1
|
/**
|
2
|
* Copyright (C) 2016 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.greece;
|
10
|
|
11
|
import java.util.Arrays;
|
12
|
import java.util.HashMap;
|
13
|
import java.util.List;
|
14
|
import java.util.Set;
|
15
|
import java.util.regex.Matcher;
|
16
|
import java.util.regex.Pattern;
|
17
|
|
18
|
import org.apache.log4j.Logger;
|
19
|
import org.springframework.stereotype.Component;
|
20
|
|
21
|
import eu.etaxonomy.cdm.api.service.config.MatchingTaxonConfigurator;
|
22
|
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState;
|
23
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
24
|
import eu.etaxonomy.cdm.model.common.Language;
|
25
|
import eu.etaxonomy.cdm.model.common.OriginalSourceType;
|
26
|
import eu.etaxonomy.cdm.model.description.Feature;
|
27
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
28
|
import eu.etaxonomy.cdm.model.description.TextData;
|
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
|
|
34
|
/**
|
35
|
* @author a.mueller
|
36
|
* @date 14.12.2016
|
37
|
*
|
38
|
*/
|
39
|
|
40
|
@Component
|
41
|
public class FloraHellenicaImageCaptionImport<CONFIG extends FloraHellenicaImportConfigurator>
|
42
|
extends FloraHellenicaImportBase<CONFIG>{
|
43
|
|
44
|
private static final long serialVersionUID = 2629253144140992196L;
|
45
|
private static final Logger logger = Logger.getLogger(FloraHellenicaImageCaptionImport.class);
|
46
|
|
47
|
private static final String TEXT = "Text";
|
48
|
protected static Integer startPage = 316;
|
49
|
|
50
|
|
51
|
private static List<String> expectedKeys= Arrays.asList(new String[]{
|
52
|
TEXT
|
53
|
});
|
54
|
|
55
|
|
56
|
@Override
|
57
|
protected String getWorksheetName() {
|
58
|
return "Captions";
|
59
|
}
|
60
|
|
61
|
/**
|
62
|
* {@inheritDoc}
|
63
|
*/
|
64
|
@Override
|
65
|
protected void firstPass(SimpleExcelTaxonImportState<CONFIG> state) {
|
66
|
|
67
|
String line = state.getCurrentLine() + ": ";
|
68
|
HashMap<String, String> record = state.getOriginalRecord();
|
69
|
|
70
|
Set<String> keys = record.keySet();
|
71
|
for (String key: keys) {
|
72
|
if (! expectedKeys.contains(key)){
|
73
|
logger.warn(line + "Unexpected key: " + key);
|
74
|
}
|
75
|
}
|
76
|
|
77
|
makeCaption(state, line, record);
|
78
|
|
79
|
}
|
80
|
|
81
|
|
82
|
|
83
|
Integer plateNo = 0;
|
84
|
private Taxon makeCaption(SimpleExcelTaxonImportState<CONFIG> state,
|
85
|
String line,
|
86
|
HashMap<String, String> record) {
|
87
|
|
88
|
String text = getValue(record, TEXT);
|
89
|
if (text.matches("Plate \\d\\d?")){
|
90
|
plateNo = Integer.valueOf(text.substring(6));
|
91
|
return null;
|
92
|
}
|
93
|
Pattern pattern = Pattern.compile(
|
94
|
"(\\d)\\. ([A-Z][a-z\\-]+ [a-z\\-]+(?: subsp. [a-z\\-]+)?) (\\([A-Z][a-z\\-]+\\))(.*)");
|
95
|
Matcher matcher = pattern.matcher(text);
|
96
|
if (!matcher.matches()){
|
97
|
logger.warn(line + "String caption not recognized: " + text);
|
98
|
return null;
|
99
|
}else{
|
100
|
String subNo = matcher.group(1);
|
101
|
String name = matcher.group(2);
|
102
|
name = adaptName(name);
|
103
|
MatchingTaxonConfigurator matchConfig = new MatchingTaxonConfigurator();
|
104
|
matchConfig.setTaxonNameTitle(name);
|
105
|
matchConfig.setIncludeSynonyms(false);
|
106
|
List<TaxonBase> taxa = getTaxonService().findTaxaByName(matchConfig);
|
107
|
TaxonBase<?> taxonBase;
|
108
|
if (taxa.isEmpty()){
|
109
|
logger.warn(line + "Taxon not found for name: " + name);
|
110
|
return null;
|
111
|
}else if (taxa.size() > 1){
|
112
|
logger.warn(line + "Found more then 1 taxon for name: " + name);
|
113
|
}
|
114
|
taxonBase = taxa.get(0);
|
115
|
Taxon taxon;
|
116
|
if (taxonBase.isInstanceOf(Synonym.class)){
|
117
|
taxon = CdmBase.deproxy(taxonBase, Synonym.class).getAcceptedTaxon();
|
118
|
logger.warn(line + "Taxon name is synonym: " + name);
|
119
|
}else{
|
120
|
taxon = CdmBase.deproxy(taxonBase, Taxon.class);
|
121
|
}
|
122
|
TaxonDescription td = getTaxonDescription(taxon, false, true);
|
123
|
|
124
|
String laterText = matcher.group(4);
|
125
|
if (laterText.startsWith(". This")){
|
126
|
laterText = laterText.substring(6);
|
127
|
}
|
128
|
text = matcher.group(2) + laterText;
|
129
|
|
130
|
Feature feature = getFeature(state, FloraHellenicaTransformer.uuidFloraHellenicaTaxonInfoFeature);
|
131
|
TextData textData = TextData.NewInstance(feature, text, Language.ENGLISH(), null);
|
132
|
td.addElement(textData);
|
133
|
Reference citation = this.getSecReference(state);
|
134
|
Integer myPage = startPage + plateNo * 2;
|
135
|
|
136
|
textData.addSource(OriginalSourceType.PrimaryTaxonomicSource, plateNo + "." +subNo
|
137
|
, "Plate", citation, "p. " + myPage);
|
138
|
|
139
|
return taxon;
|
140
|
|
141
|
}
|
142
|
}
|
143
|
|
144
|
/**
|
145
|
* @param name
|
146
|
* @return
|
147
|
*/
|
148
|
private String adaptName(String name) {
|
149
|
if (name.equals("Lathraea rhodopaea")){
|
150
|
name = "Lathraea rhodopea";
|
151
|
}else if (name.equals("Soldanella chrysosticha subsp. chrysosticha")){
|
152
|
name = "Soldanella chrysosticta subsp. chrysosticta";
|
153
|
}
|
154
|
return name;
|
155
|
}
|
156
|
|
157
|
}
|