1 |
91138b6c
|
Andreas Müller
|
/**
|
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.io;
|
11 |
|
|
|
12 |
|
|
import java.io.File;
|
13 |
|
|
import java.io.IOException;
|
14 |
|
|
import java.net.MalformedURLException;
|
15 |
|
|
import java.net.URL;
|
16 |
|
|
import java.util.ArrayList;
|
17 |
|
|
import java.util.HashMap;
|
18 |
|
|
import java.util.List;
|
19 |
|
|
import java.util.Map;
|
20 |
|
|
import java.util.Map.Entry;
|
21 |
c59ecc4a
|
Andreas Müller
|
import java.util.Set;
|
22 |
91138b6c
|
Andreas Müller
|
|
23 |
|
|
import org.apache.log4j.Logger;
|
24 |
|
|
import org.apache.sanselan.ImageInfo;
|
25 |
|
|
import org.apache.sanselan.ImageReadException;
|
26 |
|
|
import org.apache.sanselan.Sanselan;
|
27 |
|
|
import org.apache.sanselan.common.IImageMetadata;
|
28 |
|
|
import org.apache.sanselan.common.ImageMetadata.Item;
|
29 |
|
|
import org.apache.sanselan.formats.jpeg.JpegImageMetadata;
|
30 |
|
|
import org.springframework.stereotype.Component;
|
31 |
|
|
|
32 |
|
|
import eu.etaxonomy.cdm.app.images.AbstractImageImporter;
|
33 |
c59ecc4a
|
Andreas Müller
|
import eu.etaxonomy.cdm.app.images.ImageImportState;
|
34 |
91138b6c
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.agent.AgentBase;
|
35 |
|
|
import eu.etaxonomy.cdm.model.agent.Person;
|
36 |
|
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
37 |
|
|
import eu.etaxonomy.cdm.model.description.DescriptionElementBase;
|
38 |
|
|
import eu.etaxonomy.cdm.model.description.Feature;
|
39 |
|
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
40 |
|
|
import eu.etaxonomy.cdm.model.description.TextData;
|
41 |
|
|
import eu.etaxonomy.cdm.model.media.ImageFile;
|
42 |
|
|
import eu.etaxonomy.cdm.model.media.Media;
|
43 |
|
|
import eu.etaxonomy.cdm.model.media.MediaRepresentation;
|
44 |
|
|
import eu.etaxonomy.cdm.model.media.Rights;
|
45 |
9667d07d
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.media.RightsTerm;
|
46 |
401fe405
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
47 |
91138b6c
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
48 |
|
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
49 |
|
|
import eu.etaxonomy.cdm.strategy.match.DefaultMatchStrategy;
|
50 |
|
|
import eu.etaxonomy.cdm.strategy.match.IMatchStrategy;
|
51 |
|
|
|
52 |
|
|
/**
|
53 |
|
|
* TODO not working at the moment
|
54 |
|
|
*
|
55 |
|
|
* @author n.hoffmann
|
56 |
|
|
* @created 18.11.2008
|
57 |
|
|
* @version 1.0
|
58 |
|
|
*/
|
59 |
|
|
@Component
|
60 |
|
|
public class PalmaeImageImport extends AbstractImageImporter {
|
61 |
|
|
private static final Logger logger = Logger.getLogger(PalmaeImageImport.class);
|
62 |
|
|
|
63 |
|
|
enum MetaData{
|
64 |
|
|
NAME,
|
65 |
|
|
ARTIST,
|
66 |
|
|
COPYRIGHT,
|
67 |
|
|
COPYRIGHTNOTICE,
|
68 |
|
|
OBJECTNAME
|
69 |
|
|
}
|
70 |
|
|
|
71 |
|
|
private static int modCount = 300;
|
72 |
|
|
|
73 |
|
|
private static String pluralString = "images";
|
74 |
|
|
|
75 |
|
|
/**
|
76 |
|
|
* Rudimetary implementation using apache sanselan. This implementation depends
|
77 |
|
|
* on the metadata standards used in the palmae images. The IPTC field ObjectName
|
78 |
|
|
* contains a string like this: "Arecaceae; Eugeissona utilis". The string
|
79 |
|
|
* in front of the semicolon is the family name and the one behind, the taxon name.
|
80 |
|
|
* So we basically assume, that if the string gets split by ";" the element at
|
81 |
|
|
* index 1 should be the taxon name.
|
82 |
|
|
* If this format changes this method breaks!
|
83 |
|
|
*
|
84 |
|
|
* TODO The ImageMetaData class of the commons package should provide
|
85 |
|
|
* convenient access to the metadata of an image as well as all the error handling
|
86 |
|
|
*
|
87 |
|
|
* @param imageFile
|
88 |
|
|
* @return the name of the taxon as stored in ObjectName IPTC tag
|
89 |
|
|
*/
|
90 |
|
|
public String retrieveTaxonNameFromImageMetadata(File imageFile){
|
91 |
|
|
String name = null;
|
92 |
|
|
|
93 |
|
|
IImageMetadata metadata = null;
|
94 |
|
|
|
95 |
|
|
try {
|
96 |
|
|
metadata = Sanselan.getMetadata(imageFile);
|
97 |
|
|
} catch (ImageReadException e) {
|
98 |
|
|
logger.error("Error reading image" + " in " + imageFile.getName(), e);
|
99 |
|
|
} catch (IOException e) {
|
100 |
|
|
logger.error("Error reading file" + " in " + imageFile.getName(), e);
|
101 |
|
|
}
|
102 |
|
|
|
103 |
|
|
if(metadata instanceof JpegImageMetadata){
|
104 |
|
|
JpegImageMetadata jpegMetadata = (JpegImageMetadata) metadata;
|
105 |
|
|
|
106 |
|
|
for (Object object : jpegMetadata.getItems()){
|
107 |
|
|
|
108 |
|
|
Item item = (Item) object;
|
109 |
|
|
|
110 |
|
|
if(item.getKeyword().equals("ObjectName")){
|
111 |
|
|
logger.debug("File: " + imageFile.getName() + ". ObjectName string is: " + item.getText());
|
112 |
|
|
String[] objectNameSplit = item.getText().split(";");
|
113 |
|
|
|
114 |
|
|
try {
|
115 |
|
|
name = objectNameSplit[1].trim();
|
116 |
|
|
} catch (ArrayIndexOutOfBoundsException e) {
|
117 |
|
|
logger.warn("ObjectNameSplit has no second part: " + item.getText() + " in " + imageFile.getName());
|
118 |
|
|
//throw e;
|
119 |
|
|
}
|
120 |
|
|
}
|
121 |
|
|
}
|
122 |
|
|
}
|
123 |
|
|
|
124 |
|
|
|
125 |
|
|
return name;
|
126 |
|
|
}
|
127 |
|
|
|
128 |
|
|
public Map<MetaData, String> getMetaData(File imageFile, List<MetaData> metaData){
|
129 |
|
|
HashMap<MetaData, String> result = new HashMap<MetaData, String>();
|
130 |
|
|
|
131 |
|
|
IImageMetadata metadata = null;
|
132 |
|
|
List<String> metaDataStrings = new ArrayList<String>();
|
133 |
|
|
|
134 |
|
|
for (MetaData data: metaData){
|
135 |
|
|
metaDataStrings.add(data.name().toLowerCase());
|
136 |
|
|
}
|
137 |
|
|
|
138 |
|
|
|
139 |
|
|
try {
|
140 |
|
|
metadata = Sanselan.getMetadata(imageFile);
|
141 |
|
|
} catch (ImageReadException e) {
|
142 |
|
|
logger.error("Error reading image" + " in " + imageFile.getName(), e);
|
143 |
|
|
} catch (IOException e) {
|
144 |
|
|
logger.error("Error reading file" + " in " + imageFile.getName(), e);
|
145 |
|
|
}
|
146 |
|
|
|
147 |
|
|
|
148 |
|
|
|
149 |
|
|
if(metadata instanceof JpegImageMetadata){
|
150 |
|
|
JpegImageMetadata jpegMetadata = (JpegImageMetadata) metadata;
|
151 |
|
|
|
152 |
|
|
for (Object object : jpegMetadata.getItems()){
|
153 |
|
|
Item item = (Item) object;
|
154 |
|
|
|
155 |
|
|
if(metaDataStrings.contains(item.getKeyword().toLowerCase())){
|
156 |
|
|
logger.debug("File: " + imageFile.getName() + ". "+ item.getKeyword() +"string is: " + item.getText());
|
157 |
|
|
result.put(MetaData.valueOf(item.getKeyword().toUpperCase()), item.getText());
|
158 |
|
|
Set<Entry<MetaData, String>> resultSet = result.entrySet();
|
159 |
|
|
}
|
160 |
|
|
}
|
161 |
|
|
}
|
162 |
|
|
|
163 |
|
|
return result;
|
164 |
|
|
}
|
165 |
|
|
|
166 |
|
|
|
167 |
|
|
|
168 |
c59ecc4a
|
Andreas Müller
|
protected void invokeImageImport (ImageImportState state){
|
169 |
91138b6c
|
Andreas Müller
|
|
170 |
c59ecc4a
|
Andreas Müller
|
logger.info("Importing images from directory: " + state.getConfig().getSourceNameString());
|
171 |
91138b6c
|
Andreas Müller
|
|
172 |
c59ecc4a
|
Andreas Müller
|
File sourceFolder = new File(state.getConfig().getSource());
|
173 |
91138b6c
|
Andreas Müller
|
String taxonName;
|
174 |
|
|
if(sourceFolder.isDirectory()){
|
175 |
|
|
int count = 0;
|
176 |
|
|
for( File file : sourceFolder.listFiles()){
|
177 |
|
|
if(file.isFile()){
|
178 |
|
|
doCount(count++, modCount, pluralString);
|
179 |
|
|
|
180 |
|
|
taxonName= retrieveTaxonNameFromImageMetadata(file);
|
181 |
|
|
logger.debug("Looking up taxa with taxon name: " + taxonName);
|
182 |
|
|
|
183 |
|
|
//TODO:
|
184 |
|
|
ArrayList<MetaData> metaDataList = new ArrayList<MetaData>();
|
185 |
|
|
metaDataList.add (MetaData.ARTIST);
|
186 |
|
|
metaDataList.add (MetaData.COPYRIGHT);
|
187 |
|
|
metaDataList.add (MetaData.COPYRIGHTNOTICE);
|
188 |
|
|
metaDataList.add (MetaData.OBJECTNAME);
|
189 |
|
|
//metaDataList.add (MetaData.NAME);
|
190 |
|
|
|
191 |
|
|
Map<MetaData, String> metaData = getMetaData(file, metaDataList);
|
192 |
|
|
|
193 |
|
|
|
194 |
|
|
|
195 |
c59ecc4a
|
Andreas Müller
|
Reference sec = referenceService.find(state.getConfig().getSecUuid());
|
196 |
91138b6c
|
Andreas Müller
|
|
197 |
|
|
List<TaxonBase> taxa = new ArrayList<TaxonBase>();
|
198 |
|
|
if (taxonName != null){
|
199 |
|
|
taxa = taxonService.searchTaxaByName(taxonName, sec);
|
200 |
|
|
}else{
|
201 |
|
|
logger.error("TaxonName is null " + " in " + file.getName());
|
202 |
|
|
}
|
203 |
|
|
if(taxa.size() == 0){
|
204 |
|
|
logger.warn("no taxon with this name found: " + taxonName + " in " + file.getName());
|
205 |
|
|
}else if(taxa.size() > 1){
|
206 |
|
|
logger.error(taxa);
|
207 |
|
|
logger.error("multiple taxa with this name found: " + taxonName + " in " + file.getName());
|
208 |
|
|
}else{
|
209 |
|
|
Taxon taxon = (Taxon) taxa.get(0);
|
210 |
|
|
|
211 |
|
|
taxonService.saveOrUpdate(taxon);
|
212 |
|
|
|
213 |
|
|
//MetaDataFactory metaDataFactory = MetaDataFactory.getInstance();
|
214 |
|
|
//ImageMetaData imageMetaData = (ImageMetaData) metaDataFactory.readMediaData(file.toURI(), MimeType.IMAGE);
|
215 |
|
|
try{
|
216 |
|
|
ImageInfo imageinfo = Sanselan.getImageInfo(file);
|
217 |
|
|
|
218 |
|
|
String mimeType = imageinfo.getMimeType();
|
219 |
|
|
String suffix = "jpg";
|
220 |
|
|
|
221 |
|
|
|
222 |
|
|
// URL for this image
|
223 |
|
|
URL url = null;
|
224 |
|
|
try {
|
225 |
c59ecc4a
|
Andreas Müller
|
url = new URL(state.getConfig().getMediaUrlString() + file.getName());
|
226 |
91138b6c
|
Andreas Müller
|
} catch (MalformedURLException e) {
|
227 |
|
|
logger.warn("URL is malformed: "+ url);
|
228 |
|
|
}
|
229 |
|
|
|
230 |
|
|
|
231 |
401fe405
|
Andreas Müller
|
ImageFile imageFile = ImageFile.NewInstance(url.toURI(),null, imageinfo.getHeight(), imageinfo.getWidth());
|
232 |
91138b6c
|
Andreas Müller
|
|
233 |
|
|
|
234 |
|
|
MediaRepresentation representation = MediaRepresentation.NewInstance(mimeType, suffix);
|
235 |
|
|
representation.addRepresentationPart(imageFile);
|
236 |
|
|
|
237 |
|
|
Media media = Media.NewInstance();
|
238 |
|
|
media.addRepresentation(representation);
|
239 |
|
|
if (metaData.containsKey(MetaData.OBJECTNAME)){
|
240 |
|
|
media.setTitleCache(metaData.get(MetaData.OBJECTNAME).replace("'", ""), true);
|
241 |
|
|
}
|
242 |
|
|
//TODO: add the rights and the author:
|
243 |
|
|
Person artist = null;
|
244 |
|
|
if (metaData.containsKey(MetaData.ARTIST)){
|
245 |
|
|
//TODO search for the person first and then create the object...
|
246 |
|
|
artist = Person.NewTitledInstance(metaData.get(MetaData.ARTIST).replace("'", ""));
|
247 |
|
|
artist.setFirstname(getFirstName(metaData.get(MetaData.ARTIST)).replace("'", ""));
|
248 |
|
|
artist.setLastname(getLastName(metaData.get(MetaData.ARTIST)).replace("'", ""));
|
249 |
|
|
|
250 |
|
|
IMatchStrategy matchStrategy = DefaultMatchStrategy.NewInstance(AgentBase.class);
|
251 |
|
|
try{
|
252 |
|
|
List<Person> agents = commonService.findMatching(artist, matchStrategy);
|
253 |
|
|
|
254 |
|
|
if (agents.size()!= 0){
|
255 |
|
|
artist = agents.get(0);
|
256 |
|
|
}
|
257 |
|
|
}catch(eu.etaxonomy.cdm.strategy.match.MatchException e){
|
258 |
|
|
logger.warn("MatchException occurred");
|
259 |
|
|
}
|
260 |
|
|
|
261 |
|
|
media.setArtist(artist);
|
262 |
|
|
}
|
263 |
|
|
|
264 |
|
|
if (metaData.containsKey(MetaData.COPYRIGHT)){
|
265 |
|
|
//TODO: maybe search for the identic right...
|
266 |
|
|
Rights copyright = Rights.NewInstance();
|
267 |
9667d07d
|
Andreas Müller
|
copyright.setType(RightsTerm.COPYRIGHT());
|
268 |
91138b6c
|
Andreas Müller
|
Person copyrightOwner;
|
269 |
|
|
if (artist != null && !artist.getLastname().equalsIgnoreCase(getLastName(metaData.get(MetaData.COPYRIGHT)))){
|
270 |
|
|
copyrightOwner = Person.NewInstance();
|
271 |
|
|
|
272 |
|
|
copyrightOwner.setFirstname(getFirstName(metaData.get(MetaData.COPYRIGHT)));
|
273 |
|
|
copyrightOwner.setLastname(getLastName(metaData.get(MetaData.COPYRIGHT)));
|
274 |
|
|
}else
|
275 |
|
|
{
|
276 |
|
|
copyrightOwner = artist;
|
277 |
|
|
}
|
278 |
|
|
copyright.setAgent(copyrightOwner);
|
279 |
|
|
//IMatchStrategy matchStrategy = DefaultMatchStrategy.NewInstance(Rights.class);
|
280 |
|
|
media.addRights(copyright);
|
281 |
|
|
}
|
282 |
|
|
|
283 |
c59ecc4a
|
Andreas Müller
|
Reference sourceRef = state.getConfig().getSourceReference();
|
284 |
91138b6c
|
Andreas Müller
|
TaxonDescription description = taxon.getOrCreateImageGallery(sourceRef == null ? null :sourceRef.getTitleCache());
|
285 |
|
|
|
286 |
|
|
|
287 |
|
|
TextData textData = null;
|
288 |
|
|
for (DescriptionElementBase element : description.getElements()){
|
289 |
|
|
if (element.isInstanceOf(TextData.class)){
|
290 |
|
|
textData = CdmBase.deproxy(element, TextData.class);
|
291 |
|
|
}
|
292 |
|
|
}
|
293 |
|
|
if (textData == null){
|
294 |
|
|
textData = TextData.NewInstance();
|
295 |
|
|
}
|
296 |
|
|
|
297 |
|
|
|
298 |
|
|
textData.addMedia(media);
|
299 |
|
|
|
300 |
|
|
textData.setFeature(Feature.IMAGE());
|
301 |
|
|
|
302 |
|
|
description.addElement(textData);
|
303 |
|
|
|
304 |
|
|
taxonService.saveOrUpdate(taxon);
|
305 |
|
|
}catch(Exception e) {
|
306 |
|
|
e.printStackTrace();
|
307 |
|
|
}
|
308 |
|
|
}
|
309 |
|
|
}
|
310 |
|
|
}
|
311 |
|
|
}else{
|
312 |
|
|
logger.error("given source folder is not a directory");
|
313 |
|
|
}
|
314 |
c59ecc4a
|
Andreas Müller
|
return;
|
315 |
91138b6c
|
Andreas Müller
|
}
|
316 |
|
|
|
317 |
|
|
private String getFirstName(String artist){
|
318 |
|
|
if (artist == null){
|
319 |
|
|
return "";
|
320 |
|
|
}
|
321 |
|
|
if (!artist.contains(" ")) {
|
322 |
|
|
return "";
|
323 |
|
|
}
|
324 |
|
|
if (artist.contains(",")){
|
325 |
|
|
String [] artistSplits = artist.split(",");
|
326 |
|
|
artist = artistSplits[0];
|
327 |
|
|
|
328 |
|
|
}
|
329 |
|
|
|
330 |
|
|
try{
|
331 |
|
|
return artist.substring(0, artist.lastIndexOf(' ')).replace("'", "");
|
332 |
|
|
}catch (Exception e){
|
333 |
|
|
return "";
|
334 |
|
|
}
|
335 |
|
|
}
|
336 |
|
|
|
337 |
|
|
private String getLastName(String artist){
|
338 |
|
|
|
339 |
|
|
if (artist.contains(",")){
|
340 |
|
|
String [] artistSplits = artist.split(",");
|
341 |
|
|
artist = artistSplits[0];
|
342 |
|
|
|
343 |
|
|
}
|
344 |
|
|
if (!artist.contains(" ")) {
|
345 |
|
|
|
346 |
|
|
return artist;
|
347 |
|
|
}
|
348 |
|
|
try{
|
349 |
|
|
return artist.substring(artist.lastIndexOf(' ')).replace(" ", "");
|
350 |
|
|
}
|
351 |
|
|
catch(Exception e){
|
352 |
|
|
return "";
|
353 |
|
|
}
|
354 |
|
|
}
|
355 |
|
|
|
356 |
|
|
protected void doCount(int count, int modCount, String pluralString){
|
357 |
|
|
if ((count % modCount ) == 0 && count!= 0 ){ logger.info(pluralString + " handled: " + (count));}
|
358 |
|
|
}
|
359 |
|
|
|
360 |
|
|
}
|