Revision f2be28b3
Added by Andreas Müller over 11 years ago
app-import/src/main/java/eu/etaxonomy/cdm/io/globis/GlobisImageImport.java | ||
---|---|---|
18 | 18 |
import java.util.HashSet; |
19 | 19 |
import java.util.Map; |
20 | 20 |
import java.util.Set; |
21 |
import java.util.UUID; |
|
22 |
import java.util.regex.Matcher; |
|
23 |
import java.util.regex.Pattern; |
|
21 | 24 |
|
22 | 25 |
import org.apache.http.client.ClientProtocolException; |
23 | 26 |
import org.apache.log4j.Logger; |
... | ... | |
28 | 31 |
import eu.etaxonomy.cdm.io.common.IOValidator; |
29 | 32 |
import eu.etaxonomy.cdm.io.common.ResultSetPartitioner; |
30 | 33 |
import eu.etaxonomy.cdm.io.globis.validation.GlobisImageImportValidator; |
34 |
import eu.etaxonomy.cdm.model.common.Annotation; |
|
31 | 35 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
32 | 36 |
import eu.etaxonomy.cdm.model.common.Language; |
37 |
import eu.etaxonomy.cdm.model.common.Marker; |
|
38 |
import eu.etaxonomy.cdm.model.common.MarkerType; |
|
33 | 39 |
import eu.etaxonomy.cdm.model.description.DescriptionElementBase; |
34 |
import eu.etaxonomy.cdm.model.description.SpecimenDescription; |
|
40 |
import eu.etaxonomy.cdm.model.description.Feature; |
|
41 |
import eu.etaxonomy.cdm.model.description.TaxonDescription; |
|
35 | 42 |
import eu.etaxonomy.cdm.model.description.TextData; |
36 | 43 |
import eu.etaxonomy.cdm.model.media.Media; |
44 |
import eu.etaxonomy.cdm.model.name.ZoologicalName; |
|
45 |
import eu.etaxonomy.cdm.model.occurrence.Collection; |
|
46 |
import eu.etaxonomy.cdm.model.occurrence.DeterminationEvent; |
|
37 | 47 |
import eu.etaxonomy.cdm.model.occurrence.Specimen; |
38 | 48 |
import eu.etaxonomy.cdm.model.reference.Reference; |
49 |
import eu.etaxonomy.cdm.model.reference.ReferenceFactory; |
|
39 | 50 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
51 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase; |
|
40 | 52 |
|
41 | 53 |
|
42 | 54 |
/** |
... | ... | |
49 | 61 |
private static final Logger logger = Logger.getLogger(GlobisImageImport.class); |
50 | 62 |
|
51 | 63 |
private int modCount = 10000; |
64 |
|
|
65 |
private UUID uuidArtNonSpecTaxMarkerType = UUID.fromString("be362085-0f5b-4314-96d1-78b9b129ef6d") ; |
|
52 | 66 |
private static final String pluralString = "images"; |
53 | 67 |
private static final String dbTableName = "Einzelbilder"; |
54 |
private static final Class cdmTargetClass = Media |
|
55 |
.class; //not needed |
|
68 |
private static final Class cdmTargetClass = Media.class; //not needed |
|
56 | 69 |
|
57 | 70 |
private static final String IMAGE_NAMESPACE = "Einzelbilder"; |
71 |
private static UUID uuidGartRef = UUID.fromString("af85470f-6e54-4304-9d29-fd117cd56161"); |
|
58 | 72 |
|
59 | 73 |
public GlobisImageImport(){ |
60 | 74 |
super(pluralString, dbTableName, cdmTargetClass); |
... | ... | |
105 | 119 |
Map<String, Specimen> typeMap = (Map<String, Specimen>) partitioner.getObjectMap(GlobisSpecTaxImport.TYPE_NAMESPACE); |
106 | 120 |
|
107 | 121 |
Map<String, Taxon> taxonMap = (Map<String, Taxon>) partitioner.getObjectMap(TAXON_NAMESPACE); |
108 |
// Map<String, DerivedUnit> ecoFactDerivedUnitMap = (Map<String, DerivedUnit>) partitioner.getObjectMap(ECO_FACT_DERIVED_UNIT_NAMESPACE);
|
|
122 |
Map<String, ZoologicalName> specTaxNameMap = (Map<String, ZoologicalName>) partitioner.getObjectMap(GlobisSpecTaxImport.SPEC_TAX_NAMESPACE);
|
|
109 | 123 |
|
110 | 124 |
ResultSet rs = partitioner.getResultSet(); |
111 | 125 |
|
126 |
Reference<?> refGart = getReferenceService().find(uuidGartRef); |
|
127 |
|
|
128 |
|
|
112 | 129 |
try { |
113 | 130 |
|
114 | 131 |
int i = 0; |
115 | 132 |
|
116 |
//for each reference
|
|
133 |
//for each record
|
|
117 | 134 |
while (rs.next()){ |
118 | 135 |
|
119 | 136 |
if ((i++ % modCount) == 0 && i!= 1 ){ logger.info(pluralString + " handled: " + (i-1));} |
120 | 137 |
|
121 | 138 |
Integer bildID = rs.getInt("BildID"); |
122 |
|
|
123 | 139 |
Integer spectaxID = nullSafeInt(rs, "spectaxID"); |
124 |
|
|
125 | 140 |
Integer taxonID = nullSafeInt(rs, "SpecCurrspecID"); |
126 |
|
|
127 | 141 |
String copyright = rs.getString("copyright"); |
142 |
String specimenId = rs.getString("specimenID"); |
|
143 |
String bemerkungen = rs.getString("Bemerkungen"); |
|
144 |
String artNotSpecTax = rs.getString("Art non spectax"); |
|
145 |
String motiv = rs.getString("Motiv"); |
|
128 | 146 |
|
129 | 147 |
//ignore: |
130 | 148 |
// [file lab2], same as Dateiname04 but less data |
131 | 149 |
// Dateipfad |
150 |
|
|
132 | 151 |
Set<Media> recordMedia = new HashSet<Media>(); |
133 | 152 |
|
134 | 153 |
try { |
135 | 154 |
|
136 |
//make image path |
|
137 |
String pathShort = rs.getString("Dateipfad_kurz"); |
|
138 |
String fileOS = rs.getString("file OS"); |
|
139 |
pathShort= pathShort.replace(fileOS, ""); |
|
140 |
String newPath = state.getConfig().getImageBaseUrl(); |
|
141 |
String path = pathShort.replace("image:Webversionen/", newPath); |
|
142 |
|
|
143 |
Media singleMedia = makeMedia(state, rs, "file OS", "Legende 1", path, objectsToSave ); |
|
144 |
recordMedia.add(singleMedia); |
|
145 |
singleMedia = makeMedia(state, rs, "Dateinamen02", "Legende 2", path, objectsToSave ); |
|
146 |
recordMedia.add(singleMedia); |
|
147 |
singleMedia = makeMedia(state, rs, "Dateinamen03", "Legende 3", path, objectsToSave ); |
|
148 |
recordMedia.add(singleMedia); |
|
149 |
singleMedia = makeMedia(state, rs, "Dateinamen04", "Legende 4", path, objectsToSave ); |
|
150 |
recordMedia.add(singleMedia); |
|
151 |
|
|
152 |
if (spectaxID != null){ |
|
155 |
makeAllMedia(state, rs, recordMedia, objectsToSave); |
|
156 |
|
|
157 |
String title = null; |
|
158 |
|
|
159 |
Specimen specimen = null; |
|
160 |
if (spectaxID != null){ |
|
161 |
//try to find type specimen |
|
162 |
if (isNotBlank(motiv) && (motiv.startsWith("type specimen"))){ |
|
163 |
String collectionCode = transformCopyright2CollectionCode(copyright); |
|
164 |
String id = GlobisSpecTaxImport.getTypeId(spectaxID, collectionCode); |
|
165 |
specimen = typeMap.get(id); |
|
166 |
} |
|
167 |
|
|
168 |
//try to find specTaxName |
|
169 |
ZoologicalName specTaxTaxonName = specTaxNameMap.get(String.valueOf(spectaxID)); |
|
170 |
if (specTaxTaxonName != null){ |
|
171 |
title = " taxon name " + specTaxTaxonName.getTitleCache(); |
|
172 |
}else{ |
|
173 |
title = " spectaxID " + spectaxID; |
|
174 |
} |
|
175 |
}else{ |
|
176 |
title = " name " + getNameFromFileOs(rs) + (isBlank(specimenId)? "" : " (specimenId: " + specimenId + ")"); |
|
177 |
} |
|
178 |
|
|
179 |
//not type specimen |
|
180 |
if (specimen == null){ |
|
181 |
specimen = Specimen.NewInstance(); |
|
182 |
specimen.setTitleCache("Specimen for " + title ); |
|
153 | 183 |
String collectionCode = transformCopyright2CollectionCode(copyright); |
154 |
String id = GlobisSpecTaxImport.getTypeId(spectaxID, collectionCode); |
|
155 |
Specimen typeSpecimen = typeMap.get(id); |
|
156 |
if (typeSpecimen != null){ |
|
157 |
DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(typeSpecimen); |
|
158 |
for (Media media: recordMedia){ |
|
159 |
facade.addDerivedUnitMedia(media); |
|
160 |
} |
|
161 |
}else{ |
|
162 |
//TODO |
|
163 |
} |
|
164 |
}else{ |
|
165 | 184 |
//TODO |
185 |
Collection collection = getCollection(collectionCode); |
|
186 |
specimen.setCollection(collection); |
|
166 | 187 |
} |
167 | 188 |
|
168 | 189 |
|
190 |
//source |
|
191 |
specimen.addSource(String.valueOf(bildID), IMAGE_NAMESPACE, state.getTransactionalSourceReference(), null); |
|
169 | 192 |
|
193 |
//GART id (specimenID) |
|
194 |
if (isNotBlank(specimenId)){ |
|
195 |
specimen.addSource(specimenId, "", refGart, null); |
|
196 |
} |
|
197 |
//bemerkungen |
|
198 |
if (isNotBlank(bemerkungen)){ |
|
199 |
Annotation annotation = Annotation.NewInstance(bemerkungen, null, null); |
|
200 |
specimen.addAnnotation(annotation); |
|
201 |
} |
|
202 |
//media |
|
203 |
DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(specimen); |
|
204 |
for (Media media: recordMedia){ |
|
205 |
facade.addDerivedUnitMedia(media); |
|
206 |
} |
|
207 |
//art non spectax |
|
208 |
if (isNotBlank(artNotSpecTax)){ |
|
209 |
if (artNotSpecTax.equalsIgnoreCase("ja")){ |
|
210 |
MarkerType artNotSpecTaxMarker = getMarkerType(state, uuidArtNonSpecTaxMarkerType , "Art non spectax", "This marker is true if in the orginal data the 'Art non spectax' was 'ja'", null) ; |
|
211 |
specimen.addMarker(Marker.NewInstance(artNotSpecTaxMarker, true)); |
|
212 |
}else{ |
|
213 |
logger.warn(artNotSpecTax + " is not a valid value for 'Art non spectax' (BildID: " + bildID + ")" ); |
|
214 |
} |
|
215 |
} |
|
216 |
|
|
217 |
if (spectaxID != null){ |
|
218 |
|
|
219 |
//add to image gallery (discuss if this is also needed if taxon is already added to type specimen |
|
220 |
// Taxon taxon = taxonMap.get(String.valueOf(taxonID)); |
|
221 |
ZoologicalName specTaxTaxonName = specTaxNameMap.get(String.valueOf(spectaxID)); |
|
222 |
|
|
223 |
// |
|
224 |
// if (taxon == null){ |
|
225 |
//// taxon = specTaxMap.get(String.valueOf(spectaxID)); |
|
226 |
//// specTaxName = specTaxMap.g |
|
227 |
// } |
|
228 |
// if (taxon == null){ |
|
229 |
// logger.warn("No taxon available for specTaxID: " + spectaxID); |
|
230 |
// }else{ |
|
231 |
// name = CdmBase.deproxy(taxon.getName(), ZoologicalName.class); |
|
232 |
// } |
|
233 |
|
|
234 |
//TODO FIXME |
|
235 |
|
|
236 |
if (specTaxTaxonName == null){ |
|
237 |
logger.warn("Name could not be found for spectaxID: " + spectaxID + " in BildID: " + bildID); |
|
238 |
}else{ |
|
239 |
Taxon taxon = null; |
|
240 |
for (Taxon specTaxTaxon: specTaxTaxonName.getTaxa()){ |
|
241 |
taxon = specTaxTaxon; |
|
242 |
} |
|
243 |
if (taxon == null){ |
|
244 |
//FIXME |
|
245 |
Reference undefinedSec = null; |
|
246 |
taxon = Taxon.NewInstance(specTaxTaxonName, undefinedSec); |
|
247 |
} |
|
248 |
|
|
249 |
DeterminationEvent determinationEvent = DeterminationEvent.NewInstance(); |
|
250 |
determinationEvent.setTaxon(taxon); |
|
251 |
determinationEvent.setIdentifiedUnit(specimen); |
|
252 |
} |
|
253 |
|
|
254 |
|
|
255 |
|
|
256 |
|
|
257 |
// if (taxon != null){ |
|
258 |
// TaxonDescription taxonDescription = getTaxonDescription(taxon, true, true); |
|
259 |
// if (taxonDescription.getElements().size() == 0){ |
|
260 |
// TextData textData = TextData.NewInstance(Feature.IMAGE()); |
|
261 |
// taxonDescription.addElement(textData); |
|
262 |
// } |
|
263 |
// Set<DescriptionElementBase> elements = taxonDescription.getElements(); |
|
264 |
// TextData textData = CdmBase.deproxy(elements.iterator().next(), TextData.class); |
|
265 |
// for (Media media: recordMedia){ |
|
266 |
// textData.addMedia(media); |
|
267 |
// } |
|
268 |
// } |
|
269 |
} |
|
170 | 270 |
|
171 | 271 |
} catch (Exception e) { |
172 | 272 |
logger.warn("Exception in Einzelbilder: bildID " + bildID + ". " + e.getMessage()); |
... | ... | |
184 | 284 |
return false; |
185 | 285 |
} |
186 | 286 |
} |
287 |
|
|
288 |
private Collection getCollection(String collectionCode) { |
|
289 |
//TODO |
|
290 |
return null; |
|
291 |
} |
|
292 |
|
|
293 |
|
|
294 |
|
|
295 |
|
|
296 |
private String getNameFromFileOs(ResultSet rs) throws SQLException { |
|
297 |
String fileOS = rs.getString("file OS"); |
|
298 |
Pattern pattern = Pattern.compile("(.+)(_...._..\\.jpg)"); |
|
299 |
Matcher matcher = pattern.matcher(fileOS); |
|
300 |
if (matcher.matches()){ |
|
301 |
String match = matcher.group(1); |
|
302 |
return match; |
|
303 |
}else{ |
|
304 |
logger.warn("FileOS does not match: " + fileOS); |
|
305 |
return fileOS; |
|
306 |
} |
|
307 |
} |
|
308 |
|
|
309 |
|
|
310 |
|
|
311 |
|
|
312 |
private void makeAllMedia(GlobisImportState state, ResultSet rs, Set<Media> recordMedia, Set<Media> objectsToSave) throws SQLException{ |
|
313 |
//make image path |
|
314 |
String pathShort = rs.getString("Dateipfad_kurz"); |
|
315 |
String fileOS = rs.getString("file OS"); |
|
316 |
pathShort= pathShort.replace(fileOS, ""); |
|
317 |
String newPath = state.getConfig().getImageBaseUrl(); |
|
318 |
String path = pathShort.replace("image:Webversionen/", newPath); |
|
319 |
|
|
320 |
Media singleMedia = makeMedia(state, rs, "file OS", "Legende 1", path, objectsToSave ); |
|
321 |
recordMedia.add(singleMedia); |
|
322 |
singleMedia = makeMedia(state, rs, "Dateinamen02", "Legende 2", path, objectsToSave ); |
|
323 |
recordMedia.add(singleMedia); |
|
324 |
singleMedia = makeMedia(state, rs, "Dateinamen03", "Legende 3", path, objectsToSave ); |
|
325 |
recordMedia.add(singleMedia); |
|
326 |
singleMedia = makeMedia(state, rs, "Dateinamen04", "Legende 4", path, objectsToSave ); |
|
327 |
recordMedia.add(singleMedia); |
|
328 |
|
|
329 |
} |
|
187 | 330 |
|
188 | 331 |
private Media makeMedia(GlobisImportState state, ResultSet rs, String fileNameAttr, String legendAttr, String path, Set<Media> objectsToSave) throws SQLException { |
189 | 332 |
Media media = null; |
... | ... | |
191 | 334 |
String legend = rs.getString(legendAttr); |
192 | 335 |
Integer bildID = rs.getInt("BildID"); |
193 | 336 |
|
194 |
URI uri = URI.create(path+fileName); |
|
337 |
String uriStr = path+fileName; |
|
338 |
uriStr = uriStr.replace(" ", "%20"); |
|
339 |
|
|
340 |
URI uri = URI.create(uriStr); |
|
195 | 341 |
|
196 | 342 |
// Media media = ImageInfo.NewInstanceWithMetaData(uri, null); |
197 | 343 |
|
... | ... | |
283 | 429 |
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<Object, Map<String, ? extends CdmBase>>(); |
284 | 430 |
try{ |
285 | 431 |
Set<String> currSpecIdSet = new HashSet<String>(); |
432 |
Set<String> specTaxIdSet = new HashSet<String>(); |
|
286 | 433 |
Set<String> typeIdSet = new HashSet<String>(); |
287 | 434 |
|
288 | 435 |
while (rs.next()){ |
... | ... | |
290 | 437 |
handleTypeKey(rs, typeIdSet, "spectaxID", "copyright"); |
291 | 438 |
} |
292 | 439 |
|
440 |
//specTax map |
|
441 |
nameSpace = GlobisSpecTaxImport.SPEC_TAX_NAMESPACE; |
|
442 |
cdmClass = TaxonBase.class; |
|
443 |
idSet = specTaxIdSet; |
|
444 |
Map<String, TaxonBase> specTaxMap = (Map<String, TaxonBase>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace); |
|
445 |
result.put(nameSpace, specTaxMap); |
|
446 |
|
|
293 | 447 |
//taxon map |
294 | 448 |
nameSpace = TAXON_NAMESPACE; |
295 | 449 |
cdmClass = Taxon.class; |
... | ... | |
297 | 451 |
Map<String, Taxon> taxonMap = (Map<String, Taxon>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace); |
298 | 452 |
result.put(nameSpace, taxonMap); |
299 | 453 |
|
454 |
|
|
300 | 455 |
//type map |
301 | 456 |
nameSpace = GlobisSpecTaxImport.TYPE_NAMESPACE; |
302 | 457 |
cdmClass = Specimen.class; |
... | ... | |
343 | 498 |
|
344 | 499 |
|
345 | 500 |
|
501 |
/* (non-Javadoc) |
|
502 |
* @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#doInvoke(eu.etaxonomy.cdm.io.globis.GlobisImportState) |
|
503 |
*/ |
|
504 |
@Override |
|
505 |
protected void doInvoke(GlobisImportState state) { |
|
506 |
Reference refGart = ReferenceFactory.newGeneric(); |
|
507 |
refGart.setTitleCache("GART"); |
|
508 |
refGart.setUuid(uuidGartRef); |
|
509 |
getReferenceService().saveOrUpdate(refGart); |
|
510 |
super.doInvoke(state); |
|
511 |
} |
|
512 |
|
|
513 |
|
|
514 |
|
|
515 |
|
|
346 | 516 |
|
347 | 517 |
} |
Also available in: Unified diff
latest globis