1
|
// $Id$
|
2
|
/**
|
3
|
* Copyright (C) 2007 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.app.wp6.diptera;
|
11
|
|
12
|
import java.io.File;
|
13
|
import java.io.FileInputStream;
|
14
|
import java.io.InputStream;
|
15
|
import java.io.InputStreamReader;
|
16
|
import java.util.ArrayList;
|
17
|
import java.util.HashMap;
|
18
|
import java.util.List;
|
19
|
import java.util.Map;
|
20
|
|
21
|
import org.apache.commons.lang.StringUtils;
|
22
|
import org.apache.log4j.Logger;
|
23
|
import org.springframework.transaction.TransactionStatus;
|
24
|
|
25
|
import au.com.bytecode.opencsv.CSVReader;
|
26
|
import eu.etaxonomy.cdm.api.application.CdmApplicationController;
|
27
|
import eu.etaxonomy.cdm.app.common.CdmDestinations;
|
28
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
29
|
import eu.etaxonomy.cdm.database.DbSchemaValidation;
|
30
|
import eu.etaxonomy.cdm.database.ICdmDataSource;
|
31
|
import eu.etaxonomy.cdm.model.agent.Institution;
|
32
|
import eu.etaxonomy.cdm.model.occurrence.Collection;
|
33
|
import eu.etaxonomy.cdm.model.occurrence.DerivedUnit;
|
34
|
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationBase;
|
35
|
|
36
|
/**
|
37
|
* @author a.mueller
|
38
|
* @date 07.04.2010
|
39
|
*
|
40
|
*/
|
41
|
public class DipteraCollectionImport {
|
42
|
private static final Logger logger = Logger.getLogger(DipteraCollectionImport.class);
|
43
|
|
44
|
public static final File acronymsFile = new File("src/main/resources/collections/Acronyms.tab");
|
45
|
//datasource for use from local main()
|
46
|
static final ICdmDataSource cdmDestination = CdmDestinations.localH2();
|
47
|
|
48
|
|
49
|
public boolean invoke(ICdmDataSource dataSource) {
|
50
|
CdmApplicationController cdmApp = CdmApplicationController.NewInstance(dataSource, DbSchemaValidation.VALIDATE);
|
51
|
|
52
|
//create collections
|
53
|
TransactionStatus tx = cdmApp.startTransaction();
|
54
|
Map<String, Collection> colletionMap = createCollections(cdmApp);
|
55
|
|
56
|
//add collections to specimen
|
57
|
addCollectionsToSpecimen(cdmApp, colletionMap);
|
58
|
cdmApp.commitTransaction(tx);
|
59
|
|
60
|
return true;
|
61
|
|
62
|
}
|
63
|
|
64
|
|
65
|
/**
|
66
|
* @param cdmApp
|
67
|
* @param colletionMap
|
68
|
*/
|
69
|
private void addCollectionsToSpecimen(CdmApplicationController cdmApp, Map<String, Collection> colletionMap) {
|
70
|
List<DerivedUnit> specimens = cdmApp.getOccurrenceService().list(DerivedUnit.class, null, null, null, null);
|
71
|
for (SpecimenOrObservationBase<?> specOrObservBase : specimens){
|
72
|
if (specOrObservBase.getRecordBasis().isPreservedSpecimen()){
|
73
|
handleSingleSpecimen((DerivedUnit)specOrObservBase, colletionMap);
|
74
|
}else{
|
75
|
logger.warn("There are specimenOrObservationBase objects which are not of class Specimen. This is probably an error.");
|
76
|
}
|
77
|
}
|
78
|
List<SpecimenOrObservationBase> specimenList = new ArrayList<SpecimenOrObservationBase>(specimens);
|
79
|
cdmApp.getOccurrenceService().save(specimenList);
|
80
|
}
|
81
|
|
82
|
|
83
|
/**
|
84
|
* @param specimen
|
85
|
* @param colletionMap
|
86
|
*/
|
87
|
private void handleSingleSpecimen(DerivedUnit specimen, Map<String, Collection> collectionMap) {
|
88
|
String titleCache = specimen.getTitleCache();
|
89
|
String collectionCode = getCollectionCode(titleCache);
|
90
|
if (StringUtils.isBlank(collectionCode)){
|
91
|
logger.warn("Collection code is empty for: " + titleCache);
|
92
|
}else{
|
93
|
Collection collection = collectionMap.get(collectionCode);
|
94
|
if (collection != null){
|
95
|
specimen.setCollection(collection);
|
96
|
}else{
|
97
|
logger.warn("Collection not found for code: " + collectionCode + "; titleCache: " + titleCache);
|
98
|
}
|
99
|
}
|
100
|
}
|
101
|
|
102
|
|
103
|
/**
|
104
|
* @param titleCache
|
105
|
* @return
|
106
|
*/
|
107
|
private String getCollectionCode(String titleCache) {
|
108
|
String result = titleCache.trim();
|
109
|
result = replaceBracket(result);
|
110
|
result = replaceLastFullStop(result);
|
111
|
result = replaceLastQuestionMark(result);
|
112
|
result = parseLastUpperCase(result);
|
113
|
return result;
|
114
|
}
|
115
|
|
116
|
|
117
|
/**
|
118
|
* @param result
|
119
|
* @return
|
120
|
*/
|
121
|
private String parseLastUpperCase(String string) {
|
122
|
String result = "";
|
123
|
String tmpString = string;
|
124
|
int pos = tmpString.lastIndexOf(" ");
|
125
|
if (pos>-1){
|
126
|
tmpString = tmpString.substring(pos+1);
|
127
|
}
|
128
|
while (tmpString.length() > 0){
|
129
|
int len = tmpString.length();
|
130
|
char lastChar = tmpString.charAt(len-1);
|
131
|
if (Character.isUpperCase( lastChar)){
|
132
|
result = lastChar + result;
|
133
|
}else{
|
134
|
if (result.length() > 0){
|
135
|
logger.warn("Collection code is not space separated: " + string);
|
136
|
}
|
137
|
break;
|
138
|
}
|
139
|
//remove last character
|
140
|
tmpString = tmpString.substring(0, tmpString.length()-1);
|
141
|
}
|
142
|
return result;
|
143
|
}
|
144
|
|
145
|
|
146
|
|
147
|
/**
|
148
|
* @param result
|
149
|
* @return
|
150
|
*/
|
151
|
private String replaceLastQuestionMark(String string) {
|
152
|
if (string.endsWith("?")){
|
153
|
string = string.substring(0,string.length()-1).trim();
|
154
|
}
|
155
|
return string;
|
156
|
}
|
157
|
|
158
|
/**
|
159
|
* @param result
|
160
|
* @return
|
161
|
*/
|
162
|
private String replaceLastFullStop(String string) {
|
163
|
if (string.endsWith(".")){
|
164
|
string = string.substring(0,string.length()-1).trim();
|
165
|
}
|
166
|
return string;
|
167
|
}
|
168
|
|
169
|
|
170
|
/**
|
171
|
* @param result
|
172
|
* @return
|
173
|
*/
|
174
|
private String replaceBracket(String string) {
|
175
|
if (string.endsWith("]")){
|
176
|
int pos = string.indexOf("[");
|
177
|
if (pos >0){
|
178
|
string = string.substring(0, pos).trim();
|
179
|
}else{
|
180
|
logger.warn("Closing bracket has no opening bracket in: " + string);
|
181
|
}
|
182
|
}
|
183
|
return string;
|
184
|
}
|
185
|
|
186
|
|
187
|
/**
|
188
|
* @param cdmApp
|
189
|
*/
|
190
|
private Map<String, Collection> createCollections(CdmApplicationController cdmApp) {
|
191
|
Map<String, Collection> collectionMap = new HashMap<String, Collection>();
|
192
|
List<String[]> lines = getLines();
|
193
|
for (String[] line:lines){
|
194
|
Collection collection = makeLine(line);
|
195
|
collectionMap.put(collection.getCode(), collection);
|
196
|
}
|
197
|
cdmApp.getCollectionService().save(collectionMap.values());
|
198
|
// for (Collection collection: collectionMap.values()){
|
199
|
// System.out.println(collection.getTitleCache());
|
200
|
// }
|
201
|
return collectionMap;
|
202
|
}
|
203
|
|
204
|
|
205
|
private Collection makeLine(String[] line) {
|
206
|
String code = line[0];
|
207
|
String instituteName = line[1];
|
208
|
String lowerInstitutionName = line[2];
|
209
|
String higherInstitutionName = line[3];
|
210
|
String location = line[4];
|
211
|
String country = line[5];
|
212
|
//create objects
|
213
|
Collection collection = Collection.NewInstance();
|
214
|
collection.setCode(code);
|
215
|
Institution institution = Institution.NewInstance();
|
216
|
institution.setCode(code);
|
217
|
|
218
|
institution.setName(instituteName);
|
219
|
|
220
|
if (StringUtils.isNotBlank(lowerInstitutionName)){
|
221
|
Institution lowerInstitution = Institution.NewInstance();
|
222
|
lowerInstitution.setName(lowerInstitutionName);
|
223
|
lowerInstitution.setIsPartOf(institution);
|
224
|
}
|
225
|
|
226
|
if (StringUtils.isNotBlank(higherInstitutionName)){
|
227
|
Institution higherInstitution = Institution.NewInstance();
|
228
|
higherInstitution.setName(higherInstitutionName);
|
229
|
institution.setIsPartOf(higherInstitution);
|
230
|
}
|
231
|
|
232
|
collection.setInstitute(institution);
|
233
|
String locationAndCountry = CdmUtils.concat("/", location, country);
|
234
|
collection.setTownOrLocation(locationAndCountry);
|
235
|
|
236
|
String titleCache = CdmUtils.concat(", ", new String[]{instituteName, lowerInstitutionName, higherInstitutionName, location, country});
|
237
|
collection.setTitleCache(titleCache, true);
|
238
|
|
239
|
return collection;
|
240
|
}
|
241
|
|
242
|
|
243
|
|
244
|
|
245
|
private List<String[]> getLines() {
|
246
|
List<String[]> result = new ArrayList<String[]>();
|
247
|
|
248
|
try {
|
249
|
InputStream inStream = new FileInputStream(acronymsFile);
|
250
|
InputStreamReader inputStreamReader = new InputStreamReader(inStream, "UTF8");
|
251
|
CSVReader reader = new CSVReader(inputStreamReader, '\t');
|
252
|
String [] nextLine = reader.readNext();
|
253
|
|
254
|
|
255
|
while ((nextLine = reader.readNext()) != null) {
|
256
|
if (nextLine.length == 0){
|
257
|
continue;
|
258
|
}
|
259
|
result.add(nextLine);
|
260
|
}
|
261
|
return result;
|
262
|
} catch (Exception e) {
|
263
|
logger.error(e + " " + e.getCause() + " " + e.getMessage());
|
264
|
for(StackTraceElement ste : e.getStackTrace()) {
|
265
|
logger.error(ste);
|
266
|
}
|
267
|
throw new RuntimeException(e);
|
268
|
}
|
269
|
}
|
270
|
|
271
|
|
272
|
|
273
|
|
274
|
|
275
|
/**
|
276
|
* @param args
|
277
|
*/
|
278
|
public static void main(String[] args) {
|
279
|
try {
|
280
|
DipteraCollectionImport collectionImport = new DipteraCollectionImport();
|
281
|
collectionImport.invoke(cdmDestination);
|
282
|
// String titleCache = "Peru. Mouth of Rio Pachitea. ST 2R SMT. [fig. of male abdomen]";
|
283
|
// String collectionCode = collectionImport.getCollectionCode(titleCache);
|
284
|
// System.out.println(collectionCode);
|
285
|
} catch (Exception e) {
|
286
|
e.printStackTrace();
|
287
|
System.exit(-1);
|
288
|
}
|
289
|
}
|
290
|
|
291
|
}
|