|
1
|
/**
|
|
2
|
* Copyright (C) 2007 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
|
|
|
10
|
package eu.etaxonomy.cdm.app.cyprus;
|
|
11
|
|
|
12
|
import java.io.File;
|
|
13
|
import java.io.IOException;
|
|
14
|
import java.net.MalformedURLException;
|
|
15
|
import java.net.URI;
|
|
16
|
import java.net.URISyntaxException;
|
|
17
|
import java.util.ArrayList;
|
|
18
|
import java.util.HashMap;
|
|
19
|
import java.util.HashSet;
|
|
20
|
import java.util.Iterator;
|
|
21
|
import java.util.List;
|
|
22
|
import java.util.Map;
|
|
23
|
import java.util.Set;
|
|
24
|
import java.util.regex.Matcher;
|
|
25
|
import java.util.regex.Pattern;
|
|
26
|
|
|
27
|
import org.apache.log4j.Logger;
|
|
28
|
import org.apache.sanselan.ImageReadException;
|
|
29
|
import org.apache.sanselan.Sanselan;
|
|
30
|
import org.apache.sanselan.common.IImageMetadata;
|
|
31
|
import org.apache.sanselan.common.ImageMetadata.Item;
|
|
32
|
import org.joda.time.DateTime;
|
|
33
|
import org.joda.time.format.DateTimeFormat;
|
|
34
|
import org.joda.time.format.DateTimeFormatter;
|
|
35
|
import org.springframework.transaction.TransactionStatus;
|
|
36
|
|
|
37
|
import eu.etaxonomy.cdm.api.application.CdmApplicationController;
|
|
38
|
import eu.etaxonomy.cdm.api.service.config.MatchingTaxonConfigurator;
|
|
39
|
import eu.etaxonomy.cdm.app.common.CdmDestinations;
|
|
40
|
import eu.etaxonomy.cdm.common.UTF8;
|
|
41
|
import eu.etaxonomy.cdm.common.media.ImageInfo;
|
|
42
|
import eu.etaxonomy.cdm.database.DbSchemaValidation;
|
|
43
|
import eu.etaxonomy.cdm.database.ICdmDataSource;
|
|
44
|
import eu.etaxonomy.cdm.io.api.application.CdmIoApplicationController;
|
|
45
|
import eu.etaxonomy.cdm.io.common.utils.ImportDeduplicationHelper;
|
|
46
|
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState;
|
|
47
|
import eu.etaxonomy.cdm.model.agent.AgentBase;
|
|
48
|
import eu.etaxonomy.cdm.model.agent.Institution;
|
|
49
|
import eu.etaxonomy.cdm.model.agent.Person;
|
|
50
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
|
51
|
import eu.etaxonomy.cdm.model.common.Language;
|
|
52
|
import eu.etaxonomy.cdm.model.description.DescriptionElementBase;
|
|
53
|
import eu.etaxonomy.cdm.model.description.Feature;
|
|
54
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
|
55
|
import eu.etaxonomy.cdm.model.description.TextData;
|
|
56
|
import eu.etaxonomy.cdm.model.media.ImageFile;
|
|
57
|
import eu.etaxonomy.cdm.model.media.Media;
|
|
58
|
import eu.etaxonomy.cdm.model.media.MediaRepresentation;
|
|
59
|
import eu.etaxonomy.cdm.model.media.MediaRepresentationPart;
|
|
60
|
import eu.etaxonomy.cdm.model.media.Rights;
|
|
61
|
import eu.etaxonomy.cdm.model.media.RightsType;
|
|
62
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
|
63
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
|
64
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
|
65
|
|
|
66
|
/**
|
|
67
|
* @author a.mueller
|
|
68
|
* @created 05.2017
|
|
69
|
*/
|
|
70
|
public class CyprusImagesActivator {
|
|
71
|
private static final Logger logger = Logger.getLogger(CyprusImagesActivator.class);
|
|
72
|
|
|
73
|
|
|
74
|
// static final ICdmDataSource cdmDestination = CdmDestinations.cdm_cyprus_dev();
|
|
75
|
static final ICdmDataSource cdmDestination = CdmDestinations.cdm_cyprus_production();
|
|
76
|
|
|
77
|
static boolean testOnly = true;
|
|
78
|
static boolean update_notCreate = true;
|
|
79
|
//if true, data will always be updated, if false, only missing data will be updated
|
|
80
|
static boolean forceUpdate = false;
|
|
81
|
|
|
82
|
private static final String path = "//media/digitalimages/EditWP6/Zypern/photos/";
|
|
83
|
private static final String urlPath = "http://media.bgbm.org/erez/erez?src=EditWP6/zypern/photos/";
|
|
84
|
|
|
85
|
private ImportDeduplicationHelper<SimpleExcelTaxonImportState<?>> deduplicationHelper;
|
|
86
|
|
|
87
|
|
|
88
|
private void doImport(ICdmDataSource cdmDestination){
|
|
89
|
|
|
90
|
CdmApplicationController app = CdmIoApplicationController.NewInstance(cdmDestination, DbSchemaValidation.VALIDATE);
|
|
91
|
TransactionStatus tx = app.startTransaction();
|
|
92
|
|
|
93
|
deduplicationHelper = (ImportDeduplicationHelper)ImportDeduplicationHelper.NewInstance(app);
|
|
94
|
|
|
95
|
File file = new File(path);
|
|
96
|
String[] fileList = file.list();
|
|
97
|
Set<String> notFound = new HashSet<>();
|
|
98
|
|
|
99
|
String regEx = "([A-Z][a-z]+_[a-z\\-]{3,}(?:_s_[a-z\\-]{3,})?)_[A-F]\\d{1,2}\\.(?:jpg|JPG)";
|
|
100
|
Pattern pattern = Pattern.compile(regEx);
|
|
101
|
|
|
102
|
for (String fileName : fileList){
|
|
103
|
|
|
104
|
Matcher matcher = pattern.matcher(fileName);
|
|
105
|
if (matcher.matches()){
|
|
106
|
// System.out.println(fileName);
|
|
107
|
String taxonName = matcher.group(1);
|
|
108
|
taxonName = taxonName.replace("_s_", " subsp. ").replace("_", " ");
|
|
109
|
Taxon taxon = getAcceptedTaxon(app, taxonName);
|
|
110
|
if (taxon == null){
|
|
111
|
if (!notFound.contains(taxonName)){
|
|
112
|
notFound.add(taxonName);
|
|
113
|
logger.warn("Taxon not found: " + taxonName);
|
|
114
|
}
|
|
115
|
}else{
|
|
116
|
handleTaxon(app, taxon, fileName);
|
|
117
|
}
|
|
118
|
}else{
|
|
119
|
if (!fileName.matches("(?:\\.erez|Thumbs\\.db.*|zypern_.*|__Keywords_template\\.txt)")){
|
|
120
|
logger.warn("Incorrect filename:" + fileName);
|
|
121
|
}
|
|
122
|
}
|
|
123
|
}
|
|
124
|
|
|
125
|
// app.getTaxonService().saveOrUpdate(taxaToSave);
|
|
126
|
|
|
127
|
if (testOnly){
|
|
128
|
tx.setRollbackOnly();
|
|
129
|
}
|
|
130
|
app.commitTransaction(tx);
|
|
131
|
}
|
|
132
|
|
|
133
|
/**
|
|
134
|
* @param app
|
|
135
|
* @param taxon
|
|
136
|
* @param fileName
|
|
137
|
*/
|
|
138
|
private void handleTaxon(CdmApplicationController app, Taxon taxon, String fileName) {
|
|
139
|
Set<String> urlStr = getAllExistingUrls(taxon);
|
|
140
|
String fullName = urlPath + fileName;
|
|
141
|
if (urlStr.contains(fullName)){
|
|
142
|
return;
|
|
143
|
}else{
|
|
144
|
addMedia(app, taxon, fileName);
|
|
145
|
}
|
|
146
|
}
|
|
147
|
|
|
148
|
/**
|
|
149
|
* @param app
|
|
150
|
* @param taxon
|
|
151
|
* @param fileName
|
|
152
|
*/
|
|
153
|
private void addMedia(CdmApplicationController app, Taxon taxon, String fileName) {
|
|
154
|
try {
|
|
155
|
String fullName = urlPath + fileName;
|
|
156
|
Media media = getImageMedia(fullName, null, true);
|
|
157
|
makeMetaData(media, fileName, false);
|
|
158
|
makeTitle(media, fileName, false);
|
|
159
|
if (!testOnly){
|
|
160
|
makeTextData(fileName, media, taxon);
|
|
161
|
}
|
|
162
|
|
|
163
|
} catch (Exception e) {
|
|
164
|
e.printStackTrace();
|
|
165
|
return;
|
|
166
|
}
|
|
167
|
}
|
|
168
|
|
|
169
|
/**
|
|
170
|
* @param media
|
|
171
|
* @param fileName
|
|
172
|
* @param b
|
|
173
|
*/
|
|
174
|
private void makeTitle(Media media, String fileName, boolean updateOnly) {
|
|
175
|
String title = fileName.replace("_s_"," subsp. ")
|
|
176
|
.replace("_"," ").replace(".jpg","").replace(".JPG","");
|
|
177
|
if ( (!updateOnly) || media.getAllTitles().isEmpty()){
|
|
178
|
media.putTitle(Language.LATIN(), title);
|
|
179
|
}
|
|
180
|
}
|
|
181
|
|
|
182
|
/**
|
|
183
|
* @param media
|
|
184
|
*/
|
|
185
|
private void makeMetaData(Media media, String fileName, boolean updateOnly) {
|
|
186
|
|
|
187
|
File file = new File(path + fileName);
|
|
188
|
if (!file.exists()){
|
|
189
|
logger.warn("File for filename " + fileName + " does not exist.");
|
|
190
|
return;
|
|
191
|
}
|
|
192
|
|
|
193
|
Map<String, String> keywords = new HashMap<>();
|
|
194
|
String copyright = null;
|
|
195
|
String artistStr = null;
|
|
196
|
String created = null;
|
|
197
|
try{
|
|
198
|
IImageMetadata metadata = Sanselan.getMetadata(file);
|
|
199
|
ArrayList<?> items = metadata.getItems();
|
|
200
|
for (Object object : items){
|
|
201
|
Item item = (Item) object;
|
|
202
|
// System.out.println(item.getKeyword() + ": " + item.getText());
|
|
203
|
String keyword = item.getKeyword().toLowerCase();
|
|
204
|
String value =removeQuots(item.getText());
|
|
205
|
|
|
206
|
if("keywords".equals(keyword)){
|
|
207
|
String[] splits = value.split(":");
|
|
208
|
if (splits.length == 2){
|
|
209
|
keywords.put(splits[0].trim().toLowerCase(), splits[1].trim());
|
|
210
|
}else{
|
|
211
|
logger.warn("Keyword has not correct format and can not be parsed: " + value + " for file " + fileName);
|
|
212
|
}
|
|
213
|
}else if ("Copyright Notice".equalsIgnoreCase(keyword)){
|
|
214
|
copyright = value;
|
|
215
|
}else if ("artist".equals(keyword)){
|
|
216
|
artistStr = value;
|
|
217
|
}else if ("date time original".equalsIgnoreCase(item.getKeyword())){
|
|
218
|
created = value;
|
|
219
|
}
|
|
220
|
}
|
|
221
|
} catch (ImageReadException | IOException e1) {
|
|
222
|
logger.warn(" Problem (" + e1.getMessage() + ") when reading metadata from file: " + fileName);
|
|
223
|
}
|
|
224
|
|
|
225
|
|
|
226
|
AgentBase<?> artistAgent = null;
|
|
227
|
Rights right = null;
|
|
228
|
DateTime createdDate = null;
|
|
229
|
String locality = null;
|
|
230
|
|
|
231
|
//artist
|
|
232
|
if (keywords.get("photographer") != null){
|
|
233
|
String artist = keywords.get("photographer");
|
|
234
|
artistAgent = makePerson(artist, fileName);
|
|
235
|
}
|
|
236
|
if (artistStr != null){
|
|
237
|
if (keywords.get("photographer") == null){
|
|
238
|
artistAgent = makePerson(artistStr, fileName);
|
|
239
|
}else if (!keywords.get("photographer").toLowerCase().replace(" ", "")
|
|
240
|
.contains(artistStr.toLowerCase().replace(" ", ""))){
|
|
241
|
logger.warn("Artist '" + artistStr + "' could not be handled for " + fileName);
|
|
242
|
}
|
|
243
|
}
|
|
244
|
|
|
245
|
//locality
|
|
246
|
if (keywords.get("locality") != null){
|
|
247
|
locality = keywords.get("locality");
|
|
248
|
}
|
|
249
|
|
|
250
|
//copyright
|
|
251
|
if (copyright != null){
|
|
252
|
AgentBase<?> agent;
|
|
253
|
if (copyright.equals("Botanic Garden and Botanical Museum Berlin-Dahlem (BGBM)")){
|
|
254
|
agent = Institution.NewNamedInstance(copyright);
|
|
255
|
}else{
|
|
256
|
agent = makePerson(copyright, fileName);
|
|
257
|
}
|
|
258
|
right = Rights.NewInstance(null, null, RightsType.COPYRIGHT());
|
|
259
|
right.setAgent(agent);
|
|
260
|
right = deduplicationHelper.getExistingCopyright(null, right);
|
|
261
|
}
|
|
262
|
|
|
263
|
//created
|
|
264
|
if (created != null){
|
|
265
|
DateTimeFormatter f = DateTimeFormat.forPattern("yyyy:MM:dd HH:mm:ss");
|
|
266
|
try {
|
|
267
|
createdDate = f/*.withZone(DateTimeZone.forID("Europe/Athens"))*/.parseDateTime(created);
|
|
268
|
} catch (Exception e) {
|
|
269
|
logger.warn("Exception (" + e.getMessage() + ") when parsing create date " + created + " for file " + fileName);
|
|
270
|
}
|
|
271
|
}
|
|
272
|
|
|
273
|
boolean force = !updateOnly || forceUpdate;
|
|
274
|
//add to media
|
|
275
|
if (artistAgent != null && (force || media.getArtist() == null)){
|
|
276
|
media.setArtist(artistAgent);
|
|
277
|
}
|
|
278
|
if (right != null && (force || media.getRights().isEmpty())){
|
|
279
|
media.removeRights(right);
|
|
280
|
media.addRights(right);
|
|
281
|
}
|
|
282
|
if (createdDate != null && (force || media.getMediaCreated() == null)){
|
|
283
|
media.setMediaCreated(createdDate);
|
|
284
|
}
|
|
285
|
if (locality != null && (force || media.getDescription(Language.ENGLISH()) == null)){
|
|
286
|
media.putDescription(Language.ENGLISH(), locality);
|
|
287
|
}
|
|
288
|
}
|
|
289
|
|
|
290
|
/**
|
|
291
|
* @param artist
|
|
292
|
* @return
|
|
293
|
*/
|
|
294
|
private Person makePerson(String artist, String fileName) {
|
|
295
|
artist = artist.trim();
|
|
296
|
String regEx = "((?:[A-Z]\\. ?)+)([A-Z][a-z\\-\u00E4\u00F6\u00FC]+)";
|
|
297
|
Matcher matcher = Pattern.compile(regEx).matcher(artist);
|
|
298
|
Person person = Person.NewInstance();
|
|
299
|
if (matcher.matches()){
|
|
300
|
person.setFirstname(matcher.group(1).trim());
|
|
301
|
person.setLastname(matcher.group(2).trim());
|
|
302
|
}else{
|
|
303
|
person.setTitleCache(artist, true);
|
|
304
|
logger.warn("Person could not be parsed: " + artist + " for file " + fileName);
|
|
305
|
}
|
|
306
|
|
|
307
|
person = (Person)deduplicationHelper.getExistingAuthor(null, person);
|
|
308
|
return person;
|
|
309
|
}
|
|
310
|
|
|
311
|
private String removeQuots(String text) {
|
|
312
|
if (text.startsWith("'") && text.endsWith("'")){
|
|
313
|
return text.substring(1, text.length() -1);
|
|
314
|
}else{
|
|
315
|
return text;
|
|
316
|
}
|
|
317
|
}
|
|
318
|
|
|
319
|
private void makeTextData(String fileStr, Media media, Taxon taxon) {
|
|
320
|
TaxonDescription imageGallery = taxon.getImageGallery(true);
|
|
321
|
TextData textData = null;
|
|
322
|
if (!imageGallery.getElements().isEmpty()){
|
|
323
|
DescriptionElementBase el = imageGallery.getElements().iterator().next();
|
|
324
|
if (el.isInstanceOf(TextData.class)){
|
|
325
|
textData = CdmBase.deproxy(el, TextData.class);
|
|
326
|
}else{
|
|
327
|
logger.warn("Image gallery had non-textdata description elmenet: " + fileStr);
|
|
328
|
}
|
|
329
|
}
|
|
330
|
if (textData == null){
|
|
331
|
textData = TextData.NewInstance();
|
|
332
|
textData.setFeature(Feature.IMAGE());
|
|
333
|
}
|
|
334
|
imageGallery.addElement(textData);
|
|
335
|
textData.addMedia(media);
|
|
336
|
}
|
|
337
|
|
|
338
|
/**
|
|
339
|
* Creates
|
|
340
|
* @param uriString
|
|
341
|
* @param readDataFromUrl
|
|
342
|
* @see #READ_MEDIA_DATA
|
|
343
|
* @return
|
|
344
|
* @throws MalformedURLException
|
|
345
|
*/
|
|
346
|
protected Media getImageMedia(String uriString, String uriStrThumb, boolean readMediaData) throws MalformedURLException {
|
|
347
|
if( uriString == null){
|
|
348
|
return null;
|
|
349
|
} else {
|
|
350
|
uriString = uriString.replace(" ", "%20"); //replace whitespace
|
|
351
|
try {
|
|
352
|
ImageInfo imageInfo = null;
|
|
353
|
URI uri = new URI(uriString);
|
|
354
|
|
|
355
|
try {
|
|
356
|
if (readMediaData){
|
|
357
|
logger.info("Read media data from: " + uri);
|
|
358
|
imageInfo = ImageInfo.NewInstance(uri, 0);
|
|
359
|
}
|
|
360
|
} catch (Exception e) {
|
|
361
|
String message = "An error occurred when trying to read image meta data for " + uri.toString() + ": " + e.getMessage();
|
|
362
|
logger.warn(message);
|
|
363
|
}
|
|
364
|
ImageFile imageFile = ImageFile.NewInstance(uri, null, imageInfo);
|
|
365
|
|
|
366
|
MediaRepresentation representation = MediaRepresentation.NewInstance();
|
|
367
|
|
|
368
|
if(imageInfo != null){
|
|
369
|
representation.setMimeType(imageInfo.getMimeType());
|
|
370
|
representation.setSuffix(imageInfo.getSuffix());
|
|
371
|
}
|
|
372
|
representation.addRepresentationPart(imageFile);
|
|
373
|
Media media = Media.NewInstance();
|
|
374
|
media.addRepresentation(representation);
|
|
375
|
|
|
376
|
if (uriStrThumb != null){
|
|
377
|
ImageInfo imageInfoThumb = null;
|
|
378
|
uriStrThumb = uriStrThumb.replace(" ", "%20"); //replace whitespace
|
|
379
|
URI uriThumb = new URI(uriStrThumb);
|
|
380
|
try {
|
|
381
|
if (readMediaData){
|
|
382
|
logger.info("Read media data from: " + uriThumb);
|
|
383
|
imageInfoThumb = ImageInfo.NewInstance(uriThumb, 0);
|
|
384
|
}
|
|
385
|
} catch (Exception e) {
|
|
386
|
String message = "An error occurred when trying to read image meta data for " + uriThumb.toString() + ": " + e.getMessage();
|
|
387
|
logger.warn(message);
|
|
388
|
}
|
|
389
|
|
|
390
|
ImageFile imageFileFhumb = ImageFile.NewInstance(uriThumb, null, imageInfoThumb);
|
|
391
|
MediaRepresentation reprThumb = MediaRepresentation.NewInstance();
|
|
392
|
if(imageInfoThumb != null){
|
|
393
|
reprThumb.setMimeType(imageInfoThumb.getMimeType());
|
|
394
|
reprThumb.setSuffix(imageInfoThumb.getSuffix());
|
|
395
|
}
|
|
396
|
reprThumb.addRepresentationPart(imageFileFhumb);
|
|
397
|
media.addRepresentation(reprThumb);
|
|
398
|
}
|
|
399
|
|
|
400
|
return media;
|
|
401
|
} catch (URISyntaxException e1) {
|
|
402
|
String message = "An URISyntaxException occurred when trying to create uri from multimedia objcet string: " + uriString;
|
|
403
|
logger.warn(message);
|
|
404
|
return null;
|
|
405
|
}
|
|
406
|
}
|
|
407
|
}
|
|
408
|
|
|
409
|
/**
|
|
410
|
* @param taxon
|
|
411
|
* @return
|
|
412
|
*/
|
|
413
|
private Set<String> getAllExistingUrls(Taxon taxon) {
|
|
414
|
Set<String> result = new HashSet<>();
|
|
415
|
Set<TaxonDescription> descriptions = taxon.getDescriptions();
|
|
416
|
for (TaxonDescription td : descriptions){
|
|
417
|
if (td.isImageGallery()){
|
|
418
|
for (DescriptionElementBase deb : td.getElements()){
|
|
419
|
if (deb.isInstanceOf(TextData.class)){
|
|
420
|
TextData textData = CdmBase.deproxy(deb, TextData.class);
|
|
421
|
for (Media media :textData.getMedia()){
|
|
422
|
for (MediaRepresentation rep : media.getRepresentations()){
|
|
423
|
for (MediaRepresentationPart part : rep.getParts()){
|
|
424
|
URI uri = part.getUri();
|
|
425
|
if (uri != null){
|
|
426
|
String uriStr = uri.toString();
|
|
427
|
result.add(uriStr);
|
|
428
|
}
|
|
429
|
}
|
|
430
|
}
|
|
431
|
}
|
|
432
|
|
|
433
|
}
|
|
434
|
}
|
|
435
|
}
|
|
436
|
}
|
|
437
|
return result;
|
|
438
|
}
|
|
439
|
|
|
440
|
private Taxon getAcceptedTaxon(CdmApplicationController app, String taxonNameStr) {
|
|
441
|
|
|
442
|
MatchingTaxonConfigurator config = new MatchingTaxonConfigurator();
|
|
443
|
taxonNameStr = adaptName(taxonNameStr);
|
|
444
|
config.setTaxonNameTitle(taxonNameStr);
|
|
445
|
config.setIncludeSynonyms(false);
|
|
446
|
List<TaxonBase> list = app.getTaxonService().findTaxaByName(config);
|
|
447
|
if (list.isEmpty()){
|
|
448
|
// logger.warn("Taxon not found for media: " + taxonNameStr);
|
|
449
|
taxonNameStr = taxonNameStr.replaceFirst(" ", " " + UTF8.HYBRID.toString());
|
|
450
|
config.setTaxonNameTitle(taxonNameStr);
|
|
451
|
list = app.getTaxonService().findTaxaByName(config);
|
|
452
|
if (list.isEmpty()){
|
|
453
|
return null;
|
|
454
|
}else if (list.size() > 1){
|
|
455
|
logger.warn("After searching for hybrids more than 1 taxon was foung: " + taxonNameStr);
|
|
456
|
}
|
|
457
|
}
|
|
458
|
if (list.size()>1){
|
|
459
|
Iterator<TaxonBase> it = list.iterator();
|
|
460
|
while (it.hasNext()){
|
|
461
|
Taxon next = (Taxon)it.next();
|
|
462
|
if (next.getTaxonNodes().isEmpty() && !next.getTaxonForMisappliedName().isEmpty()){
|
|
463
|
it.remove();
|
|
464
|
}
|
|
465
|
}
|
|
466
|
if (list.size()>1){
|
|
467
|
logger.warn("More than 1 taxon found for media: " + taxonNameStr + " . Will now try to use only taxon with taxon node.");
|
|
468
|
it = list.iterator();
|
|
469
|
while (it.hasNext()){
|
|
470
|
Taxon next = (Taxon)it.next();
|
|
471
|
if (next.getTaxonNodes().isEmpty()){
|
|
472
|
it.remove();
|
|
473
|
}
|
|
474
|
}
|
|
475
|
if (list.size()>1){
|
|
476
|
logger.warn("Still more than 1 taxon found for media: " + taxonNameStr);
|
|
477
|
}else if (list.size() < 1){
|
|
478
|
logger.warn("After removing nodeless taxa no taxon was left: " + taxonNameStr);
|
|
479
|
return null;
|
|
480
|
}
|
|
481
|
}else if (list.size() < 1){
|
|
482
|
logger.warn("After removing misapplications no taxon was left: " + taxonNameStr);
|
|
483
|
return null;
|
|
484
|
}
|
|
485
|
}
|
|
486
|
TaxonBase<?> taxonBase = list.get(0);
|
|
487
|
Taxon result;
|
|
488
|
if (taxonBase.isInstanceOf(Synonym.class)){
|
|
489
|
result = CdmBase.deproxy(taxonBase, Synonym.class).getAcceptedTaxon();
|
|
490
|
}else{
|
|
491
|
result = CdmBase.deproxy(taxonBase, Taxon.class);
|
|
492
|
}
|
|
493
|
return result;
|
|
494
|
}
|
|
495
|
|
|
496
|
/**
|
|
497
|
* @param taxonNameStr
|
|
498
|
* @return
|
|
499
|
*/
|
|
500
|
private String adaptName(String taxonNameStr) {
|
|
501
|
// if (taxonNameStr.equals("Hypericum cerastoides")){
|
|
502
|
// taxonNameStr = "Hypericum cerastioides";
|
|
503
|
// }
|
|
504
|
return taxonNameStr;
|
|
505
|
}
|
|
506
|
|
|
507
|
private void test(){
|
|
508
|
File f = new File(path);
|
|
509
|
String[] list = f.list();
|
|
510
|
List<String> fullFileNames = new ArrayList<>();
|
|
511
|
for (String fileName : list){
|
|
512
|
fullFileNames.add(path + fileName);
|
|
513
|
if (! fileName.matches("([A-Z][a-z]+_[a-z\\-]{3,}(?:_s_[a-z\\-]{3,})?)_[A-F]\\d{1,2}\\.(jpg|JPG)")){
|
|
514
|
System.out.println(fileName);
|
|
515
|
}
|
|
516
|
}
|
|
517
|
}
|
|
518
|
|
|
519
|
private void updateMetadata(ICdmDataSource cdmDestination){
|
|
520
|
CdmApplicationController app = CdmIoApplicationController.NewInstance(cdmDestination, DbSchemaValidation.VALIDATE);
|
|
521
|
TransactionStatus tx = app.startTransaction();
|
|
522
|
|
|
523
|
deduplicationHelper = (ImportDeduplicationHelper<SimpleExcelTaxonImportState<?>>)ImportDeduplicationHelper.NewInstance(app);
|
|
524
|
|
|
525
|
List<Media> list = app.getMediaService().list(Media.class, null, null, null, null);
|
|
526
|
for (Media media : list){
|
|
527
|
String fileName = getUrlStringForMedia(media);
|
|
528
|
if (fileName.startsWith(urlPath)){
|
|
529
|
fileName = fileName.replace(urlPath, "");
|
|
530
|
if (fileName.equals("Acinos_exiguus_C1.jpg")){ //for debugging only
|
|
531
|
// System.out.println(fileName);
|
|
532
|
makeMetaData(media, fileName, true);
|
|
533
|
makeTitle(media, fileName, true);
|
|
534
|
}
|
|
535
|
}else{
|
|
536
|
logger.warn("Filename does not start with standard url path: " + fileName);
|
|
537
|
}
|
|
538
|
}
|
|
539
|
|
|
540
|
if (testOnly){
|
|
541
|
tx.setRollbackOnly();
|
|
542
|
}
|
|
543
|
app.commitTransaction(tx);
|
|
544
|
|
|
545
|
}
|
|
546
|
|
|
547
|
/**
|
|
548
|
* @param media
|
|
549
|
* @return
|
|
550
|
*/
|
|
551
|
private String getUrlStringForMedia(Media media) {
|
|
552
|
String result = null;
|
|
553
|
for (MediaRepresentation rep : media.getRepresentations()){
|
|
554
|
for (MediaRepresentationPart part : rep.getParts()){
|
|
555
|
URI uri = part.getUri();
|
|
556
|
if (uri != null){
|
|
557
|
if (result != null){
|
|
558
|
logger.warn("More than 1 uri exists for media "+ media.getId());
|
|
559
|
}else{
|
|
560
|
result = uri.toString();
|
|
561
|
}
|
|
562
|
}
|
|
563
|
}
|
|
564
|
}
|
|
565
|
return result;
|
|
566
|
}
|
|
567
|
|
|
568
|
/**
|
|
569
|
* @param args
|
|
570
|
*/
|
|
571
|
public static void main(String[] args) {
|
|
572
|
CyprusImagesActivator me = new CyprusImagesActivator();
|
|
573
|
if (update_notCreate){
|
|
574
|
me.updateMetadata(cdmDestination);
|
|
575
|
}else{
|
|
576
|
me.doImport(cdmDestination);
|
|
577
|
}
|
|
578
|
// me.test();
|
|
579
|
System.exit(0);
|
|
580
|
}
|
|
581
|
|
|
582
|
}
|