1 |
d5d79aea
|
Andreas Müller
|
/**
|
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 |
|
|
|
10 |
eb1471a5
|
Andreas Müller
|
package eu.etaxonomy.cdm.compare.taxon;
|
11 |
d5d79aea
|
Andreas Müller
|
|
12 |
|
|
import java.util.Comparator;
|
13 |
|
|
import java.util.HashSet;
|
14 |
|
|
import java.util.Set;
|
15 |
|
|
|
16 |
23de68fc
|
Andreas Müller
|
import org.apache.logging.log4j.LogManager;import org.apache.logging.log4j.Logger;
|
17 |
d5d79aea
|
Andreas Müller
|
|
18 |
|
|
import eu.etaxonomy.cdm.model.name.NameRelationship;
|
19 |
|
|
import eu.etaxonomy.cdm.model.name.NameRelationshipType;
|
20 |
9dc896c9
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
21 |
eb1471a5
|
Andreas Müller
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
22 |
d5d79aea
|
Andreas Müller
|
|
23 |
|
|
/**
|
24 |
|
|
* This class orders synonyms of a homotypic group,
|
25 |
|
|
* first by
|
26 |
|
|
* <ul>
|
27 |
|
|
* <li>Basionym groups (the basionym and all names derived from this basionym)
|
28 |
|
|
* should be kept together in a subgroup</li>
|
29 |
|
|
* <li>The order of the subgroups is defined by the ordering of their
|
30 |
|
|
* basionyms (according to the following ordering)</li>
|
31 |
|
|
* <li>If a name is illegitimate or not does play a role for ordering</li>
|
32 |
|
|
* <li>Names with publication year should always come first</li>
|
33 |
82ca4447
|
Andreas Müller
|
* <li>Names with no publication year are sorted by rank</li>
|
34 |
|
|
* <li>Names with no publication year and equal rank are sorted alphabetically</li>
|
35 |
d5d79aea
|
Andreas Müller
|
* <li>If 2 names have a replaced synonym relationship the replaced synonym comes first,
|
36 |
|
|
* the replacement name comes later as this reflects the order of publication</li>
|
37 |
|
|
* </ul>
|
38 |
|
|
*
|
39 |
138715e0
|
Andreas Müller
|
* Details on ordering are explained at https://dev.e-taxonomy.eu/redmine/issues/3338<BR>
|
40 |
d5d79aea
|
Andreas Müller
|
*
|
41 |
|
|
* @author a.mueller
|
42 |
a88578ce
|
Andreas Müller
|
* @since 02.03.2016
|
43 |
d5d79aea
|
Andreas Müller
|
*/
|
44 |
|
|
public class HomotypicGroupTaxonComparator extends TaxonComparator {
|
45 |
035e8577
|
Andreas Müller
|
|
46 |
d5d79aea
|
Andreas Müller
|
private static final long serialVersionUID = -5088210641256430878L;
|
47 |
23de68fc
|
Andreas Müller
|
private static final Logger logger = LogManager.getLogger(HomotypicGroupTaxonComparator.class);
|
48 |
d5d79aea
|
Andreas Müller
|
|
49 |
|
|
private final TaxonBase<?> firstTaxonInGroup;
|
50 |
9dc896c9
|
Andreas Müller
|
private final TaxonName firstNameInGroup;
|
51 |
d5d79aea
|
Andreas Müller
|
|
52 |
|
|
public HomotypicGroupTaxonComparator(@SuppressWarnings("rawtypes") TaxonBase firstTaxonInGroup) {
|
53 |
82ca4447
|
Andreas Müller
|
super(true);
|
54 |
|
|
this.firstTaxonInGroup = firstTaxonInGroup;
|
55 |
|
|
this.firstNameInGroup = firstTaxonInGroup == null ? null: firstTaxonInGroup.getName();
|
56 |
|
|
}
|
57 |
|
|
|
58 |
|
|
public HomotypicGroupTaxonComparator(@SuppressWarnings("rawtypes") TaxonBase firstTaxonInGroup, boolean includeRanks) {
|
59 |
|
|
super(includeRanks);
|
60 |
d5d79aea
|
Andreas Müller
|
this.firstTaxonInGroup = firstTaxonInGroup;
|
61 |
|
|
this.firstNameInGroup = firstTaxonInGroup == null ? null: firstTaxonInGroup.getName();
|
62 |
|
|
}
|
63 |
|
|
|
64 |
8432293f
|
Andreas Müller
|
public HomotypicGroupTaxonComparator(TaxonName firstNameInGroup, boolean includeRanks) {
|
65 |
|
|
super(includeRanks);
|
66 |
|
|
firstTaxonInGroup = null;
|
67 |
|
|
this.firstNameInGroup = firstNameInGroup;
|
68 |
|
|
}
|
69 |
|
|
|
70 |
d5d79aea
|
Andreas Müller
|
/**
|
71 |
|
|
*
|
72 |
|
|
* @see TaxonComparator#compare(TaxonBase, TaxonBase)
|
73 |
|
|
* @see java.lang.String#compareTo(String)
|
74 |
|
|
* @see java.util.Comparator#compare(java.lang.Object, java.lang.Object)
|
75 |
|
|
*/
|
76 |
|
|
@Override
|
77 |
|
|
public int compare(
|
78 |
|
|
@SuppressWarnings("rawtypes") TaxonBase taxonBase1,
|
79 |
|
|
@SuppressWarnings("rawtypes") TaxonBase taxonBase2) {
|
80 |
|
|
|
81 |
9dc896c9
|
Andreas Müller
|
TaxonName name1 = taxonBase1.getName();
|
82 |
|
|
TaxonName name2 = taxonBase2.getName();
|
83 |
8432293f
|
Andreas Müller
|
|
84 |
|
|
return compareNames(name1, name2, taxonBase1, taxonBase2);
|
85 |
|
|
}
|
86 |
|
|
|
87 |
|
|
public int compareNames(TaxonName name1, TaxonName name2, TaxonBase<?> taxonBase1, TaxonBase<?> taxonBase2) {
|
88 |
1b5c38e1
|
Andreas Müller
|
if (logger.isDebugEnabled()){logger.debug(name1.getTitleCache() +" : "+ name2.getTitleCache());}
|
89 |
8432293f
|
Andreas Müller
|
if (name1 == null && taxonBase1 == null ||
|
90 |
|
|
name2 == null && taxonBase2 == null){
|
91 |
|
|
throw new IllegalArgumentException("There should always be either a name or a taxon to be compared");
|
92 |
|
|
}
|
93 |
d5d79aea
|
Andreas Müller
|
|
94 |
|
|
int compareStatus = compareStatus(name1, name2);
|
95 |
|
|
if (compareStatus != 0){
|
96 |
|
|
return compareStatus;
|
97 |
|
|
}
|
98 |
|
|
|
99 |
|
|
//not same homotypical group -
|
100 |
|
|
//NOTE: this comparator should usually not be used
|
101 |
82ca4447
|
Andreas Müller
|
// for comparing names of different homotypical groups.
|
102 |
d5d79aea
|
Andreas Müller
|
// The following is only to have a defined compare behavior
|
103 |
|
|
// which follows the contract of Comparator#compare.
|
104 |
|
|
if (name1 == null ||
|
105 |
|
|
name2 == null ||
|
106 |
|
|
! name1.getHomotypicalGroup().equals(name2.getHomotypicalGroup())){
|
107 |
|
|
|
108 |
b8c0c883
|
Andreas Müller
|
String compareString1 = name1 != null ?
|
109 |
|
|
name1.getHomotypicalGroup().getUuid().toString() :
|
110 |
|
|
taxonBase1.getUuid().toString();
|
111 |
|
|
String compareString2 = name2 != null ?
|
112 |
|
|
name2.getHomotypicalGroup().getUuid().toString() :
|
113 |
|
|
taxonBase2.getUuid().toString();
|
114 |
|
|
int result = compareString1.compareTo(compareString2);
|
115 |
d5d79aea
|
Andreas Müller
|
return result;
|
116 |
|
|
}
|
117 |
|
|
|
118 |
|
|
//same homotypical group ...
|
119 |
|
|
//one taxon is first in group
|
120 |
8432293f
|
Andreas Müller
|
if (isFirstInGroup(taxonBase1, name1)){
|
121 |
d5d79aea
|
Andreas Müller
|
return -1;
|
122 |
8432293f
|
Andreas Müller
|
}else if (taxonBase2 != null && taxonBase2.equals(firstTaxonInGroup)){
|
123 |
d5d79aea
|
Andreas Müller
|
return 1;
|
124 |
|
|
}
|
125 |
|
|
|
126 |
|
|
//same name => compare on taxon level
|
127 |
|
|
if (name1.equals(name2)){
|
128 |
|
|
return super.compare(taxonBase1, taxonBase2); //if name is the same compare on taxon level
|
129 |
|
|
}
|
130 |
|
|
|
131 |
9dc896c9
|
Andreas Müller
|
TaxonName basionym1 = getPreferredInBasionymGroup(name1);
|
132 |
|
|
TaxonName basionym2 = getPreferredInBasionymGroup(name2);
|
133 |
d5d79aea
|
Andreas Müller
|
|
134 |
|
|
int compareResult;
|
135 |
|
|
if (basionym1.equals(basionym2)){
|
136 |
|
|
//both names belong to same basionym sub-group
|
137 |
|
|
compareResult = handleSameBasionym(basionym1, name1, name2);
|
138 |
|
|
}else{
|
139 |
|
|
compareResult = compareBasionyms(basionym1, basionym2);
|
140 |
|
|
}
|
141 |
|
|
|
142 |
|
|
if (compareResult != 0){
|
143 |
1b5c38e1
|
Andreas Müller
|
// if (logger.isDebugEnabled()){logger.debug(": " + compareResult);}
|
144 |
d5d79aea
|
Andreas Müller
|
return compareResult;
|
145 |
|
|
}else{
|
146 |
|
|
//names are uncomparable on name level (except for uuid, id, etc.)
|
147 |
|
|
int result = super.compare(taxonBase1, taxonBase2);
|
148 |
1b5c38e1
|
Andreas Müller
|
if (logger.isDebugEnabled()){logger.debug(": = " + result);}
|
149 |
d5d79aea
|
Andreas Müller
|
return result;
|
150 |
|
|
}
|
151 |
|
|
}
|
152 |
|
|
|
153 |
8432293f
|
Andreas Müller
|
private boolean isFirstInGroup(TaxonBase<?> taxonBase, TaxonName name) {
|
154 |
|
|
if (taxonBase != null){
|
155 |
|
|
return taxonBase.equals(firstTaxonInGroup);
|
156 |
|
|
}else{
|
157 |
|
|
return name.equals(firstNameInGroup);
|
158 |
|
|
}
|
159 |
|
|
}
|
160 |
c1f24dc3
|
Andreas Müller
|
|
161 |
d5d79aea
|
Andreas Müller
|
/**
|
162 |
|
|
* Compare 2 names which have the same basionym.
|
163 |
|
|
* The names must not be equal to each other but may be equal
|
164 |
|
|
* to the basionym.
|
165 |
882b5f7f
|
Andreas Müller
|
*
|
166 |
d5d79aea
|
Andreas Müller
|
* @param basionym the basionym
|
167 |
|
|
* @param name1 first name to compare
|
168 |
|
|
* @param name2 second name to compare
|
169 |
|
|
* @return compare value according to the {@link Comparator#compare(Object, Object)} contract.
|
170 |
|
|
*/
|
171 |
9dc896c9
|
Andreas Müller
|
private int handleSameBasionym(TaxonName basionym,
|
172 |
|
|
TaxonName name1,
|
173 |
|
|
TaxonName name2) {
|
174 |
d5d79aea
|
Andreas Müller
|
|
175 |
|
|
if (basionym.equals(name1)){
|
176 |
|
|
return -1;
|
177 |
|
|
}else if (basionym.equals(name2)){
|
178 |
|
|
return 1;
|
179 |
|
|
}else{
|
180 |
ee68d8ba
|
Katja Luther
|
return super.compare(name1, name2, false);
|
181 |
d5d79aea
|
Andreas Müller
|
}
|
182 |
ee68d8ba
|
Katja Luther
|
|
183 |
d5d79aea
|
Andreas Müller
|
}
|
184 |
|
|
|
185 |
9dc896c9
|
Andreas Müller
|
private int compareBasionyms(TaxonName basionym1Orig, TaxonName basionym2Orig) {
|
186 |
d5d79aea
|
Andreas Müller
|
//one taxon is first in group
|
187 |
9dc896c9
|
Andreas Müller
|
TaxonName basionym1 = getFirstNameInGroup(basionym1Orig);
|
188 |
|
|
TaxonName basionym2 = getFirstNameInGroup(basionym2Orig);
|
189 |
d5d79aea
|
Andreas Müller
|
|
190 |
1c7946d9
|
Andreas Müller
|
//handle accepted taxon case
|
191 |
d5d79aea
|
Andreas Müller
|
if (basionym1.equals(firstNameInGroup)){
|
192 |
|
|
return -1;
|
193 |
051d0697
|
Andreas Müller
|
}else if (basionym2.equals(firstNameInGroup)){
|
194 |
d5d79aea
|
Andreas Müller
|
return 1;
|
195 |
|
|
}
|
196 |
|
|
|
197 |
1c7946d9
|
Andreas Müller
|
//handle replaced synonyms
|
198 |
1a10708d
|
Andreas Müller
|
boolean basio2IsReplacedSynForBasio1 = getReplacedSynonymClosure(basionym1).contains(basionym2);
|
199 |
|
|
boolean basio1IsReplacedSynForBasio2 = getReplacedSynonymClosure(basionym2).contains(basionym1);
|
200 |
|
|
|
201 |
|
|
if (basio2IsReplacedSynForBasio1 && !basio1IsReplacedSynForBasio2){
|
202 |
d5d79aea
|
Andreas Müller
|
return 1;
|
203 |
c1f24dc3
|
Andreas Müller
|
}else if (basio1IsReplacedSynForBasio2 && !basio2IsReplacedSynForBasio1){
|
204 |
d5d79aea
|
Andreas Müller
|
return -1;
|
205 |
|
|
}
|
206 |
|
|
|
207 |
1c7946d9
|
Andreas Müller
|
//compare by date, nom. illeg., rank and alphabetically
|
208 |
|
|
return super.compare(basionym1, basionym2, true);
|
209 |
d5d79aea
|
Andreas Müller
|
}
|
210 |
|
|
|
211 |
db183545
|
Andreas Müller
|
private TaxonName getFirstNameInGroup(TaxonName basionym) {
|
212 |
d5d79aea
|
Andreas Müller
|
for (NameRelationship nameRel : basionym.getRelationsFromThisName()){
|
213 |
|
|
if (nameRel.getType() != null && nameRel.getType().equals(NameRelationshipType.BASIONYM())){
|
214 |
|
|
if (nameRel.getToName().equals(firstNameInGroup)){
|
215 |
|
|
return firstNameInGroup;
|
216 |
|
|
}
|
217 |
|
|
}
|
218 |
|
|
}
|
219 |
|
|
return basionym;
|
220 |
|
|
}
|
221 |
|
|
|
222 |
db183545
|
Andreas Müller
|
private Set<TaxonName> getReplacedSynonymClosure(TaxonName name) {
|
223 |
9dc896c9
|
Andreas Müller
|
Set<TaxonName> set = name.getReplacedSynonyms();
|
224 |
d5d79aea
|
Andreas Müller
|
if (set.isEmpty()){
|
225 |
|
|
return set;
|
226 |
|
|
}
|
227 |
9dc896c9
|
Andreas Müller
|
Set<TaxonName> result = new HashSet<>();
|
228 |
db183545
|
Andreas Müller
|
for (TaxonName replSyn : set){
|
229 |
d5d79aea
|
Andreas Müller
|
boolean notYetContained = result.add(replSyn);
|
230 |
|
|
if (notYetContained){
|
231 |
|
|
result.addAll(replSyn.getReplacedSynonyms());
|
232 |
|
|
}
|
233 |
|
|
}
|
234 |
|
|
return result;
|
235 |
|
|
}
|
236 |
|
|
|
237 |
db183545
|
Andreas Müller
|
private TaxonName getPreferredInBasionymGroup(TaxonName name) {
|
238 |
|
|
Set<TaxonName> candidates = new HashSet<>();
|
239 |
c1f24dc3
|
Andreas Müller
|
//get all final basionyms, except for those being part of a basionym circle
|
240 |
db183545
|
Andreas Müller
|
for (TaxonName candidate : name.getBasionyms()){
|
241 |
c1f24dc3
|
Andreas Müller
|
if (candidate != null
|
242 |
|
|
&& candidate.getHomotypicalGroup().equals(name.getHomotypicalGroup())
|
243 |
|
|
&& !hasBasionymCircle(candidate, null)){
|
244 |
|
|
candidate = getPreferredInBasionymGroup(candidate);
|
245 |
|
|
candidates.add(candidate);
|
246 |
d5d79aea
|
Andreas Müller
|
}
|
247 |
|
|
}
|
248 |
|
|
|
249 |
|
|
if (candidates.isEmpty()){
|
250 |
|
|
return name;
|
251 |
c1f24dc3
|
Andreas Müller
|
}else if (candidates.size() == 1){
|
252 |
|
|
return candidates.iterator().next();
|
253 |
|
|
}else{
|
254 |
9dc896c9
|
Andreas Müller
|
TaxonName result = candidates.iterator().next();
|
255 |
c1f24dc3
|
Andreas Müller
|
candidates.remove(result);
|
256 |
db183545
|
Andreas Müller
|
for (TaxonName candidate : candidates){
|
257 |
1c7946d9
|
Andreas Müller
|
if (super.compare(result, candidate, false) > 0){
|
258 |
c1f24dc3
|
Andreas Müller
|
result = candidate;
|
259 |
|
|
}
|
260 |
|
|
}
|
261 |
|
|
return result;
|
262 |
|
|
}
|
263 |
|
|
}
|
264 |
|
|
|
265 |
db183545
|
Andreas Müller
|
private boolean hasBasionymCircle(TaxonName name, Set<TaxonName> existing) {
|
266 |
c1f24dc3
|
Andreas Müller
|
if (existing == null){
|
267 |
9dc896c9
|
Andreas Müller
|
existing = new HashSet<>();
|
268 |
c1f24dc3
|
Andreas Müller
|
}
|
269 |
|
|
if (existing.contains(name)){
|
270 |
|
|
return true;
|
271 |
d5d79aea
|
Andreas Müller
|
}else{
|
272 |
9dc896c9
|
Andreas Müller
|
Set<TaxonName> basionyms = name.getBasionyms();
|
273 |
c1f24dc3
|
Andreas Müller
|
if (basionyms.isEmpty()){
|
274 |
|
|
return false;
|
275 |
|
|
}
|
276 |
|
|
existing.add(name);
|
277 |
9dc896c9
|
Andreas Müller
|
for (TaxonName basionym : basionyms){
|
278 |
c1f24dc3
|
Andreas Müller
|
if (hasBasionymCircle(basionym, existing)){
|
279 |
|
|
return true;
|
280 |
|
|
}
|
281 |
|
|
}
|
282 |
|
|
return false;
|
283 |
d5d79aea
|
Andreas Müller
|
}
|
284 |
|
|
}
|
285 |
882b5f7f
|
Andreas Müller
|
}
|