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.stream;
|
10
|
|
11
|
import java.util.HashMap;
|
12
|
import java.util.HashSet;
|
13
|
import java.util.List;
|
14
|
import java.util.Map;
|
15
|
import java.util.Set;
|
16
|
|
17
|
import org.apache.commons.lang.StringUtils;
|
18
|
import org.apache.log4j.Logger;
|
19
|
|
20
|
import eu.etaxonomy.cdm.io.common.events.IIoEvent;
|
21
|
import eu.etaxonomy.cdm.io.common.events.IIoObserver;
|
22
|
import eu.etaxonomy.cdm.io.common.events.IoProblemEvent;
|
23
|
import eu.etaxonomy.cdm.io.dwca.in.DwcaDataImportConfiguratorBase;
|
24
|
import eu.etaxonomy.cdm.io.stream.terms.TermUri;
|
25
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
26
|
import eu.etaxonomy.cdm.model.common.Language;
|
27
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
28
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
29
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
30
|
|
31
|
/**
|
32
|
* @author a.mueller
|
33
|
* @since 23.11.2011
|
34
|
*
|
35
|
* FIXME URI
|
36
|
*/
|
37
|
public abstract class PartitionableConverterBase<CONFIG extends DwcaDataImportConfiguratorBase, STATE extends StreamImportStateBase<CONFIG,StreamImportBase>>
|
38
|
implements IPartitionableConverter<StreamItem, IReader<CdmBase>, String> {
|
39
|
|
40
|
private static final Logger logger = Logger.getLogger(PartitionableConverterBase.class);
|
41
|
|
42
|
protected STATE state;
|
43
|
protected CONFIG config; //for convenience only (must always be same as state.getConfig())
|
44
|
|
45
|
|
46
|
|
47
|
public PartitionableConverterBase(STATE state) {
|
48
|
super();
|
49
|
this.state = state;
|
50
|
this.config = state.getConfig(); //for fast access
|
51
|
}
|
52
|
|
53
|
protected void fireWarningEvent(String message, StreamItem item, Integer severity) {
|
54
|
fireWarningEvent(message, getDataLocation(item), severity, 1);
|
55
|
}
|
56
|
|
57
|
private String getDataLocation(StreamItem item) {
|
58
|
String location = item.getLocation();
|
59
|
return location;
|
60
|
}
|
61
|
|
62
|
protected void fireWarningEvent(String message, String dataLocation, Integer severity) {
|
63
|
fireWarningEvent(message, dataLocation, severity, 1);
|
64
|
}
|
65
|
|
66
|
protected void fireWarningEvent(String message, String dataLocation, Integer severity, int stackDepth) {
|
67
|
stackDepth++;
|
68
|
StackTraceElement[] stackTrace = new Exception().getStackTrace();
|
69
|
int lineNumber = stackTrace[stackDepth].getLineNumber();
|
70
|
String methodName = stackTrace[stackDepth].getMethodName();
|
71
|
String className = stackTrace[stackDepth].getClassName();
|
72
|
Class<?> declaringClass;
|
73
|
try {
|
74
|
declaringClass = Class.forName(className);
|
75
|
} catch (ClassNotFoundException e) {
|
76
|
declaringClass = this.getClass();
|
77
|
}
|
78
|
|
79
|
IoProblemEvent event = IoProblemEvent.NewInstance(declaringClass, message, dataLocation,
|
80
|
lineNumber, severity, methodName);
|
81
|
|
82
|
//for performance improvement one may read:
|
83
|
//http://stackoverflow.com/questions/421280/in-java-how-do-i-find-the-caller-of-a-method-using-stacktrace-or-reflection
|
84
|
// Object o = new SecurityManager().getSecurityContext();
|
85
|
|
86
|
fire(event);
|
87
|
}
|
88
|
|
89
|
protected void fire(IIoEvent event){
|
90
|
Set<IIoObserver> observers = state.getConfig().getObservers();
|
91
|
for (IIoObserver observer: observers){
|
92
|
observer.handleEvent(event);
|
93
|
}
|
94
|
if (observers.size() == 0){
|
95
|
logger.warn(event.getMessage() + " (no observer for message!).");
|
96
|
}
|
97
|
}
|
98
|
|
99
|
|
100
|
|
101
|
/**
|
102
|
* Returns the value for the given term in the item.
|
103
|
*/
|
104
|
protected String getValue(StreamItem item, TermUri term) {
|
105
|
return item.get(term.getUriString());
|
106
|
}
|
107
|
|
108
|
/**
|
109
|
* Checks if the given term has a value in item that is not blank (null, empty or whitespace only).
|
110
|
* @param term
|
111
|
* @param item
|
112
|
* @return true if value is not blank
|
113
|
*/
|
114
|
protected boolean exists(TermUri term, StreamItem item) {
|
115
|
return ! StringUtils.isBlank(getValue(item, term));
|
116
|
}
|
117
|
|
118
|
/*
|
119
|
* To be implemented by subclasses if needed.
|
120
|
*/
|
121
|
@Override
|
122
|
public ItemFilter<StreamItem> getItemFilter(){
|
123
|
return null;
|
124
|
}
|
125
|
|
126
|
@Override
|
127
|
public Map<String, Set<String>> getPartitionForeignKeys(IReader<StreamItem> instream) {
|
128
|
Map<String, Set<String>> result = new HashMap<>();
|
129
|
|
130
|
while (instream.hasNext()){
|
131
|
StreamItem next = instream.read();
|
132
|
makeForeignKeysForItem(next, result);
|
133
|
}
|
134
|
return result;
|
135
|
}
|
136
|
|
137
|
|
138
|
/**
|
139
|
* Fills the the foreign key map with foreign keys required for this item.
|
140
|
* @param next
|
141
|
* @param foreignKeyMap
|
142
|
*/
|
143
|
protected abstract void makeForeignKeysForItem(StreamItem next, Map<String, Set<String>> foreignKeyMap);
|
144
|
|
145
|
|
146
|
/**
|
147
|
* False if string is null, empty or whitespace only. True otherwise.
|
148
|
* @param string String to test.
|
149
|
*/
|
150
|
protected boolean hasValue(String string) {
|
151
|
return StringUtils.isNotBlank(string);
|
152
|
}
|
153
|
|
154
|
|
155
|
/**
|
156
|
* Returns the key set for a given key or creates a new one.
|
157
|
* @param key
|
158
|
* @param fkMap
|
159
|
* @return
|
160
|
*/
|
161
|
protected Set<String> getKeySet(String key, Map<String, Set<String>> fkMap) {
|
162
|
Set<String> keySet = fkMap.get(key);
|
163
|
if (keySet == null){
|
164
|
keySet = new HashSet<>();
|
165
|
fkMap.put(key, keySet);
|
166
|
}
|
167
|
return keySet;
|
168
|
}
|
169
|
|
170
|
|
171
|
protected <T extends TaxonBase<?>> T getTaxonBase(String id, StreamItem item, Class<T> clazz, STATE state) {
|
172
|
if (clazz == null){
|
173
|
clazz = (Class)TaxonBase.class;
|
174
|
}
|
175
|
List<T> taxonList = state.get(TermUri.DWC_TAXON.toString(), id, clazz);
|
176
|
if (taxonList.size() > 1){
|
177
|
String message = "Ambigous taxon mapping for id %s. There is more than 1 matching taxon.";
|
178
|
message = String.format(message, id);
|
179
|
fireWarningEvent(message, item, 8);
|
180
|
logger.warn(message); //TODO remove when events are handled correctly
|
181
|
return null;
|
182
|
}else if (taxonList.isEmpty()){
|
183
|
return null;
|
184
|
}else{
|
185
|
return taxonList.get(0);
|
186
|
}
|
187
|
}
|
188
|
|
189
|
|
190
|
protected TaxonDescription getTaxonDescription(Taxon taxon, boolean isImageGallery) {
|
191
|
TaxonDescription result = null;
|
192
|
Set<TaxonDescription> descs = taxon.getDescriptions();
|
193
|
for (TaxonDescription desc : descs){
|
194
|
if (desc.isImageGallery() == isImageGallery){
|
195
|
result = desc;
|
196
|
break;
|
197
|
}
|
198
|
}
|
199
|
if (result == null){
|
200
|
result = TaxonDescription.NewInstance(taxon);
|
201
|
result.setImageGallery(isImageGallery);
|
202
|
}
|
203
|
return result;
|
204
|
}
|
205
|
|
206
|
|
207
|
protected Language getDcLanguage(StreamItem item, List<MappedCdmBase<? extends CdmBase>> resultList ) {
|
208
|
String namespace = TermUri.DC_LANGUAGE.toString();
|
209
|
String languageStr = item.get(TermUri.DC_LANGUAGE);
|
210
|
if (isBlank(languageStr)){
|
211
|
return null;
|
212
|
}
|
213
|
|
214
|
List<Language> list = state.get(namespace, languageStr, Language.class);
|
215
|
if (list.isEmpty()){
|
216
|
//try to find in cdm
|
217
|
Language lang = Language.getLanguageByIsoCode(languageStr);
|
218
|
if (lang == null){
|
219
|
lang = Language.getLanguageByLabel(languageStr);
|
220
|
}
|
221
|
if (lang == null){
|
222
|
lang = state.getCurrentIO().getLanguage(state, null, languageStr, languageStr, languageStr, null);
|
223
|
}
|
224
|
state.putMapping(namespace, languageStr, lang);
|
225
|
list.add(lang);
|
226
|
// MappedCdmBase<? extends CdmBase> languageMcb = new MappedCdmBase<>(TermUri.DC_LANGUAGE, languageStr, lang);
|
227
|
// resultList.add(languageMcb);
|
228
|
}
|
229
|
if (list.size() > 1){
|
230
|
String message = "There is more than 1 cdm entity matching given language '%s'. I take an arbitrary one.";
|
231
|
fireWarningEvent(String.format(message, languageStr), item, 4);
|
232
|
}
|
233
|
Language result = list.iterator().next();
|
234
|
return result;
|
235
|
}
|
236
|
|
237
|
protected boolean isNotBlank(String str){
|
238
|
return StringUtils.isNotBlank(str);
|
239
|
}
|
240
|
|
241
|
protected boolean isBlank(String str){
|
242
|
return StringUtils.isBlank(str);
|
243
|
}
|
244
|
|
245
|
}
|