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
|
package eu.etaxonomy.cdm.io.pilotOutputHtml;
|
10
|
|
11
|
import java.io.File;
|
12
|
import java.io.FileOutputStream;
|
13
|
import java.io.OutputStreamWriter;
|
14
|
import java.io.PrintWriter;
|
15
|
import java.util.List;
|
16
|
|
17
|
import org.apache.logging.log4j.LogManager;import org.apache.logging.log4j.Logger;
|
18
|
import org.springframework.stereotype.Component;
|
19
|
import org.springframework.transaction.TransactionStatus;
|
20
|
|
21
|
import eu.etaxonomy.cdm.io.common.CdmExportBase;
|
22
|
import eu.etaxonomy.cdm.io.common.ICdmExport;
|
23
|
import eu.etaxonomy.cdm.io.common.IExportConfigurator;
|
24
|
import eu.etaxonomy.cdm.io.common.mapping.out.IExportTransformer;
|
25
|
import eu.etaxonomy.cdm.io.sdd.out.SDDDataSet;
|
26
|
import eu.etaxonomy.cdm.model.agent.AgentBase;
|
27
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
28
|
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationBase;
|
29
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
30
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
31
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
32
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
33
|
import eu.etaxonomy.cdm.model.term.DefinedTermBase;
|
34
|
import eu.etaxonomy.cdm.model.term.TermType;
|
35
|
|
36
|
/**
|
37
|
* @author h.fradin (from a.babadshanjan)
|
38
|
* @since 10.12.2008
|
39
|
*/
|
40
|
@Component("pilotOutputExporter")
|
41
|
public class PilotOutputExporter
|
42
|
extends CdmExportBase<PilotOutputExportConfigurator, PilotOutputExportState, IExportTransformer, File>
|
43
|
implements ICdmExport<PilotOutputExportConfigurator, PilotOutputExportState> {
|
44
|
// public class JaxbExport extends CdmIoBase implements ICdmIoExport {
|
45
|
// TODO: public class JaxbExport extends CdmIoBase implements ICdmIO {
|
46
|
|
47
|
private static final long serialVersionUID = -5556376038773540067L;
|
48
|
private static final Logger logger = LogManager.getLogger(PilotOutputExporter.class);
|
49
|
private PilotOutputDocumentBuilder pilotOutputDocumentBuilder = null;
|
50
|
|
51
|
private boolean doOccurrence = true;
|
52
|
|
53
|
|
54
|
public PilotOutputExporter() {
|
55
|
super();
|
56
|
this.ioName = this.getClass().getSimpleName();
|
57
|
}
|
58
|
|
59
|
/** Retrieves data from a CDM DB and serializes them CDM to XML.
|
60
|
* Starts with root taxa and traverses the classification to retrieve children taxa, synonyms and relationships.
|
61
|
* Taxa that are not part of the classification are not found.
|
62
|
*
|
63
|
* @param exImpConfig
|
64
|
* @param dbname
|
65
|
* @param filename
|
66
|
*/
|
67
|
@Override
|
68
|
protected void doInvoke(PilotOutputExportState state){
|
69
|
// protected boolean doInvoke(IExportConfigurator config,
|
70
|
// Map<String, MapWrapper<? extends CdmBase>> stores) {
|
71
|
|
72
|
PilotOutputExportConfigurator pilotOutputExpConfig = state.getConfig();
|
73
|
String dbname = pilotOutputExpConfig.getSource().getName();
|
74
|
String fileName = pilotOutputExpConfig.getDestinationNameString();
|
75
|
logger.info("Serializing DB " + dbname + " to file " + fileName);
|
76
|
logger.debug("DbSchemaValidation = " + pilotOutputExpConfig.getDbSchemaValidation());
|
77
|
|
78
|
TransactionStatus txStatus = startTransaction(true);
|
79
|
SDDDataSet dataSet = new SDDDataSet();
|
80
|
List<Taxon> taxa = null;
|
81
|
List<DefinedTermBase> terms = null;
|
82
|
|
83
|
// get data from DB
|
84
|
|
85
|
try {
|
86
|
logger.info("Retrieving data from DB");
|
87
|
|
88
|
retrieveData(pilotOutputExpConfig, dataSet);
|
89
|
|
90
|
} catch (Exception e) {
|
91
|
logger.error("Error retrieving data");
|
92
|
e.printStackTrace();
|
93
|
}
|
94
|
|
95
|
logger.info("All data retrieved");
|
96
|
|
97
|
try {
|
98
|
pilotOutputDocumentBuilder = new PilotOutputDocumentBuilder();
|
99
|
File f = new File(fileName);
|
100
|
FileOutputStream fos = new FileOutputStream(f);
|
101
|
PrintWriter writer = new PrintWriter(new OutputStreamWriter(fos, "UTF8"), true);
|
102
|
pilotOutputDocumentBuilder.marshal(dataSet, fileName);
|
103
|
|
104
|
// TODO: Split into one file per data set member to see whether performance improves?
|
105
|
|
106
|
logger.info("XML file written");
|
107
|
logger.info("Filename is: " + fileName);
|
108
|
|
109
|
} catch (Exception e) {
|
110
|
logger.error("Marshalling error");
|
111
|
e.printStackTrace();
|
112
|
}
|
113
|
commitTransaction(txStatus);
|
114
|
return;
|
115
|
}
|
116
|
|
117
|
private void retrieveData (IExportConfigurator config, SDDDataSet sddDataSet) {
|
118
|
|
119
|
PilotOutputExportConfigurator pilotOutputExpConfig = (PilotOutputExportConfigurator)config;
|
120
|
final int MAX_ROWS = 50000;
|
121
|
int numberOfRows = pilotOutputExpConfig.getMaxRows();
|
122
|
// TODO:
|
123
|
//CdmApplicationController appCtr = config.getCdmAppController(false, true);
|
124
|
|
125
|
int agentRows = numberOfRows;
|
126
|
int definedTermBaseRows = numberOfRows;
|
127
|
int referenceBaseRows = numberOfRows;
|
128
|
int taxonNameRows = numberOfRows;
|
129
|
int taxonBaseRows = numberOfRows;
|
130
|
int relationshipRows = numberOfRows;
|
131
|
int occurrencesRows = numberOfRows;
|
132
|
int mediaRows = numberOfRows;
|
133
|
int featureDataRows = numberOfRows;
|
134
|
int languageDataRows = numberOfRows;
|
135
|
int termVocabularyRows = numberOfRows;
|
136
|
int homotypicalGroupRows = numberOfRows;
|
137
|
|
138
|
if (pilotOutputExpConfig.isDoTermVocabularies() == true) {
|
139
|
if (termVocabularyRows == 0) { termVocabularyRows = MAX_ROWS; }
|
140
|
logger.info("# TermVocabulary");
|
141
|
sddDataSet.setTermVocabularies((List)getVocabularyService().list(null,MAX_ROWS, 0,null,null));;
|
142
|
}
|
143
|
|
144
|
if (pilotOutputExpConfig.isDoLanguageData() == true) {
|
145
|
if (languageDataRows == 0) { languageDataRows = MAX_ROWS; }
|
146
|
logger.info("# Representation, Language String");
|
147
|
sddDataSet.setLanguageData(getTermService().getAllRepresentations(MAX_ROWS, 0));
|
148
|
sddDataSet.addLanguageData(getTermService().getAllLanguageStrings(MAX_ROWS, 0));
|
149
|
}
|
150
|
|
151
|
if (pilotOutputExpConfig.isDoTerms() == true) {
|
152
|
if (definedTermBaseRows == 0) { definedTermBaseRows = getTermService().count(DefinedTermBase.class); }
|
153
|
logger.info("# DefinedTermBase: " + definedTermBaseRows);
|
154
|
sddDataSet.setTerms(getTermService().list(null,definedTermBaseRows, 0,null,null));
|
155
|
}
|
156
|
|
157
|
if (pilotOutputExpConfig.isDoAuthors() == true) {
|
158
|
if (agentRows == 0) { agentRows = getAgentService().count(AgentBase.class); }
|
159
|
logger.info("# Agents: " + agentRows);
|
160
|
//logger.info(" # Team: " + getAgentService().count(Team.class));
|
161
|
sddDataSet.setAgents(getAgentService().list(null,agentRows, 0,null,null));
|
162
|
}
|
163
|
|
164
|
if (pilotOutputExpConfig.getDoReferences() != IExportConfigurator.DO_REFERENCES.NONE) {
|
165
|
if (referenceBaseRows == 0) { referenceBaseRows = getReferenceService().count(Reference.class); }
|
166
|
logger.info("# Reference: " + referenceBaseRows);
|
167
|
sddDataSet.setReferences(getReferenceService().list(null,referenceBaseRows, 0,null,null));
|
168
|
}
|
169
|
|
170
|
if (pilotOutputExpConfig.isDoTaxonNames() == true) {
|
171
|
if (taxonNameRows == 0) { taxonNameRows = getNameService().count(TaxonName.class); }
|
172
|
logger.info("# TaxonName: " + taxonNameRows);
|
173
|
//logger.info(" # Taxon: " + getNameService().count(BotanicalName.class));
|
174
|
sddDataSet.setTaxonomicNames(getNameService().list(null,taxonNameRows, 0,null,null));
|
175
|
}
|
176
|
|
177
|
if (pilotOutputExpConfig.isDoHomotypicalGroups() == true) {
|
178
|
if (homotypicalGroupRows == 0) { homotypicalGroupRows = MAX_ROWS; }
|
179
|
logger.info("# Homotypical Groups");
|
180
|
sddDataSet.setHomotypicalGroups(getNameService().getAllHomotypicalGroups(homotypicalGroupRows, 0));
|
181
|
}
|
182
|
|
183
|
if (pilotOutputExpConfig.isDoTaxa() == true) {
|
184
|
if (taxonBaseRows == 0) { taxonBaseRows = getTaxonService().count(TaxonBase.class); }
|
185
|
logger.info("# TaxonBase: " + taxonBaseRows);
|
186
|
// dataSet.setTaxa(new ArrayList<Taxon>());
|
187
|
// dataSet.setSynonyms(new ArrayList<Synonym>());
|
188
|
List<TaxonBase> tb = getTaxonService().list(null,taxonBaseRows, 0,null,null);
|
189
|
for (TaxonBase taxonBase : tb) {
|
190
|
if (taxonBase instanceof Taxon) {
|
191
|
sddDataSet.addTaxon((Taxon)taxonBase);
|
192
|
} else if (taxonBase instanceof Synonym) {
|
193
|
sddDataSet.addSynonym((Synonym)taxonBase);
|
194
|
} else {
|
195
|
logger.error("entry of wrong type: " + taxonBase.toString());
|
196
|
}
|
197
|
}
|
198
|
}
|
199
|
|
200
|
// TODO:
|
201
|
// retrieve taxa and synonyms separately
|
202
|
// need correct count for taxa and synonyms
|
203
|
// if (taxonBaseRows == 0) { taxonBaseRows = getTaxonService().count(TaxonBase.class); }
|
204
|
// logger.info("# Synonym: " + taxonBaseRows);
|
205
|
// dataSet.setSynonyms(new ArrayList<Synonym>());
|
206
|
// dataSet.setSynonyms(getTaxonService().getAllSynonyms(taxonBaseRows, 0));
|
207
|
//
|
208
|
// if (pilotOutputExpConfig.isDoRelTaxa() == true) {
|
209
|
// if (relationshipRows == 0) { relationshipRows = MAX_ROWS; }
|
210
|
// logger.info("# Relationships");
|
211
|
// List<RelationshipBase> relationList = getTaxonService().getAllRelationships(relationshipRows, 0);
|
212
|
// Set<RelationshipBase> relationSet = new HashSet<RelationshipBase>(relationList);
|
213
|
// sddDataSet.setRelationships(relationSet);
|
214
|
// }
|
215
|
|
216
|
if (pilotOutputExpConfig.isDoReferencedEntities() == true) {
|
217
|
logger.info("# Referenced Entities");
|
218
|
sddDataSet.setReferencedEntities(getNameService().getAllNomenclaturalStatus(MAX_ROWS, 0));
|
219
|
sddDataSet.addSourcedEntities(getNameService().getAllTypeDesignations(MAX_ROWS, 0));
|
220
|
}
|
221
|
|
222
|
if (pilotOutputExpConfig.isDoOccurrence() == true) {
|
223
|
if (occurrencesRows == 0) { occurrencesRows = getOccurrenceService().count(SpecimenOrObservationBase.class); }
|
224
|
logger.info("# SpecimenOrObservationBase: " + occurrencesRows);
|
225
|
sddDataSet.setOccurrences(getOccurrenceService().list(null,occurrencesRows, 0,null,null));
|
226
|
}
|
227
|
|
228
|
if (pilotOutputExpConfig.isDoMedia() == true) {
|
229
|
if (mediaRows == 0) { mediaRows = MAX_ROWS; }
|
230
|
logger.info("# Media");
|
231
|
sddDataSet.setMedia(getMediaService().list(null,mediaRows, 0,null,null));
|
232
|
// dataSet.addMedia(getMediaService().getAllMediaRepresentations(mediaRows, 0));
|
233
|
// dataSet.addMedia(getMediaService().getAllMediaRepresentationParts(mediaRows, 0));
|
234
|
}
|
235
|
|
236
|
if (pilotOutputExpConfig.isDoFeatureData() == true) {
|
237
|
if (featureDataRows == 0) { featureDataRows = MAX_ROWS; }
|
238
|
logger.info("# Feature Tree, Feature Node");
|
239
|
sddDataSet.setFeatureData(getTermNodeService().list(TermType.Feature,null,null,null,null));
|
240
|
sddDataSet.addFeatureData(getTermTreeService().list(TermType.Feature,null,null,null,null));
|
241
|
}
|
242
|
}
|
243
|
|
244
|
@Override
|
245
|
protected boolean doCheck(PilotOutputExportState state) {
|
246
|
boolean result = true;
|
247
|
logger.warn("No check implemented for Jaxb export");
|
248
|
return result;
|
249
|
}
|
250
|
|
251
|
@Override
|
252
|
protected boolean isIgnore(PilotOutputExportState state) {
|
253
|
return false;
|
254
|
}
|
255
|
|
256
|
@Override
|
257
|
public byte[] getByteArray() {
|
258
|
// TODO Auto-generated method stub
|
259
|
return null;
|
260
|
}
|
261
|
}
|