Project

General

Profile

Download (6.99 KB) Statistics
| Branch: | Revision:
1 acdc7437 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.app.dwca;
11
12
import java.net.URI;
13
import java.util.UUID;
14
15
import org.apache.log4j.Logger;
16
17
import eu.etaxonomy.cdm.api.application.ICdmApplicationConfiguration;
18
import eu.etaxonomy.cdm.app.common.CdmDestinations;
19
import eu.etaxonomy.cdm.database.DbSchemaValidation;
20
import eu.etaxonomy.cdm.database.ICdmDataSource;
21
import eu.etaxonomy.cdm.io.common.CdmDefaultImport;
22
import eu.etaxonomy.cdm.io.common.IImportConfigurator.CHECK;
23
import eu.etaxonomy.cdm.io.common.events.LoggingIoObserver;
24
import eu.etaxonomy.cdm.io.dwca.in.DwcaImportConfigurator;
25 9638d8f7 Andreas Müller
import eu.etaxonomy.cdm.io.dwca.in.DwcaImportConfigurator.DatasetUse;
26 8d477b8a Andreas Müller
import eu.etaxonomy.cdm.io.dwca.in.IImportMapping.MappingType;
27 acdc7437 Andreas Müller
import eu.etaxonomy.cdm.model.agent.Person;
28 9638d8f7 Andreas Müller
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
29 acdc7437 Andreas Müller
import eu.etaxonomy.cdm.model.reference.Reference;
30
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
31
32
/**
33
 * @author a.mueller
34 7e96b3c1 Andreas Müller
 * @created 10.11.2011
35 acdc7437 Andreas Müller
 */
36
public class DwcaImportActivator {
37
	private static final Logger logger = Logger.getLogger(DwcaImportActivator.class);
38
	
39
	//database validation status (create, update, validate ...)
40
	static DbSchemaValidation hbm2dll = DbSchemaValidation.CREATE;
41 ba37ba74 Andreas Müller
//	static final URI source = dwca_test_in();
42 b6050fc2 Andreas Müller
//	static final URI source = dwca_test_cich_len();
43 7e96b3c1 Andreas Müller
	static final URI source = dwca_test_col_sapindaceae();
44
//	static final URI source = dwca_test_col_cichorium();
45 898b2185 Andreas Müller
//	static final URI source = dwca_test_scratch_test();
46
//	static final URI source = dwca_test_col_All_Pesi2();
47 8d477b8a Andreas Müller
//	static final URI source = dwca_test_col_All();
48 0d21320d Andreas Müller
//	static final URI source = dwca_test_col_All_Pesi2();
49 ba37ba74 Andreas Müller
	
50 acdc7437 Andreas Müller
	
51 898b2185 Andreas Müller
	static final ICdmDataSource cdmDestination = CdmDestinations.localH2();
52
//	static final ICdmDataSource cdmDestination = CdmDestinations.cdm_test_local_mysql_dwca();
53
//	static final ICdmDataSource cdmDestination = CdmDestinations.cdm_test_local_mysql_test();
54 acdc7437 Andreas Müller
//	static final ICdmDataSource cdmDestination = CdmDestinations.localH2();
55 0d21320d Andreas Müller
//	static final ICdmDataSource cdmDestination = CdmDestinations.cdm_test_local_mysql_pesi_test();
56 7e96b3c1 Andreas Müller
//	static final ICdmDataSource cdmDestination = CdmDestinations.cdm_test_local_mysql();
57 acdc7437 Andreas Müller
58
	
59
	//classification
60 c9978a69 Andreas Müller
	static final UUID classificationUuid = UUID.fromString("29d4011f-a6dd-4081-beb8-559ba6b84a6b");
61 acdc7437 Andreas Müller
	
62 9638d8f7 Andreas Müller
	//default nom code is ICZN as it allows adding publication year 
63
	static final NomenclaturalCode defaultNomCode = NomenclaturalCode.ICZN;
64
65
	
66 acdc7437 Andreas Müller
	//check - import
67
	static final CHECK check = CHECK.IMPORT_WITHOUT_CHECK;
68 7e96b3c1 Andreas Müller
	static int partitionSize = 5;
69 acdc7437 Andreas Müller
	
70 c9978a69 Andreas Müller
	//config
71 9638d8f7 Andreas Müller
	static DatasetUse datasetUse = DatasetUse.SECUNDUM;
72 c9978a69 Andreas Müller
	static boolean scientificNameIdAsOriginalSourceId = true;
73 0d21320d Andreas Müller
	static boolean guessNomRef = true;
74
	private boolean handleAllRefsAsCitation = true;
75 c9978a69 Andreas Müller
	
76
	//validate
77
	static boolean validateRankConsistency = false;
78
	
79
	
80 acdc7437 Andreas Müller
	//taxa
81
	static final boolean doTaxa = true;
82
	static final boolean doDistribution = false;
83 c9978a69 Andreas Müller
	//deduplicate
84
	static final boolean doDeduplicate = false;
85
	
86
	
87
	
88 8d477b8a Andreas Müller
	static final MappingType mappingType = MappingType.InMemoryMapping;
89 acdc7437 Andreas Müller
	
90
	private void doImport(ICdmDataSource cdmDestination){
91
		
92
		//make Source
93
		DwcaImportConfigurator config= DwcaImportConfigurator.NewInstance(source, cdmDestination);
94
		config.addObserver(new LoggingIoObserver());
95
		config.setClassificationUuid(classificationUuid);
96
		config.setCheck(check);
97
		config.setDbSchemaValidation(hbm2dll);
98 8d477b8a Andreas Müller
		config.setMappingType(mappingType);
99 0d21320d Andreas Müller
		
100 c9978a69 Andreas Müller
		config.setScientificNameIdAsOriginalSourceId(scientificNameIdAsOriginalSourceId);
101
		config.setValidateRankConsistency(validateRankConsistency);
102 7e96b3c1 Andreas Müller
		config.setDefaultPartitionSize(partitionSize);
103 9638d8f7 Andreas Müller
		config.setNomenclaturalCode(defaultNomCode);
104 0d21320d Andreas Müller
		config.setDatasetUse(datasetUse);
105
		config.setGuessNomenclaturalReferences(guessNomRef);
106
		config.setHandleAllRefsAsCitation(handleAllRefsAsCitation);
107 acdc7437 Andreas Müller
		
108
		CdmDefaultImport myImport = new CdmDefaultImport();
109
110
		
111
		//...
112
		if (true){
113
			System.out.println("Start import from ("+ source.toString() + ") ...");
114
			config.setSourceReference(getSourceReference(config.getSourceReferenceTitle()));
115
			myImport.invoke(config);
116
			System.out.println("End import from ("+ source.toString() + ")...");
117
		}
118
		
119
		
120
		
121
		//deduplicate
122
		if (doDeduplicate){
123
			ICdmApplicationConfiguration app = myImport.getCdmAppController();
124
			int count = app.getAgentService().deduplicate(Person.class, null, null);
125
			logger.warn("Deduplicated " + count + " persons.");
126
//			count = app.getAgentService().deduplicate(Team.class, null, null);
127
//			logger.warn("Deduplicated " + count + " teams.");
128
			count = app.getReferenceService().deduplicate(Reference.class, null, null);
129
			logger.warn("Deduplicated " + count + " references.");
130
		}
131
		
132
	}
133
134 9638d8f7 Andreas Müller
	private Reference<?> getSourceReference(String string) {
135
		Reference<?> result = ReferenceFactory.newGeneric();
136 acdc7437 Andreas Müller
		result.setTitleCache(string);
137
		return result;
138
	}
139
140
	//Dwca
141
	public static URI dwca_test_in() {
142
//		URI sourceUrl = URI.create("http://dev.e-taxonomy.eu/trac/export/14463/trunk/cdmlib/cdmlib-io/src/test/resources/eu/etaxonomy/cdm/io/dwca/in/DwcaZipToStreamConverterTest-input.zip");
143
		URI sourceUrl = URI.create("file:///C:/Users/pesiimport/Documents/pesi_cdmlib/cdmlib-io/src/test/resources/eu/etaxonomy/cdm/io/dwca/in/DwcaZipToStreamConverterTest-input.zip");
144
		return sourceUrl;
145
	}
146 ba37ba74 Andreas Müller
	
147
	
148
	//Dwca
149
	public static URI dwca_test_cich() {
150
		URI sourceUrl = URI.create("file:///E:/opt/data/dwca/20110621_1400_cichorieae_dwca.zip");
151
		return sourceUrl;
152
	}
153 80467b49 Andreas Müller
	
154
	//Dwca
155
	public static URI dwca_test_cich_len() {
156
		URI sourceUrl = URI.create("file:///C:/localCopy/Data/dwca/export/20110621_1400_cichorieae_dwca.zip");
157
		return sourceUrl;
158
	}
159
	
160 b6050fc2 Andreas Müller
	//Dwca
161
	public static URI dwca_test_col_cichorium() {
162
		URI sourceUrl = URI.create("file:///C:/localCopy/Data/dwca/import/CoL/Cichorium/archive-genus-Cichorium-bl3.zip");
163
		return sourceUrl;
164
	}
165 80467b49 Andreas Müller
	
166 7e96b3c1 Andreas Müller
	//Dwca
167
	public static URI dwca_test_col_sapindaceae() {
168
		URI sourceUrl = URI.create("file:///C:/localCopy/Data/dwca/import/CoL/Sapindaceae/archive-family-Sapindaceae-bl3.zip");
169
		return sourceUrl;
170
	}
171
	
172 b6050fc2 Andreas Müller
	//CoL
173
	public static URI dwca_test_col_All() {
174
		URI sourceUrl = URI.create("file:///C:/localCopy/Data/dwca/import/CoL/All/archive-complete.zip");
175
		return sourceUrl;
176
	}
177 acdc7437 Andreas Müller
178 8d477b8a Andreas Müller
	//CoL
179
	public static URI dwca_test_col_All_Pesi2() {
180
		URI sourceUrl = URI.create("file:///C:/opt/data/CoL/All/archive-complete.zip");
181
		return sourceUrl;
182
	}
183 acdc7437 Andreas Müller
	
184 898b2185 Andreas Müller
	//Scratchpads test
185
	public static URI dwca_test_scratch_test() {
186
		URI sourceUrl = URI.create("file:////PESIIMPORT3/vibrant/dwca/dwca_export_scratchpads_test.zip");
187
		return sourceUrl;
188
	}
189
	
190 acdc7437 Andreas Müller
191
	/**
192
	 * @param args
193
	 */
194
	public static void main(String[] args) {
195
		DwcaImportActivator me = new DwcaImportActivator();
196
		me.doImport(cdmDestination);
197
	}
198
	
199
}