Project

General

Profile

Download (8.62 KB) Statistics
| Branch: | Tag: | Revision:
1
/**
2
* Copyright (C) 2009 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
package eu.etaxonomy.cdm.io.dwca.in;
10

    
11
import java.net.URI;
12
import java.util.ArrayList;
13
import java.util.HashSet;
14
import java.util.List;
15
import java.util.Map;
16
import java.util.Set;
17

    
18
import org.apache.commons.lang.StringUtils;
19
import org.apache.log4j.Logger;
20

    
21
import com.ibm.lsid.MalformedLSIDException;
22

    
23
import eu.etaxonomy.cdm.io.dwca.TermUri;
24
import eu.etaxonomy.cdm.io.stream.StreamItem;
25
import eu.etaxonomy.cdm.model.agent.Team;
26
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
27
import eu.etaxonomy.cdm.model.common.CdmBase;
28
import eu.etaxonomy.cdm.model.common.LSID;
29
import eu.etaxonomy.cdm.model.common.TimePeriod;
30
import eu.etaxonomy.cdm.model.description.DescriptionElementSource;
31
import eu.etaxonomy.cdm.model.description.Feature;
32
import eu.etaxonomy.cdm.model.description.TaxonDescription;
33
import eu.etaxonomy.cdm.model.description.TextData;
34
import eu.etaxonomy.cdm.model.name.INonViralName;
35
import eu.etaxonomy.cdm.model.name.IZoologicalName;
36
import eu.etaxonomy.cdm.model.name.TaxonName;
37
import eu.etaxonomy.cdm.model.reference.Reference;
38
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
39
import eu.etaxonomy.cdm.model.taxon.Synonym;
40
import eu.etaxonomy.cdm.model.taxon.Taxon;
41
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
42
import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
43

    
44
/**
45
 * @author a.mueller
46
 * @date 22.11.2011
47
 *
48
 */
49
public class GbifReferenceCsv2CdmConverter extends PartitionableConverterBase<DwcaDataImportConfiguratorBase, DwcaDataImportStateBase<DwcaDataImportConfiguratorBase>>
50
						implements IPartitionableConverter<StreamItem, IReader<CdmBase>, String>{
51

    
52
	private static final Logger logger = Logger.getLogger(GbifReferenceCsv2CdmConverter.class);
53

    
54
	private static final String CORE_ID = "coreId";
55

    
56
	/**
57
	 * @param state
58
	 */
59
	public GbifReferenceCsv2CdmConverter(DwcaDataImportStateBase state) {
60
		super(state);
61
	}
62

    
63
	@Override
64
    public IReader<MappedCdmBase> map(StreamItem item ){
65
		List<MappedCdmBase> resultList = new ArrayList<MappedCdmBase>();
66

    
67
		Map<String, String> csv = item.map;
68
		Reference sourceReference = state.getTransactionalSourceReference();
69
		String sourceReferecenDetail = null;
70

    
71
		String id = getSourceId(item);
72
		TaxonBase<?> taxon = getTaxonBase(id, item, TaxonBase.class, state);
73
		if (isNotBlank(id) && taxon == null){
74
			String message = "Taxon for id %s could not be found";
75
			message = String.format(message, id);
76
			fireWarningEvent(message, item, 8);
77
		}
78

    
79
		String strCreator = getValue(item, TermUri.DC_CREATOR);
80
		String strDate = getValue(item, TermUri.DC_DATE);
81
		String strTitle = getValue(item, TermUri.DC_TITLE);
82
		String strSource = getValue(item, TermUri.DC_SOURCE);
83
		String strIdentifier = getValue(item, TermUri.DC_IDENTIFIER);
84
		String strType = getValue(item, TermUri.DC_TYPE);
85

    
86
		Reference reference = ReferenceFactory.newGeneric();
87
		resultList.add(new MappedCdmBase<CdmBase>(reference));
88

    
89
		//author
90
		TeamOrPersonBase<?> author = handleCreator(strCreator);
91
		reference.setAuthorship(author);
92
		//date
93
		TimePeriod publicationDate = handleDate(strDate);
94
		reference.setDatePublished(publicationDate);
95
		//title
96
		reference.setTitle(strTitle);
97
		//inreference
98
		Reference inRef = handleInRef(strSource);
99
		if (inRef != null){
100
			reference.setInReference(inRef);
101
			resultList.add(new MappedCdmBase<CdmBase>(inRef));
102
		}
103

    
104
		//URI
105
		handleIdentifier(strIdentifier, reference);
106

    
107
		//type
108
		handleType(reference, strType, taxon, resultList, item);
109

    
110

    
111
		return new ListReader<MappedCdmBase>(resultList);
112
	}
113

    
114

    
115
	private void handleType(Reference reference, String strType, TaxonBase<?> taxon, List<MappedCdmBase> resultList, StreamItem item) {
116
		// TODO handleType not yet implemented
117

    
118
		if (taxon == null){
119
			String message = "Taxon is null. Reference not imported.";
120
			fireWarningEvent(message,item, 4);
121
			//do nothing
122
		}else{
123
			boolean isNomRef = false;
124
			if (isNotBlank(strType)){
125
				if (strType.matches("Botanical Protologue")){
126
					if (taxon.getName() != null && reference != null && taxon.getName().isNonViral()){
127
						INonViralName nvn = taxon.getName();
128
						nvn.setNomenclaturalReference(reference);
129
						isNomRef = true;
130
					}else{
131
						//TODO
132
					}
133
				}
134
			}
135

    
136
			//guess a nom ref
137
			if (isNomRef == false && config.isGuessNomenclaturalReferences()){
138
				//if reference equals in author and year we assume that it is the nom ref
139
				//this information is usually only available for ICZN names
140
				if (taxon.getName() != null && reference != null && taxon.getName().isNonViral()){
141
					INonViralName nvn = taxon.getName();
142
					String taxonAuthor = nvn.getAuthorshipCache();
143
					String refAuthor = reference.getAuthorship().getNomenclaturalTitle();
144
					Integer combYear = null;
145
					Integer origYear = null;
146
					if (nvn.isZoological()){
147
						IZoologicalName zooName = (IZoologicalName)CdmBase.deproxy(nvn);
148
						combYear = zooName.getPublicationYear();
149
						origYear = zooName.getOriginalPublicationYear();
150
					}
151
					String refYear = reference.getYear();
152

    
153
					//combination compare
154
					if (taxonAuthor != null && taxonAuthor.equals(refAuthor)){
155
						if (combYear != null && String.valueOf(combYear).equals(refYear)){
156
							//is nom Ref
157
							isNomRef = true;
158
							nvn.setNomenclaturalReference(reference);
159
						}else if (origYear != null && String.valueOf(origYear).equals(refYear)){
160
							//TODO not yet handled by CDM
161
						}
162
					}
163

    
164
				}
165
			}
166
			if (config.isHandleAllRefsAsCitation()){
167
				if (taxon.isInstanceOf(Taxon.class)){
168
					TaxonDescription desc = getTaxonDescription(CdmBase.deproxy(taxon, Taxon.class), false);
169
					createCitation(desc, reference, taxon.getName());
170
					resultList.add(new MappedCdmBase<CdmBase>(desc));
171
				}else if (taxon.isInstanceOf(Synonym.class)){
172
					Synonym syn = CdmBase.deproxy(taxon, Synonym.class);
173
					Taxon tax = syn.getAcceptedTaxon();
174
					if (tax != null){
175
    					TaxonDescription desc = getTaxonDescription(tax, false);
176
    					createCitation(desc, reference, syn.getName());
177
    					resultList.add(new MappedCdmBase<CdmBase>(desc));
178
					}
179
				}
180

    
181
			}
182

    
183
		}
184

    
185

    
186
	}
187

    
188
	private void createCitation(TaxonDescription desc, Reference ref, TaxonName nameUsedInSource) {
189
		Feature feature = Feature.CITATION();
190
		TextData textData = TextData.NewInstance(feature);
191
		DescriptionElementSource source = DescriptionElementSource.NewPrimarySourceInstance(ref, null, nameUsedInSource, null);
192
		textData.addSource(source);
193
		desc.addElement(textData);
194
	}
195

    
196
	private void handleIdentifier(String strIdentifier, Reference reference) {
197
		if (StringUtils.isBlank(strIdentifier)){
198
			return;
199
		}else if (LSID.isLsid(strIdentifier)){
200
			LSID lsid;
201
			try {
202
				lsid = new LSID(strIdentifier);
203
				reference.setLsid(lsid);
204
			} catch (MalformedLSIDException e) {
205
				//TODO should not happen as we have checked before
206
				throw new RuntimeException(e);
207
			}
208
		}
209
		try {
210
			URI uri = URI.create(strIdentifier);
211
			reference.setUri(uri);
212
		} catch (Exception e) {
213
			logger.debug("Reference is not an URI");
214
		}
215
		//TODO further identifier types
216

    
217
	}
218

    
219
	private Reference handleInRef(String strSource) {
220
		if (StringUtils.isBlank(strSource)){
221
			return null;
222
		}else{
223
			Reference inRef = ReferenceFactory.newGeneric();
224
			return inRef;
225
		}
226
	}
227

    
228

    
229
	private TimePeriod handleDate(String strDate) {
230
		TimePeriod tp = TimePeriodParser.parseString(strDate);
231
		return tp;
232
	}
233

    
234
	private TeamOrPersonBase handleCreator(String strCreator) {
235
		Team team = Team.NewTitledInstance(strCreator, strCreator);
236
		return team;
237
	}
238

    
239
	@Override
240
	public String getSourceId(StreamItem item) {
241
		String id = item.get(CORE_ID);
242
		return id;
243
	}
244

    
245

    
246
//********************** PARTITIONABLE **************************************/
247

    
248
	@Override
249
	protected void makeForeignKeysForItem(StreamItem item, Map<String, Set<String>> fkMap) {
250
		String value;
251
		String key;
252
		if ( hasValue(value = item.get(CORE_ID))){
253
			key = TermUri.DWC_TAXON.toString();
254
			Set<String> keySet = getKeySet(key, fkMap);
255
			keySet.add(value);
256
		}
257
	}
258

    
259

    
260
	@Override
261
	public Set<String> requiredSourceNamespaces() {
262
		Set<String> result = new HashSet<String>();
263
 		result.add(TermUri.DWC_TAXON.toString());
264
 		return result;
265
	}
266

    
267
//******************* TO STRING ******************************************/
268

    
269
	@Override
270
	public String toString(){
271
		return this.getClass().getName();
272
	}
273

    
274

    
275
}
(19-19/37)