1 |
275ada95
|
Andreas Müller
|
// $Id$
|
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 |
50818ab7
|
Katja Luther
|
import eu.etaxonomy.cdm.io.common.ImportResult;
|
32 |
275ada95
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.agent.Institution;
|
33 |
|
|
import eu.etaxonomy.cdm.model.occurrence.Collection;
|
34 |
|
|
import eu.etaxonomy.cdm.model.occurrence.DerivedUnit;
|
35 |
|
|
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationBase;
|
36 |
|
|
|
37 |
|
|
/**
|
38 |
|
|
* @author a.mueller
|
39 |
|
|
* @date 07.04.2010
|
40 |
|
|
*
|
41 |
|
|
*/
|
42 |
|
|
public class DipteraCollectionImport {
|
43 |
|
|
private static final Logger logger = Logger.getLogger(DipteraCollectionImport.class);
|
44 |
|
|
|
45 |
|
|
public static final File acronymsFile = new File("src/main/resources/collections/Acronyms.tab");
|
46 |
|
|
//datasource for use from local main()
|
47 |
|
|
static final ICdmDataSource cdmDestination = CdmDestinations.localH2();
|
48 |
|
|
|
49 |
|
|
|
50 |
50818ab7
|
Katja Luther
|
public ImportResult invoke(ICdmDataSource dataSource) {
|
51 |
|
|
ImportResult result = new ImportResult();
|
52 |
275ada95
|
Andreas Müller
|
CdmApplicationController cdmApp = CdmApplicationController.NewInstance(dataSource, DbSchemaValidation.VALIDATE);
|
53 |
|
|
|
54 |
|
|
//create collections
|
55 |
|
|
TransactionStatus tx = cdmApp.startTransaction();
|
56 |
|
|
Map<String, Collection> colletionMap = createCollections(cdmApp);
|
57 |
|
|
|
58 |
|
|
//add collections to specimen
|
59 |
|
|
addCollectionsToSpecimen(cdmApp, colletionMap);
|
60 |
|
|
cdmApp.commitTransaction(tx);
|
61 |
|
|
|
62 |
50818ab7
|
Katja Luther
|
return result;
|
63 |
275ada95
|
Andreas Müller
|
|
64 |
|
|
}
|
65 |
|
|
|
66 |
|
|
|
67 |
|
|
/**
|
68 |
|
|
* @param cdmApp
|
69 |
|
|
* @param colletionMap
|
70 |
|
|
*/
|
71 |
|
|
private void addCollectionsToSpecimen(CdmApplicationController cdmApp, Map<String, Collection> colletionMap) {
|
72 |
|
|
List<DerivedUnit> specimens = cdmApp.getOccurrenceService().list(DerivedUnit.class, null, null, null, null);
|
73 |
|
|
for (SpecimenOrObservationBase<?> specOrObservBase : specimens){
|
74 |
|
|
if (specOrObservBase.getRecordBasis().isPreservedSpecimen()){
|
75 |
|
|
handleSingleSpecimen((DerivedUnit)specOrObservBase, colletionMap);
|
76 |
|
|
}else{
|
77 |
|
|
logger.warn("There are specimenOrObservationBase objects which are not of class Specimen. This is probably an error.");
|
78 |
|
|
}
|
79 |
|
|
}
|
80 |
|
|
List<SpecimenOrObservationBase> specimenList = new ArrayList<SpecimenOrObservationBase>(specimens);
|
81 |
|
|
cdmApp.getOccurrenceService().save(specimenList);
|
82 |
|
|
}
|
83 |
|
|
|
84 |
|
|
|
85 |
|
|
/**
|
86 |
|
|
* @param specimen
|
87 |
|
|
* @param colletionMap
|
88 |
|
|
*/
|
89 |
|
|
private void handleSingleSpecimen(DerivedUnit specimen, Map<String, Collection> collectionMap) {
|
90 |
|
|
String titleCache = specimen.getTitleCache();
|
91 |
|
|
String collectionCode = getCollectionCode(titleCache);
|
92 |
|
|
if (StringUtils.isBlank(collectionCode)){
|
93 |
|
|
logger.warn("Collection code is empty for: " + titleCache);
|
94 |
|
|
}else{
|
95 |
|
|
Collection collection = collectionMap.get(collectionCode);
|
96 |
|
|
if (collection != null){
|
97 |
|
|
specimen.setCollection(collection);
|
98 |
|
|
}else{
|
99 |
|
|
logger.warn("Collection not found for code: " + collectionCode + "; titleCache: " + titleCache);
|
100 |
|
|
}
|
101 |
|
|
}
|
102 |
|
|
}
|
103 |
|
|
|
104 |
|
|
|
105 |
|
|
/**
|
106 |
|
|
* @param titleCache
|
107 |
|
|
* @return
|
108 |
|
|
*/
|
109 |
|
|
private String getCollectionCode(String titleCache) {
|
110 |
|
|
String result = titleCache.trim();
|
111 |
|
|
result = replaceBracket(result);
|
112 |
|
|
result = replaceLastFullStop(result);
|
113 |
|
|
result = replaceLastQuestionMark(result);
|
114 |
|
|
result = parseLastUpperCase(result);
|
115 |
|
|
return result;
|
116 |
|
|
}
|
117 |
|
|
|
118 |
|
|
|
119 |
|
|
/**
|
120 |
|
|
* @param result
|
121 |
|
|
* @return
|
122 |
|
|
*/
|
123 |
|
|
private String parseLastUpperCase(String string) {
|
124 |
|
|
String result = "";
|
125 |
|
|
String tmpString = string;
|
126 |
|
|
int pos = tmpString.lastIndexOf(" ");
|
127 |
|
|
if (pos>-1){
|
128 |
|
|
tmpString = tmpString.substring(pos+1);
|
129 |
|
|
}
|
130 |
|
|
while (tmpString.length() > 0){
|
131 |
|
|
int len = tmpString.length();
|
132 |
|
|
char lastChar = tmpString.charAt(len-1);
|
133 |
|
|
if (Character.isUpperCase( lastChar)){
|
134 |
|
|
result = lastChar + result;
|
135 |
|
|
}else{
|
136 |
|
|
if (result.length() > 0){
|
137 |
|
|
logger.warn("Collection code is not space separated: " + string);
|
138 |
|
|
}
|
139 |
|
|
break;
|
140 |
|
|
}
|
141 |
|
|
//remove last character
|
142 |
|
|
tmpString = tmpString.substring(0, tmpString.length()-1);
|
143 |
|
|
}
|
144 |
|
|
return result;
|
145 |
|
|
}
|
146 |
|
|
|
147 |
|
|
|
148 |
|
|
|
149 |
|
|
/**
|
150 |
|
|
* @param result
|
151 |
|
|
* @return
|
152 |
|
|
*/
|
153 |
|
|
private String replaceLastQuestionMark(String string) {
|
154 |
|
|
if (string.endsWith("?")){
|
155 |
|
|
string = string.substring(0,string.length()-1).trim();
|
156 |
|
|
}
|
157 |
|
|
return string;
|
158 |
|
|
}
|
159 |
|
|
|
160 |
|
|
/**
|
161 |
|
|
* @param result
|
162 |
|
|
* @return
|
163 |
|
|
*/
|
164 |
|
|
private String replaceLastFullStop(String string) {
|
165 |
|
|
if (string.endsWith(".")){
|
166 |
|
|
string = string.substring(0,string.length()-1).trim();
|
167 |
|
|
}
|
168 |
|
|
return string;
|
169 |
|
|
}
|
170 |
|
|
|
171 |
|
|
|
172 |
|
|
/**
|
173 |
|
|
* @param result
|
174 |
|
|
* @return
|
175 |
|
|
*/
|
176 |
|
|
private String replaceBracket(String string) {
|
177 |
|
|
if (string.endsWith("]")){
|
178 |
|
|
int pos = string.indexOf("[");
|
179 |
|
|
if (pos >0){
|
180 |
|
|
string = string.substring(0, pos).trim();
|
181 |
|
|
}else{
|
182 |
|
|
logger.warn("Closing bracket has no opening bracket in: " + string);
|
183 |
|
|
}
|
184 |
|
|
}
|
185 |
|
|
return string;
|
186 |
|
|
}
|
187 |
|
|
|
188 |
|
|
|
189 |
|
|
/**
|
190 |
|
|
* @param cdmApp
|
191 |
|
|
*/
|
192 |
|
|
private Map<String, Collection> createCollections(CdmApplicationController cdmApp) {
|
193 |
|
|
Map<String, Collection> collectionMap = new HashMap<String, Collection>();
|
194 |
|
|
List<String[]> lines = getLines();
|
195 |
|
|
for (String[] line:lines){
|
196 |
|
|
Collection collection = makeLine(line);
|
197 |
|
|
collectionMap.put(collection.getCode(), collection);
|
198 |
|
|
}
|
199 |
|
|
cdmApp.getCollectionService().save(collectionMap.values());
|
200 |
|
|
// for (Collection collection: collectionMap.values()){
|
201 |
|
|
// System.out.println(collection.getTitleCache());
|
202 |
|
|
// }
|
203 |
|
|
return collectionMap;
|
204 |
|
|
}
|
205 |
|
|
|
206 |
|
|
|
207 |
|
|
private Collection makeLine(String[] line) {
|
208 |
|
|
String code = line[0];
|
209 |
|
|
String instituteName = line[1];
|
210 |
|
|
String lowerInstitutionName = line[2];
|
211 |
|
|
String higherInstitutionName = line[3];
|
212 |
|
|
String location = line[4];
|
213 |
|
|
String country = line[5];
|
214 |
|
|
//create objects
|
215 |
|
|
Collection collection = Collection.NewInstance();
|
216 |
|
|
collection.setCode(code);
|
217 |
|
|
Institution institution = Institution.NewInstance();
|
218 |
|
|
institution.setCode(code);
|
219 |
|
|
|
220 |
|
|
institution.setName(instituteName);
|
221 |
|
|
|
222 |
|
|
if (StringUtils.isNotBlank(lowerInstitutionName)){
|
223 |
|
|
Institution lowerInstitution = Institution.NewInstance();
|
224 |
|
|
lowerInstitution.setName(lowerInstitutionName);
|
225 |
|
|
lowerInstitution.setIsPartOf(institution);
|
226 |
|
|
}
|
227 |
|
|
|
228 |
|
|
if (StringUtils.isNotBlank(higherInstitutionName)){
|
229 |
|
|
Institution higherInstitution = Institution.NewInstance();
|
230 |
|
|
higherInstitution.setName(higherInstitutionName);
|
231 |
|
|
institution.setIsPartOf(higherInstitution);
|
232 |
|
|
}
|
233 |
|
|
|
234 |
|
|
collection.setInstitute(institution);
|
235 |
|
|
String locationAndCountry = CdmUtils.concat("/", location, country);
|
236 |
|
|
collection.setTownOrLocation(locationAndCountry);
|
237 |
|
|
|
238 |
|
|
String titleCache = CdmUtils.concat(", ", new String[]{instituteName, lowerInstitutionName, higherInstitutionName, location, country});
|
239 |
|
|
collection.setTitleCache(titleCache, true);
|
240 |
|
|
|
241 |
|
|
return collection;
|
242 |
|
|
}
|
243 |
|
|
|
244 |
|
|
|
245 |
|
|
|
246 |
|
|
|
247 |
|
|
private List<String[]> getLines() {
|
248 |
|
|
List<String[]> result = new ArrayList<String[]>();
|
249 |
|
|
|
250 |
|
|
try {
|
251 |
|
|
InputStream inStream = new FileInputStream(acronymsFile);
|
252 |
|
|
InputStreamReader inputStreamReader = new InputStreamReader(inStream, "UTF8");
|
253 |
|
|
CSVReader reader = new CSVReader(inputStreamReader, '\t');
|
254 |
|
|
String [] nextLine = reader.readNext();
|
255 |
|
|
|
256 |
|
|
|
257 |
|
|
while ((nextLine = reader.readNext()) != null) {
|
258 |
|
|
if (nextLine.length == 0){
|
259 |
|
|
continue;
|
260 |
|
|
}
|
261 |
|
|
result.add(nextLine);
|
262 |
|
|
}
|
263 |
|
|
return result;
|
264 |
|
|
} catch (Exception e) {
|
265 |
|
|
logger.error(e + " " + e.getCause() + " " + e.getMessage());
|
266 |
|
|
for(StackTraceElement ste : e.getStackTrace()) {
|
267 |
|
|
logger.error(ste);
|
268 |
|
|
}
|
269 |
|
|
throw new RuntimeException(e);
|
270 |
|
|
}
|
271 |
|
|
}
|
272 |
|
|
|
273 |
|
|
|
274 |
|
|
|
275 |
|
|
|
276 |
|
|
|
277 |
|
|
/**
|
278 |
|
|
* @param args
|
279 |
|
|
*/
|
280 |
|
|
public static void main(String[] args) {
|
281 |
|
|
try {
|
282 |
|
|
DipteraCollectionImport collectionImport = new DipteraCollectionImport();
|
283 |
|
|
collectionImport.invoke(cdmDestination);
|
284 |
|
|
// String titleCache = "Peru. Mouth of Rio Pachitea. ST 2R SMT. [fig. of male abdomen]";
|
285 |
|
|
// String collectionCode = collectionImport.getCollectionCode(titleCache);
|
286 |
|
|
// System.out.println(collectionCode);
|
287 |
|
|
} catch (Exception e) {
|
288 |
|
|
e.printStackTrace();
|
289 |
|
|
System.exit(-1);
|
290 |
|
|
}
|
291 |
|
|
}
|
292 |
|
|
|
293 |
|
|
}
|