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.dwca.out;
|
10
|
|
11
|
import java.io.FileNotFoundException;
|
12
|
import java.io.IOException;
|
13
|
import java.io.UnsupportedEncodingException;
|
14
|
import java.util.ArrayList;
|
15
|
import java.util.HashSet;
|
16
|
import java.util.List;
|
17
|
import java.util.Set;
|
18
|
import java.util.UUID;
|
19
|
|
20
|
import org.apache.commons.lang.StringUtils;
|
21
|
import org.apache.log4j.Logger;
|
22
|
import org.springframework.beans.factory.annotation.Autowired;
|
23
|
|
24
|
import eu.etaxonomy.cdm.api.service.IClassificationService;
|
25
|
import eu.etaxonomy.cdm.api.service.ITaxonNodeService;
|
26
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
27
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
28
|
import eu.etaxonomy.cdm.model.common.IOriginalSource;
|
29
|
import eu.etaxonomy.cdm.model.common.ISourceable;
|
30
|
import eu.etaxonomy.cdm.model.description.DescriptionElementSource;
|
31
|
import eu.etaxonomy.cdm.model.location.Country;
|
32
|
import eu.etaxonomy.cdm.model.location.NamedArea;
|
33
|
import eu.etaxonomy.cdm.model.taxon.Classification;
|
34
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
35
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
36
|
import eu.etaxonomy.cdm.model.taxon.TaxonNode;
|
37
|
import eu.etaxonomy.cdm.persistence.dto.TaxonNodeDto;
|
38
|
|
39
|
/**
|
40
|
* @author a.mueller
|
41
|
* @date 18.04.2011
|
42
|
*
|
43
|
*/
|
44
|
public abstract class DwcaDataExportBase extends DwcaExportBase{
|
45
|
|
46
|
private static final long serialVersionUID = -5467295551060073610L;
|
47
|
|
48
|
@SuppressWarnings("unused")
|
49
|
private static final Logger logger = Logger.getLogger(DwcaDataExportBase.class);
|
50
|
|
51
|
@Autowired
|
52
|
private IClassificationService classificationService;
|
53
|
|
54
|
@Autowired
|
55
|
private ITaxonNodeService taxonNodeService;
|
56
|
|
57
|
abstract protected void handleTaxonNode(DwcaTaxExportState state, TaxonNode node)throws IOException, FileNotFoundException, UnsupportedEncodingException;
|
58
|
|
59
|
|
60
|
/**
|
61
|
* {@inheritDoc}
|
62
|
*/
|
63
|
@Override
|
64
|
public int countSteps(DwcaTaxExportState state) {
|
65
|
//FIXME count without initialization
|
66
|
List<TaxonNode> allNodes = allNodes(state);
|
67
|
return allNodes.size();
|
68
|
}
|
69
|
|
70
|
/**
|
71
|
* Returns the list of {@link TaxonNode taxon nodes} that correspond to the
|
72
|
* given filter criteria (e.g. subtreeUUids). If no filter is given
|
73
|
* all taxon nodes of all classifications are returned. If the list has been
|
74
|
* computed before it is taken from the state cache. Nodes that do not have
|
75
|
* a taxon attached are not returned. Instead a warning is given that the node is
|
76
|
* ommitted (empty taxon nodes should not but do exist in CDM databases).
|
77
|
* <BR>
|
78
|
* Preliminary implementation. Better implement API method for this.
|
79
|
*/
|
80
|
//TODO unify with similar methods for other exports
|
81
|
protected List<TaxonNode> allNodes(DwcaTaxExportState state) {
|
82
|
|
83
|
Set<UUID> subtreeUuidSet = state.getConfig().getSubtreeUuids();
|
84
|
if (subtreeUuidSet == null){
|
85
|
subtreeUuidSet = new HashSet<>();
|
86
|
}
|
87
|
//handle empty list as no filter defined
|
88
|
if (subtreeUuidSet.isEmpty()){
|
89
|
List<Classification> classificationList = getClassificationService().list(Classification.class, null, 0, null, null);
|
90
|
for (Classification classification : classificationList){
|
91
|
subtreeUuidSet.add(classification.getRootNode().getUuid());
|
92
|
}
|
93
|
}
|
94
|
|
95
|
//TODO memory critical to store ALL node
|
96
|
if (state.getAllNodes().isEmpty()){
|
97
|
makeAllNodes(state, subtreeUuidSet);
|
98
|
}
|
99
|
List<TaxonNode> allNodes = state.getAllNodes();
|
100
|
return allNodes;
|
101
|
}
|
102
|
|
103
|
private void makeAllNodes(DwcaTaxExportState state, Set<UUID> subtreeSet) {
|
104
|
|
105
|
try {
|
106
|
boolean doSynonyms = false;
|
107
|
boolean recursive = true;
|
108
|
Set<UUID> uuidSet = new HashSet<>();
|
109
|
|
110
|
for (UUID subtreeUuid : subtreeSet){
|
111
|
UUID tnUuuid = taxonNodeUuid(subtreeUuid);
|
112
|
uuidSet.add(tnUuuid);
|
113
|
List<TaxonNodeDto> records = getTaxonNodeService().pageChildNodesDTOs(tnUuuid,
|
114
|
recursive, doSynonyms, null, null, null).getRecords();
|
115
|
for (TaxonNodeDto dto : records){
|
116
|
uuidSet.add(dto.getUuid());
|
117
|
}
|
118
|
}
|
119
|
List<TaxonNode> allNodes = getTaxonNodeService().find(uuidSet);
|
120
|
|
121
|
List<TaxonNode> result = new ArrayList<>();
|
122
|
for (TaxonNode node : allNodes){
|
123
|
if(node.getParent()== null){ //root (or invalid) node
|
124
|
continue;
|
125
|
}
|
126
|
node = CdmBase.deproxy(node);
|
127
|
Taxon taxon = CdmBase.deproxy(node.getTaxon());
|
128
|
if (taxon == null){
|
129
|
String message = "There is a taxon node without taxon. id=" + node.getId();
|
130
|
state.getResult().addWarning(message);
|
131
|
continue;
|
132
|
}
|
133
|
result.add(node);
|
134
|
}
|
135
|
state.setAllNodes(result);
|
136
|
} catch (Exception e) {
|
137
|
String message = "Unexpected exception when trying to compute all taxon nodes";
|
138
|
state.getResult().addException(e, message);
|
139
|
}
|
140
|
}
|
141
|
|
142
|
|
143
|
/**
|
144
|
* @param subtreeUuid
|
145
|
* @return
|
146
|
*/
|
147
|
private UUID taxonNodeUuid(UUID subtreeUuid) {
|
148
|
TaxonNode node = taxonNodeService.find(subtreeUuid);
|
149
|
if (node == null){
|
150
|
Classification classification = classificationService.find(subtreeUuid);
|
151
|
if (classification != null){
|
152
|
node = classification.getRootNode();
|
153
|
}else{
|
154
|
throw new IllegalArgumentException("Subtree identifier does not exist: " + subtreeUuid);
|
155
|
}
|
156
|
}
|
157
|
return node.getUuid();
|
158
|
}
|
159
|
|
160
|
/**
|
161
|
* Creates the locationId, locality, countryCode triple
|
162
|
* @param state
|
163
|
* @param record
|
164
|
* @param area
|
165
|
*/
|
166
|
protected void handleArea(DwcaTaxExportState state, IDwcaAreaRecord record, NamedArea area, TaxonBase<?> taxon, boolean required) {
|
167
|
if (area != null){
|
168
|
record.setLocationId(area);
|
169
|
record.setLocality(area.getLabel());
|
170
|
if (area.isInstanceOf(Country.class)){
|
171
|
Country country = CdmBase.deproxy(area, Country.class);
|
172
|
record.setCountryCode(country.getIso3166_A2());
|
173
|
}
|
174
|
}else{
|
175
|
if (required){
|
176
|
String message = "Description requires area but area does not exist for taxon " + getTaxonLogString(taxon);
|
177
|
state.getResult().addWarning(message);
|
178
|
}
|
179
|
}
|
180
|
}
|
181
|
|
182
|
|
183
|
protected String getTaxonLogString(TaxonBase<?> taxon) {
|
184
|
return taxon.getTitleCache() + "(" + taxon.getId() + ")";
|
185
|
}
|
186
|
|
187
|
|
188
|
protected String getSources(ISourceable<?> sourceable, DwcaTaxExportConfigurator config) {
|
189
|
String result = "";
|
190
|
for (IOriginalSource<?> source: sourceable.getSources()){
|
191
|
if (StringUtils.isBlank(source.getIdInSource())){//idInSource indicates that this source is only data provenance, may be changed in future
|
192
|
if (source.getCitation() != null){
|
193
|
String ref = source.getCitation().getTitleCache();
|
194
|
result = CdmUtils.concat(config.getSetSeparator(), result, ref);
|
195
|
}
|
196
|
}
|
197
|
}
|
198
|
return result;
|
199
|
}
|
200
|
|
201
|
protected String getSources3(ISourceable<?> sourceable, DwcaTaxExportConfigurator config) {
|
202
|
String result = "";
|
203
|
for (IOriginalSource<?> source: sourceable.getSources()){
|
204
|
if (source.getCitation() != null){
|
205
|
String ref = source.getCitation().getTitleCache();
|
206
|
result = CdmUtils.concat(config.getSetSeparator(), result, ref);
|
207
|
}
|
208
|
}
|
209
|
return result;
|
210
|
}
|
211
|
|
212
|
protected String getSources2(Set<DescriptionElementSource> set, DwcaTaxExportConfigurator config) {
|
213
|
String result = "";
|
214
|
for(DescriptionElementSource source: set){
|
215
|
if (StringUtils.isBlank(source.getIdInSource())){//idInSource indicates that this source is only data provenance, may be changed in future
|
216
|
if (source.getCitation() != null){
|
217
|
String ref = source.getCitation().getTitleCache();
|
218
|
result = CdmUtils.concat(config.getSetSeparator(), result, ref);
|
219
|
}
|
220
|
}
|
221
|
}
|
222
|
return result;
|
223
|
}
|
224
|
|
225
|
}
|