
|
import org.springframework.transaction.TransactionStatus;
|
24 |
|
|
|
25 |
|
|
import eu.etaxonomy.cdm.api.application.CdmApplicationController;
|
26 |
|
|
import eu.etaxonomy.cdm.app.common.CdmDestinations;
|
27 |
|
|
import eu.etaxonomy.cdm.common.ExcelUtils;
|
28 |
|
|
import eu.etaxonomy.cdm.database.DbSchemaValidation;
|
29 |
|
|
import eu.etaxonomy.cdm.database.ICdmDataSource;
|
30 |
|
|
import eu.etaxonomy.cdm.io.api.application.CdmIoApplicationController;
|
31 |
|
|
import eu.etaxonomy.cdm.io.common.IImportConfigurator.CHECK;
|
32 |
|
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
33 |
|
|
import eu.etaxonomy.cdm.model.description.Feature;
|
34 |
|
|
import eu.etaxonomy.cdm.model.description.MeasurementUnit;
|
35 |
|
|
import eu.etaxonomy.cdm.model.description.QuantitativeData;
|
36 |
|
|
import eu.etaxonomy.cdm.model.description.StatisticalMeasure;
|
37 |
|
|
import eu.etaxonomy.cdm.model.description.StatisticalMeasurementValue;
|
38 |
|
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
39 |
86536e03
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
40 |
de81ab2a
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.reference.OriginalSourceType;
|
41 |
cce11612
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
42 |
|
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
43 |
|
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
44 |
|
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
45 |
|
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
46 |
|
|
|
47 |
|
|
/**
|
48 |
|
|
* @author a.mueller
|
49 |
a13538c8
|
Andreas Müller
|
* @since 16.12.2010
|
50 |
cce11612
|
Andreas Müller
|
*/
|
51 |
|
|
public class CyprusAltitudeActivator {
|
52 |
|
|
private static final Logger logger = Logger.getLogger(CyprusAltitudeActivator.class);
|
53 |
8422c0cd
|
Andreas Müller
|
|
54 |
cce11612
|
Andreas Müller
|
//database validation status (create, update, validate ...)
|
55 |
|
|
static DbSchemaValidation hbm2dll = DbSchemaValidation.VALIDATE;
|
56 |
|
|
static final URI source = cyprus_altitude();
|
57 |
|
|
|
58 |
8422c0cd
|
Andreas Müller
|
|
59 |
9cb92253
|
Andreas Müller
|
static final ICdmDataSource cdmDestination = CdmDestinations.localH2();
|
60 |
|
|
// static final ICdmDataSource cdmDestination = CdmDestinations.cdm_test_local_mysql_test();
|
61 |
|
|
// static final ICdmDataSource cdmDestination = CdmDestinations.cdm_cyprus_dev();
|
62 |
cce11612
|
Andreas Müller
|
// static final ICdmDataSource cdmDestination = CdmDestinations.cdm_cyprus_production();
|
63 |
|
|
|
64 |
8422c0cd
|
Andreas Müller
|
|
65 |
cce11612
|
Andreas Müller
|
//feature tree uuid
|
66 |
|
|
public static final UUID featureTreeUuid = UUID.fromString("14d1e912-5ec2-4d10-878b-828788b70a87");
|
67 |
8422c0cd
|
Andreas Müller
|
|
68 |
cce11612
|
Andreas Müller
|
//classification
|
69 |
|
|
static final UUID classificationUuid = UUID.fromString("0c2b5d25-7b15-4401-8b51-dd4be0ee5cab");
|
70 |
8422c0cd
|
Andreas Müller
|
|
71 |
cce11612
|
Andreas Müller
|
private static final String sourceReferenceTitle = "Cyprus Excel Altitude Import";
|
72 |
8422c0cd
|
Andreas Müller
|
|
73 |
|
|
|
74 |
cce11612
|
Andreas Müller
|
//TODO move to Feature vocabulary
|
75 |
8969b71c
|
Andreas Müller
|
private static final UUID uuidAltitudeFeature = UUID.fromString("1a28ed59-e15f-4001-b5c2-ea89f0012671");
|
76 |
8422c0cd
|
Andreas Müller
|
|
77 |
cce11612
|
Andreas Müller
|
//check - import
|
78 |
|
|
static final CHECK check = CHECK.IMPORT_WITHOUT_CHECK;
|
79 |
8422c0cd
|
Andreas Müller
|
|
80 |
cce11612
|
Andreas Müller
|
private void doImport(ICdmDataSource cdmDestination){
|
81 |
8422c0cd
|
Andreas Müller
|
|
82 |
539b7820
|
Andreas Müller
|
List<Map<String, String>> excel;
|
83 |
cce11612
|
Andreas Müller
|
try {
|
84 |
|
|
excel = ExcelUtils.parseXLS(source, "coreTax");
|
85 |
|
|
} catch (FileNotFoundException e) {
|
86 |
|
|
e.printStackTrace();
|
87 |
|
|
return;
|
88 |
|
|
}
|
89 |
8422c0cd
|
Andreas Müller
|
|
90 |
cce11612
|
Andreas Müller
|
CdmApplicationController app = CdmIoApplicationController.NewInstance(cdmDestination, hbm2dll);
|
91 |
|
|
|
92 |
2e477cf4
|
Andreas Müller
|
Set<TaxonBase> taxaToSave = new HashSet<>();
|
93 |
8422c0cd
|
Andreas Müller
|
|
94 |
cce11612
|
Andreas Müller
|
TransactionStatus tx = app.startTransaction();
|
95 |
8422c0cd
|
Andreas Müller
|
|
96 |
cce11612
|
Andreas Müller
|
UUID uuidMikle77 = UUID.fromString("9f5fa7ee-538b-4ae5-bd82-2a9503fea1d6");
|
97 |
|
|
UUID uuidMikle85 = UUID.fromString("994403c4-c400-413d-9a1a-8531a40bfd8c");
|
98 |
8422c0cd
|
Andreas Müller
|
|
99 |
|
|
Reference mikle77 = app.getReferenceService().find(uuidMikle77);
|
100 |
|
|
Reference mikle85 = app.getReferenceService().find(uuidMikle85);
|
101 |
|
|
|
102 |
cce11612
|
Andreas Müller
|
Feature altitudeFeature = (Feature) app.getTermService().find(uuidAltitudeFeature);
|
103 |
|
|
if (altitudeFeature == null){
|
104 |
8969b71c
|
Andreas Müller
|
throw new RuntimeException("Could not find altitudinal range feature");
|
105 |
cce11612
|
Andreas Müller
|
}
|
106 |
8422c0cd
|
Andreas Müller
|
|
107 |
cce11612
|
Andreas Müller
|
MeasurementUnit meter = (MeasurementUnit)app.getTermService().find(UUID.fromString("8bef5055-789c-41e5-bea2-8dc2ea8ecdf6"));
|
108 |
8422c0cd
|
Andreas Müller
|
|
109 |
cce11612
|
Andreas Müller
|
int count =1;
|
110 |
539b7820
|
Andreas Müller
|
for (Map<String, String> row : excel){
|
111 |
cce11612
|
Andreas Müller
|
count++;
|
112 |
|
|
UUID baseUuid = makeUuid(row, "uuid");
|
113 |
|
|
UUID acceptedUuid = makeUuid(row, "acceptedNameUuid");
|
114 |
|
|
UUID parentUuid = makeUuid(row, "parentUuid");
|
115 |
8422c0cd
|
Andreas Müller
|
|
116 |
8969b71c
|
Andreas Müller
|
String altitudeMin = row.get("Min");
|
117 |
|
|
String altitudeMax = row.get("Max");
|
118 |
|
|
String acceptedName = row.get("AcceptedName");
|
119 |
8422c0cd
|
Andreas Müller
|
|
120 |
cce11612
|
Andreas Müller
|
String source = row.get("Source");
|
121 |
8422c0cd
|
Andreas Müller
|
|
122 |
8969b71c
|
Andreas Müller
|
if (StringUtils.isBlank(altitudeMin)){
|
123 |
cce11612
|
Andreas Müller
|
continue;
|
124 |
|
|
}
|
125 |
8422c0cd
|
Andreas Müller
|
|
126 |
cce11612
|
Andreas Müller
|
boolean hasAltitude = false;
|
127 |
8422c0cd
|
Andreas Müller
|
Reference sourceRef = getSource(source, mikle77, mikle85);
|
128 |
8969b71c
|
Andreas Müller
|
Taxon taxon = getTaxon(app, baseUuid, acceptedUuid, parentUuid, acceptedName, count);
|
129 |
cce11612
|
Andreas Müller
|
if (taxon != null){
|
130 |
|
|
TaxonDescription desc = getDescription(taxon, sourceRef);
|
131 |
8422c0cd
|
Andreas Müller
|
|
132 |
8969b71c
|
Andreas Müller
|
hasAltitude = makeAltitude(altitudeMin, altitudeMax, altitudeFeature, sourceRef, desc, meter, count);
|
133 |
cce11612
|
Andreas Müller
|
if (hasAltitude){
|
134 |
|
|
if(desc.getTaxon() == null){
|
135 |
|
|
taxon.addDescription(desc);
|
136 |
|
|
}
|
137 |
|
|
taxaToSave.add(taxon);
|
138 |
8969b71c
|
Andreas Müller
|
}else{
|
139 |
|
|
logger.warn("HasALtitude is false in " + count);
|
140 |
cce11612
|
Andreas Müller
|
}
|
141 |
|
|
}else{
|
142 |
|
|
logger.warn("Taxon not recognized in line " + count);
|
143 |
|
|
}
|
144 |
|
|
}
|
145 |
|
|
|
146 |
|
|
app.getTaxonService().saveOrUpdate(taxaToSave);
|
147 |
8422c0cd
|
Andreas Müller
|
|
148 |
cce11612
|
Andreas Müller
|
app.commitTransaction(tx);
|
149 |
|
|
}
|
150 |
|
|
|
151 |
8969b71c
|
Andreas Müller
|
|
152 |
|
|
private Taxon getTaxon(CdmApplicationController app, UUID baseUuid, UUID acceptedUuid, UUID parentUuid, String acceptedName, int row) {
|
153 |
cce11612
|
Andreas Müller
|
TaxonBase<?> base = app.getTaxonService().find(baseUuid);
|
154 |
8422c0cd
|
Andreas Müller
|
|
155 |
cce11612
|
Andreas Müller
|
Taxon result = null;
|
156 |
|
|
if (base.isInstanceOf(Taxon.class)){
|
157 |
|
|
Taxon t = CdmBase.deproxy(base, Taxon.class);
|
158 |
|
|
if (t.getTaxonNodes().size() == 1 && t.getTaxonNodes().iterator().next().getClassification().getUuid().equals(classificationUuid)){
|
159 |
|
|
result = t;
|
160 |
|
|
}else{
|
161 |
|
|
logger.warn("Base taxon (uuid) not in classification. Row: " + row + ", Taxon: " + base.getTitleCache());
|
162 |
|
|
}
|
163 |
|
|
}
|
164 |
|
|
if (result == null){
|
165 |
|
|
TaxonBase<?> accepted = app.getTaxonService().find(acceptedUuid);
|
166 |
|
|
Taxon t = CdmBase.deproxy(accepted, Taxon.class);;
|
167 |
|
|
if (t.getTaxonNodes().size() == 1 && t.getTaxonNodes().iterator().next().getClassification().getUuid().equals(classificationUuid)){
|
168 |
|
|
if (hasSynonym(t, base)){
|
169 |
|
|
result = t;
|
170 |
|
|
}else{
|
171 |
|
|
logger.warn("Synonym relation has changed somehow. Row: " + row + ", Taxon: " + base.getTitleCache());
|
172 |
|
|
}
|
173 |
8422c0cd
|
Andreas Müller
|
|
174 |
cce11612
|
Andreas Müller
|
}else{
|
175 |
|
|
logger.warn("Accepted taxon not in classification. Row: " + row + ", Taxon: " + base.getTitleCache());
|
176 |
|
|
}
|
177 |
|
|
}
|
178 |
8422c0cd
|
Andreas Müller
|
|
179 |
8969b71c
|
Andreas Müller
|
if (result != null){
|
180 |
|
|
if (! result.getName().getTitleCache().equals(acceptedName)){
|
181 |
|
|
logger.warn("AcceptedName and taxon name is not equal in " + row + ".\n" +
|
182 |
|
|
" Accepted Name: " + acceptedName + ";\n" +
|
183 |
|
|
" Taxon Name: " + result.getName().getTitleCache());
|
184 |
|
|
}
|
185 |
|
|
}
|
186 |
8422c0cd
|
Andreas Müller
|
|
187 |
cce11612
|
Andreas Müller
|
return result;
|
188 |
|
|
}
|
189 |
|
|
|
190 |
|
|
private boolean hasSynonym(Taxon t, TaxonBase<?> base) {
|
191 |
|
|
if (base.isInstanceOf(Synonym.class)){
|
192 |
3ef2e1bd
|
Andreas Müller
|
for (Synonym syn : t.getSynonyms()){
|
193 |
|
|
if (syn.equals(base)){
|
194 |
cce11612
|
Andreas Müller
|
return true;
|
195 |
|
|
}
|
196 |
|
|
}
|
197 |
|
|
}
|
198 |
|
|
return false;
|
199 |
|
|
}
|
200 |
|
|
|
201 |
9cb92253
|
Andreas Müller
|
// private static final Pattern altitudePattern = Pattern.compile("\\d{1,4}(-\\d{1,4})?");
|
202 |
cce11612
|
Andreas Müller
|
|
203 |
|
|
|
204 |
8422c0cd
|
Andreas Müller
|
private boolean makeAltitude(String altitudeMin, String altitudeMax, Feature altitudeFeature,
|
205 |
|
|
Reference sourceRef, TaxonDescription desc, MeasurementUnit meter, int row) {
|
206 |
|
|
|
207 |
8969b71c
|
Andreas Müller
|
QuantitativeData data = QuantitativeData.NewInstance(altitudeFeature);
|
208 |
8422c0cd
|
Andreas Müller
|
|
209 |
8969b71c
|
Andreas Müller
|
//Meikle
|
210 |
|
|
if (source != null){
|
211 |
ea7deae0
|
Andreas Müller
|
TaxonName nameUsedInSource = null; //TODO
|
212 |
8969b71c
|
Andreas Müller
|
data.addSource(OriginalSourceType.PrimaryTaxonomicSource, null, null, sourceRef, null, nameUsedInSource, null);
|
213 |
|
|
}
|
214 |
|
|
data.setUnit(meter);
|
215 |
8422c0cd
|
Andreas Müller
|
|
216 |
8969b71c
|
Andreas Müller
|
Integer min = Integer.valueOf(altitudeMin);
|
217 |
|
|
StatisticalMeasurementValue minValue = StatisticalMeasurementValue.NewInstance(StatisticalMeasure.MIN(), min);
|
218 |
|
|
data.addStatisticalValue(minValue);
|
219 |
|
|
|
220 |
|
|
Integer max = Integer.valueOf(altitudeMax);
|
221 |
|
|
StatisticalMeasurementValue maxValue = StatisticalMeasurementValue.NewInstance(StatisticalMeasure.MAX(), max);
|
222 |
|
|
data.addStatisticalValue(maxValue);
|
223 |
8422c0cd
|
Andreas Müller
|
|
224 |
8969b71c
|
Andreas Müller
|
desc.addElement(data);
|
225 |
|
|
return true;
|
226 |
|
|
}
|
227 |
8422c0cd
|
Andreas Müller
|
|
228 |
|
|
private TaxonDescription getDescription(Taxon taxon, Reference sourceRef) {
|
229 |
cce11612
|
Andreas Müller
|
if (taxon != null){
|
230 |
|
|
//TODO Mikle existiert derzeit nicht also Source
|
231 |
8422c0cd
|
Andreas Müller
|
|
232 |
cce11612
|
Andreas Müller
|
TaxonDescription desc = TaxonDescription.NewInstance();
|
233 |
|
|
desc.setTitleCache("Import from " + getSourceReference().getTitleCache(), true);
|
234 |
|
|
desc.addSource(OriginalSourceType.PrimaryTaxonomicSource, null, null, sourceRef,null);
|
235 |
|
|
desc.addSource(OriginalSourceType.Import, null, null, getSourceReference(), null);
|
236 |
8422c0cd
|
Andreas Müller
|
|
237 |
cce11612
|
Andreas Müller
|
return desc;
|
238 |
|
|
}
|
239 |
|
|
return null;
|
240 |
|
|
}
|
241 |
|
|
|
242 |
8422c0cd
|
Andreas Müller
|
private Reference getSource(String source, Reference m77, Reference m85) {
|
243 |
cce11612
|
Andreas Müller
|
if(StringUtils.isNotBlank(source)){
|
244 |
|
|
if (source.equals("Meikle 1977")){
|
245 |
|
|
return m77;
|
246 |
|
|
}else if (source.equals("Meikle 1985")){
|
247 |
|
|
return m85;
|
248 |
|
|
}else{
|
249 |
|
|
logger.warn("Source not recognized: " + source);
|
250 |
|
|
}
|
251 |
|
|
}
|
252 |
|
|
return null;
|
253 |
|
|
}
|
254 |
|
|
|
255 |
|
|
/**
|
256 |
|
|
* @param row
|
257 |
|
|
* @return
|
258 |
|
|
*/
|
259 |
539b7820
|
Andreas Müller
|
private UUID makeUuid(Map<String, String> row, String colName) {
|
260 |
cce11612
|
Andreas Müller
|
if (StringUtils.isBlank(row.get(colName))){
|
261 |
|
|
return null;
|
262 |
|
|
}else{
|
263 |
|
|
return UUID.fromString(row.get(colName));
|
264 |
|
|
}
|
265 |
|
|
}
|
266 |
|
|
|
267 |
8422c0cd
|
Andreas Müller
|
Reference sourceReference;
|
268 |
|
|
private Reference getSourceReference() {
|
269 |
cce11612
|
Andreas Müller
|
if (sourceReference == null){
|
270 |
|
|
sourceReference = ReferenceFactory.newGeneric();
|
271 |
|
|
sourceReference.setTitleCache(sourceReferenceTitle, true);
|
272 |
|
|
}
|
273 |
|
|
return sourceReference;
|
274 |
8422c0cd
|
Andreas Müller
|
|
275 |
cce11612
|
Andreas Müller
|
}
|
276 |
|
|
|
277 |
|
|
//Cyprus
|
278 |
|
|
public static URI cyprus_altitude() {
|
279 |
|
|
URI sourceUrl;
|
280 |
|
|
try {
|
281 |
8969b71c
|
Andreas Müller
|
sourceUrl = new URI("file:/F:/data/cyprus/Cyprus-altitude-import-neu.xls");
|
282 |
|
|
// sourceUrl = new URI("file:/F:/data/cyprus/Zypern-Altitude.xls");
|
283 |
cce11612
|
Andreas Müller
|
return sourceUrl;
|
284 |
|
|
} catch (URISyntaxException e) {
|
285 |
|
|
e.printStackTrace();
|
286 |
|
|
return null;
|
287 |
|
|
}
|
288 |
|
|
}
|
289 |
|
|
|
290 |
|
|
|
291 |
|
|
/**
|
292 |
|
|
* @param args
|
293 |
|
|
*/
|
294 |
|
|
public static void main(String[] args) {
|
295 |
|
|
CyprusAltitudeActivator me = new CyprusAltitudeActivator();
|
296 |
|
|
me.doImport(cdmDestination);
|
297 |
|
|
}
|
298 |
8422c0cd
|
Andreas Müller
|
|
299 |
cce11612
|
Andreas Müller
|
}
|