1 |
91904a96
|
Andreas Müller
|
/**
|
2 |
|
|
* Copyright (C) 2009 EDIT
|
3 |
7d882578
|
Andreas Müller
|
* European Distributed Institute of Taxonomy
|
4 |
91904a96
|
Andreas Müller
|
* http://www.e-taxonomy.eu
|
5 |
7d882578
|
Andreas Müller
|
*
|
6 |
91904a96
|
Andreas Müller
|
* 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.io.wp6;
|
11 |
|
|
|
12 |
|
|
import java.util.ArrayList;
|
13 |
|
|
import java.util.HashMap;
|
14 |
|
|
import java.util.List;
|
15 |
|
|
import java.util.Map;
|
16 |
|
|
import java.util.regex.Matcher;
|
17 |
|
|
import java.util.regex.Pattern;
|
18 |
|
|
|
19 |
|
|
import org.apache.commons.lang.StringUtils;
|
20 |
|
|
import org.apache.log4j.Logger;
|
21 |
|
|
|
22 |
7d882578
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.name.IBotanicalName;
|
23 |
91904a96
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
24 |
|
|
import eu.etaxonomy.cdm.strategy.parser.INonViralNameParser;
|
25 |
|
|
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl;
|
26 |
|
|
|
27 |
|
|
/**
|
28 |
|
|
* @author a.babadshanjan
|
29 |
|
|
* @created 13.01.2009
|
30 |
|
|
* @version 1.0
|
31 |
|
|
*/
|
32 |
|
|
public class CommonNameRow {
|
33 |
|
|
private static final Logger logger = Logger.getLogger(CommonNameRow.class);
|
34 |
|
|
|
35 |
|
|
|
36 |
|
|
private String species;
|
37 |
|
|
private String reference;
|
38 |
|
|
private String area;
|
39 |
218a4795
|
Andreas Müller
|
private String nameUsedInSource;
|
40 |
7d882578
|
Andreas Müller
|
|
41 |
91904a96
|
Andreas Müller
|
private Map<String, List<String>> commonNames = new HashMap<String, List<String>>();
|
42 |
7d882578
|
Andreas Müller
|
|
43 |
|
|
|
44 |
91904a96
|
Andreas Müller
|
public CommonNameRow() {
|
45 |
|
|
this.species = "";
|
46 |
|
|
this.reference = "";
|
47 |
|
|
this.area = "";
|
48 |
|
|
commonNames = new HashMap<String, List<String>>();
|
49 |
|
|
}
|
50 |
7d882578
|
Andreas Müller
|
|
51 |
|
|
// **************************** GETTER / SETTER *********************************/
|
52 |
|
|
|
53 |
91904a96
|
Andreas Müller
|
public void setCommonNames(String commonNamesString){
|
54 |
218a4795
|
Andreas Müller
|
commonNamesString = makeNameUsedInSource(commonNamesString);
|
55 |
91904a96
|
Andreas Müller
|
String[] split = commonNamesString.split(";");
|
56 |
|
|
for (String oneLanguage : split){
|
57 |
|
|
oneLanguage = oneLanguage.trim();
|
58 |
34b56211
|
Andreas Müller
|
String reLangPattern = "\\((\\*|[a-z]{2,3}|.{2,},\\sno\\sISO-Code)\\)$";
|
59 |
91904a96
|
Andreas Müller
|
String pattern = ".+" + reLangPattern;
|
60 |
|
|
if (! oneLanguage.matches(pattern)){
|
61 |
|
|
logger.warn("Common name does not match: " + oneLanguage);
|
62 |
|
|
}else{
|
63 |
|
|
Pattern langPattern = Pattern.compile(reLangPattern);
|
64 |
|
|
Matcher matcher = langPattern.matcher(oneLanguage);
|
65 |
|
|
if (matcher.find()){
|
66 |
|
|
String lang = matcher.group().substring(1);
|
67 |
|
|
lang = lang.substring(0, lang.length()-1);
|
68 |
|
|
String names = "";
|
69 |
|
|
try {
|
70 |
|
|
names = oneLanguage.substring(0,matcher.start()-1);
|
71 |
|
|
} catch (Exception e) {
|
72 |
|
|
e.printStackTrace();
|
73 |
|
|
}
|
74 |
|
|
String[] splitNames = names.split(",");
|
75 |
|
|
List<String> nameList = new ArrayList<String>();
|
76 |
|
|
for (String singleName : splitNames){
|
77 |
|
|
if (StringUtils.isNotBlank(singleName)){
|
78 |
|
|
nameList.add(singleName.trim());
|
79 |
|
|
}
|
80 |
|
|
}
|
81 |
|
|
if (!nameList.isEmpty()){
|
82 |
|
|
this.commonNames.put(lang, nameList);
|
83 |
|
|
}
|
84 |
|
|
}else{
|
85 |
|
|
logger.warn("Common name does not match: " + oneLanguage);
|
86 |
|
|
}
|
87 |
|
|
}
|
88 |
|
|
}
|
89 |
|
|
}
|
90 |
7d882578
|
Andreas Müller
|
|
91 |
218a4795
|
Andreas Müller
|
private String makeNameUsedInSource(String commonNamesString) {
|
92 |
|
|
String[] split = commonNamesString.split(":");
|
93 |
|
|
if (split.length > 1){
|
94 |
001bef66
|
Andreas Müller
|
logger.debug("NameUsedInSource: " + split[0]);
|
95 |
218a4795
|
Andreas Müller
|
this.nameUsedInSource = split[0].trim();
|
96 |
|
|
if (split.length > 2){
|
97 |
|
|
logger.warn("XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:" + commonNamesString);
|
98 |
|
|
}
|
99 |
|
|
return split[1].trim();
|
100 |
|
|
}else{
|
101 |
|
|
return split[0].trim();
|
102 |
|
|
}
|
103 |
|
|
}
|
104 |
|
|
|
105 |
91904a96
|
Andreas Müller
|
public Map<String, List<String>> getCommonNames() {
|
106 |
|
|
return commonNames;
|
107 |
|
|
}
|
108 |
7d882578
|
Andreas Müller
|
|
109 |
91904a96
|
Andreas Müller
|
public void parseSpecies(String species){
|
110 |
7d882578
|
Andreas Müller
|
INonViralNameParser<?> parser = NonViralNameParserImpl.NewInstance();
|
111 |
|
|
IBotanicalName name = (IBotanicalName)parser.parseFullName(species, NomenclaturalCode.ICNAFP, null);
|
112 |
91904a96
|
Andreas Müller
|
if (name.isProtectedTitleCache()){
|
113 |
|
|
logger.warn("Name could not be parsed: " + species);
|
114 |
|
|
}
|
115 |
c273c734
|
Andreas Müller
|
this.species = species;
|
116 |
91904a96
|
Andreas Müller
|
}
|
117 |
|
|
|
118 |
|
|
public String getSpecies() {
|
119 |
|
|
return species;
|
120 |
|
|
}
|
121 |
7d882578
|
Andreas Müller
|
|
122 |
91904a96
|
Andreas Müller
|
public void setSpecies(String species) {
|
123 |
|
|
this.species = species;
|
124 |
|
|
}
|
125 |
|
|
|
126 |
|
|
public void setReference(String reference) {
|
127 |
|
|
reference = reference.replace(".", "");
|
128 |
|
|
if (! reference.matches("\\d{7}")){
|
129 |
|
|
logger.warn("Unexpected reference");
|
130 |
|
|
}
|
131 |
218a4795
|
Andreas Müller
|
this.reference = reference.substring(0,6);
|
132 |
91904a96
|
Andreas Müller
|
}
|
133 |
|
|
|
134 |
|
|
public String getReference() {
|
135 |
|
|
return reference;
|
136 |
|
|
}
|
137 |
|
|
|
138 |
|
|
public void setArea(String area) {
|
139 |
|
|
this.area = area;
|
140 |
|
|
}
|
141 |
7d882578
|
Andreas Müller
|
|
142 |
91904a96
|
Andreas Müller
|
public String getArea() {
|
143 |
|
|
return area;
|
144 |
|
|
}
|
145 |
|
|
|
146 |
218a4795
|
Andreas Müller
|
public void setNameUsedInSource(String nameUsedInSource) {
|
147 |
|
|
this.nameUsedInSource = nameUsedInSource;
|
148 |
|
|
}
|
149 |
|
|
|
150 |
|
|
public String getNameUsedInSource() {
|
151 |
|
|
return nameUsedInSource;
|
152 |
|
|
}
|
153 |
|
|
|
154 |
91904a96
|
Andreas Müller
|
|
155 |
7d882578
|
Andreas Müller
|
|
156 |
91904a96
|
Andreas Müller
|
}
|