1
|
/**
|
2
|
* Copyright (C) 2018 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.strategy.match;
|
10
|
|
11
|
import eu.etaxonomy.cdm.model.agent.Person;
|
12
|
import eu.etaxonomy.cdm.model.agent.Team;
|
13
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
14
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
15
|
import eu.etaxonomy.cdm.model.reference.ReferenceType;
|
16
|
|
17
|
/**
|
18
|
* @author a.mueller
|
19
|
* @since 17.10.2018
|
20
|
*/
|
21
|
public class MatchStrategyFactory {
|
22
|
|
23
|
public static IMatchStrategyEqual NewDefaultInstance(Class<? extends IMatchable> matchClass){
|
24
|
return new DefaultMatchStrategy(matchClass);
|
25
|
}
|
26
|
|
27
|
public static IParsedMatchStrategy NewParsedInstance(Class<? extends IMatchable> matchClass){
|
28
|
IParsedMatchStrategy parsedMatchStrategy = new ParsedBaseMatchStrategy(matchClass);
|
29
|
return parsedMatchStrategy;
|
30
|
}
|
31
|
|
32
|
public static IParsedMatchStrategy NewParsedPersonInstance(){
|
33
|
try {
|
34
|
IParsedMatchStrategy parsedPersonMatchStrategy = NewParsedInstance(Person.class);
|
35
|
|
36
|
addParsedAgentBaseMatchModes(parsedPersonMatchStrategy);
|
37
|
|
38
|
//FIXME adapt for inRef authors
|
39
|
parsedPersonMatchStrategy.setMatchMode("familyName", MatchMode.EQUAL_OR_FIRST_NULL);
|
40
|
|
41
|
//TODO lifespan may implement MATCH_OR_ONE_NULL
|
42
|
String[] equalOrNullParams = new String[]{"collectorTitle","givenName","initials",
|
43
|
"lifespan","orcid","prefix","suffix"};
|
44
|
for(String param : equalOrNullParams){
|
45
|
parsedPersonMatchStrategy.setMatchMode(param, MatchMode.EQUAL_OR_FIRST_NULL);
|
46
|
}
|
47
|
|
48
|
String[] ignoreParams = new String[]{"institutionalMemberships"};
|
49
|
for(String param : ignoreParams){
|
50
|
parsedPersonMatchStrategy.setMatchMode(param, MatchMode.IGNORE);
|
51
|
}
|
52
|
|
53
|
return parsedPersonMatchStrategy;
|
54
|
} catch (MatchException e) {
|
55
|
throw new RuntimeException("Exception when creating parsed person match strategy.", e);
|
56
|
}
|
57
|
}
|
58
|
|
59
|
public static IParsedMatchStrategy NewParsedTeamInstance(){
|
60
|
IParsedMatchStrategy parsedTeamMatchStrategy = NewParsedInstance(Team.class);
|
61
|
try {
|
62
|
addParsedAgentBaseMatchModes(parsedTeamMatchStrategy);
|
63
|
|
64
|
parsedTeamMatchStrategy.setMatchMode("teamMembers", MatchMode.MATCH, NewParsedPersonInstance());
|
65
|
|
66
|
parsedTeamMatchStrategy.setMatchMode("hasMoreMembers", MatchMode.EQUAL);
|
67
|
|
68
|
//TODO is this correct?
|
69
|
parsedTeamMatchStrategy.setMatchMode("protectedCollectorTitleCache", MatchMode.EQUAL_OR_FIRST_NULL);
|
70
|
parsedTeamMatchStrategy.setMatchMode("protectedNomenclaturalTitleCache", MatchMode.EQUAL);
|
71
|
|
72
|
String[] equalOrNullParams = new String[]{"collectorTitle"};
|
73
|
for(String param : equalOrNullParams){
|
74
|
parsedTeamMatchStrategy.setMatchMode(param, MatchMode.EQUAL_OR_FIRST_NULL);
|
75
|
}
|
76
|
|
77
|
String[] ignoreParams = new String[]{};
|
78
|
for(String param : ignoreParams){
|
79
|
parsedTeamMatchStrategy.setMatchMode(param, MatchMode.IGNORE);
|
80
|
}
|
81
|
|
82
|
return parsedTeamMatchStrategy;
|
83
|
} catch (MatchException e) {
|
84
|
throw new RuntimeException("Exception when creating parsed team match strategy.", e);
|
85
|
}
|
86
|
}
|
87
|
|
88
|
public static IParsedMatchStrategy NewParsedTeamOrPersonInstance(){
|
89
|
@SuppressWarnings("rawtypes")
|
90
|
SubClassMatchStrategy<TeamOrPersonBase> parsedAuthorMatchStrategy = SubClassMatchStrategy.NewInstance(TeamOrPersonBase.class);
|
91
|
try {
|
92
|
parsedAuthorMatchStrategy.putStrategy(Person.class, NewParsedPersonInstance());
|
93
|
parsedAuthorMatchStrategy.putStrategy(Team.class, NewParsedTeamInstance());
|
94
|
return parsedAuthorMatchStrategy;
|
95
|
} catch (MatchException e) {
|
96
|
throw new RuntimeException("Exception when creating parsed author match strategy.", e);
|
97
|
}
|
98
|
}
|
99
|
|
100
|
public static IParsedMatchStrategy NewParsedBookInstance(){
|
101
|
try {
|
102
|
IParsedMatchStrategy parsedBookMatchStrategy = NewParsedInstance(Reference.class);
|
103
|
|
104
|
addParsedReferenceMatchModes(parsedBookMatchStrategy);
|
105
|
|
106
|
//author (as it should always be known for a parsed book)
|
107
|
parsedBookMatchStrategy.setMatchMode("authorship", MatchMode.MATCH_REQUIRED, NewParsedTeamOrPersonInstance());
|
108
|
//in-reference , parsed book has no in-Reference
|
109
|
parsedBookMatchStrategy.setMatchMode("inReference", MatchMode.EQUAL_OR_FIRST_NULL);
|
110
|
parsedBookMatchStrategy.setMatchMode("title", MatchMode.EQUAL_OR_FIRST_NULL); //not available in parsed book
|
111
|
parsedBookMatchStrategy.setMatchMode("abbrevTitle", MatchMode.EQUAL_REQUIRED);
|
112
|
|
113
|
//TODO caches
|
114
|
|
115
|
return parsedBookMatchStrategy;
|
116
|
} catch (MatchException e) {
|
117
|
throw new RuntimeException("Exception when creating parsed team match strategy.", e);
|
118
|
}
|
119
|
}
|
120
|
|
121
|
/**
|
122
|
* As the book section is not clearly defined if parsed (usually only the author and publication date is known),
|
123
|
* it can only match exactly. If pages, title, abbrevTitle are known for the more complete references
|
124
|
* it will match only if they are also known for the parsed instance.
|
125
|
*/
|
126
|
public static IParsedMatchStrategy NewParsedBookSectionInstance(){
|
127
|
try {
|
128
|
IParsedMatchStrategy parsedBookSectionMatchStrategy = (IParsedMatchStrategy)NewDefaultInstance(Reference.class);
|
129
|
|
130
|
addParsedSectionReferenceMatchModes(parsedBookSectionMatchStrategy);
|
131
|
|
132
|
//author (or either title or authorship match, but usually only authorship is known for parsed instances)
|
133
|
parsedBookSectionMatchStrategy.setMatchMode("authorship", MatchMode.MATCH_REQUIRED, NewParsedTeamOrPersonInstance());
|
134
|
//in-reference
|
135
|
parsedBookSectionMatchStrategy.setMatchMode("inReference", MatchMode.MATCH_REQUIRED, NewParsedBookInstance());
|
136
|
//see comment on title and abbrevTitle in #NewParsedArticleInstance
|
137
|
parsedBookSectionMatchStrategy.setMatchMode("title", MatchMode.EQUAL);
|
138
|
parsedBookSectionMatchStrategy.setMatchMode("abbrevTitle", MatchMode.EQUAL);
|
139
|
|
140
|
//TODO caches
|
141
|
|
142
|
return parsedBookSectionMatchStrategy;
|
143
|
|
144
|
} catch (MatchException e) {
|
145
|
throw new RuntimeException("Exception when creating parsed team match strategy.", e);
|
146
|
}
|
147
|
}
|
148
|
|
149
|
public static IParsedMatchStrategy NewParsedArticleInstance(){ //or should it better be Section (in Journal)
|
150
|
try {
|
151
|
IParsedMatchStrategy articleMatchStrategy = (IParsedMatchStrategy)NewDefaultInstance(Reference.class);
|
152
|
|
153
|
addParsedSectionReferenceMatchModes(articleMatchStrategy);
|
154
|
|
155
|
//if a title or abbrevTitle exists for the existing (first) article
|
156
|
//we can not guarantee that the article is really the one that is
|
157
|
//meant in the context, there could be other articles with the same
|
158
|
//author.
|
159
|
//Having a soft matching for possible candidates may solve this issue.
|
160
|
//Also comparing the detail (page) with a filled field "pages" may solve
|
161
|
//it in many cases.
|
162
|
//However, we need to accept that articles as nomenclatural references are not
|
163
|
//clearly enough defined so exact matching with rich data is not possible at
|
164
|
//this point
|
165
|
//TODO improvement, check page number
|
166
|
articleMatchStrategy.setMatchMode("title", MatchMode.EQUAL);
|
167
|
articleMatchStrategy.setMatchMode("abbrevTitle", MatchMode.EQUAL);
|
168
|
|
169
|
articleMatchStrategy.setMatchMode("authorship", MatchMode.MATCH_REQUIRED, NewParsedTeamOrPersonInstance());
|
170
|
articleMatchStrategy.setMatchMode("inReference", MatchMode.MATCH, NewParsedJournalInstance());
|
171
|
|
172
|
//TODO caches
|
173
|
|
174
|
return articleMatchStrategy;
|
175
|
} catch (MatchException e) {
|
176
|
throw new RuntimeException("Problems when creating parsed article match strategy.");
|
177
|
}
|
178
|
}
|
179
|
|
180
|
/**
|
181
|
* Adds all typical parsed reference match modes which are equal for non section references
|
182
|
* (journal, book, (generic(?))).
|
183
|
* The following fields need to be handled explicitly by each calling method:<BR>
|
184
|
*
|
185
|
* <LI>authorship</LI>
|
186
|
* <LI>title</LI>
|
187
|
* <LI>abbrevTitle</LI>
|
188
|
* <LI>titleCache</LI>
|
189
|
* <LI>abbrevTitleCache</LI>
|
190
|
* <LI>protectedTitleCache</LI>
|
191
|
* <LI>protectedAbbrevTitleCache</LI>
|
192
|
* <LI>inReference</LI>
|
193
|
*
|
194
|
* @see #addParsedSectionReferenceMatchModes(IParsedMatchStrategy)
|
195
|
* @param referenceMatchStrategy the strategy to fill
|
196
|
* @throws MatchException
|
197
|
*/
|
198
|
private static void addParsedReferenceMatchModes(IParsedMatchStrategy referenceMatchStrategy) throws MatchException {
|
199
|
|
200
|
//"placePublished" should be MatchMode.EQUAL if parser parses place published
|
201
|
//TODO datePublished could also be more detailed in first then in second, e.g. Apr 2008 <-> 2008,
|
202
|
//something like MatchMode.IncludedIn is needed here
|
203
|
|
204
|
//TODO externally managed
|
205
|
|
206
|
addParsedIdentifiableEntityModes(referenceMatchStrategy);
|
207
|
|
208
|
String[] equalOrNullParams = new String[]{"accessed","doi",
|
209
|
"institution","isbn","issn","organization",
|
210
|
"pages","publisher","placePublished",
|
211
|
"referenceAbstract","school","uri"};
|
212
|
for(String param : equalOrNullParams){
|
213
|
referenceMatchStrategy.setMatchMode(param, MatchMode.EQUAL_OR_FIRST_NULL);
|
214
|
}
|
215
|
|
216
|
String[] equalParams = new String[]{"datePublished","edition",
|
217
|
"editor","seriesPart","volume"};
|
218
|
for(String param : equalParams){
|
219
|
referenceMatchStrategy.setMatchMode(param, MatchMode.EQUAL);
|
220
|
}
|
221
|
String[] matchOrNullParams = new String[]{"institution"};
|
222
|
for(String param : matchOrNullParams){
|
223
|
referenceMatchStrategy.setMatchMode(param, MatchMode.MATCH_OR_FIRST_NULL);
|
224
|
}
|
225
|
}
|
226
|
|
227
|
/**
|
228
|
* Adds all typical parsed reference match modes which are equal for section references
|
229
|
* (article, book section).
|
230
|
* The following fields need to be handled explicitly by each calling method:<BR>
|
231
|
*
|
232
|
* For further information see: #9157
|
233
|
*
|
234
|
* <LI>authorship</LI>
|
235
|
* <LI>title</LI>
|
236
|
* <LI>abbrevTitle</LI>
|
237
|
* <LI>titleCache</LI>
|
238
|
* <LI>abbrevTitleCache</LI>
|
239
|
* <LI>protectedTitleCache</LI>
|
240
|
* <LI>protectedAbbrevTitleCache</LI>
|
241
|
* <LI>inReference</LI>
|
242
|
*
|
243
|
* @see #addParsedReferenceMatchModes(IParsedMatchStrategy)
|
244
|
* @param referenceMatchStrategy the strategy to fill
|
245
|
* @throws MatchException
|
246
|
*/
|
247
|
private static void addParsedSectionReferenceMatchModes(IParsedMatchStrategy referenceMatchStrategy) throws MatchException {
|
248
|
|
249
|
//"placePublished" should be MatchMode.EQUAL if parser parses place published
|
250
|
//TODO datePublished could also be more detailed in first then in second, e.g. Apr 2008 <-> 2008,
|
251
|
//something like MatchMode.IncludedIn is needed here
|
252
|
//TODO pages if a page number is given the parsed reference may not be unidentified anymore
|
253
|
|
254
|
//TODO externally managed
|
255
|
|
256
|
addParsedIdentifiableEntityModes(referenceMatchStrategy);
|
257
|
|
258
|
//here the result differs from addParsedReferenceMatchModes
|
259
|
String[] equalOrNullParams = new String[]{"accessed","doi",
|
260
|
"institution","isbn","issn","organization",
|
261
|
"pages","publisher","placePublished",
|
262
|
"referenceAbstract","school","uri"};
|
263
|
for(String param : equalOrNullParams){
|
264
|
referenceMatchStrategy.setMatchMode(param, MatchMode.EQUAL); //EQUAL !!
|
265
|
}
|
266
|
|
267
|
String[] equalParams = new String[]{"datePublished","edition",
|
268
|
"editor","seriesPart","volume"};
|
269
|
for(String param : equalParams){
|
270
|
referenceMatchStrategy.setMatchMode(param, MatchMode.EQUAL);
|
271
|
}
|
272
|
String[] matchOrNullParams = new String[]{"institution"};
|
273
|
for(String param : matchOrNullParams){
|
274
|
referenceMatchStrategy.setMatchMode(param, MatchMode.MATCH); //MATCH!!
|
275
|
}
|
276
|
}
|
277
|
|
278
|
private static void addParsedAgentBaseMatchModes(IParsedMatchStrategy matchStrategy) throws MatchException {
|
279
|
|
280
|
//FIXME adapt for inRef authors
|
281
|
//TODO how to initialize this cache field in a generic way
|
282
|
matchStrategy.setMatchMode("nomenclaturalTitle", MatchMode.EQUAL_REQUIRED);
|
283
|
|
284
|
addParsedIdentifiableEntityModes(matchStrategy);
|
285
|
|
286
|
//TODO contact does not yet works, also not with EQUAL_OR_ONE_NULL, leads to agent.id=? or agent.id is null query
|
287
|
//better should be even handled with MATCH.Equal_OR_ONE_NULL
|
288
|
String[] equalOrNullParams = new String[]{"collectorTitle"};
|
289
|
for(String param : equalOrNullParams){
|
290
|
matchStrategy.setMatchMode(param, MatchMode.EQUAL_OR_FIRST_NULL);
|
291
|
}
|
292
|
String[] ignoreCollectionParams = new String[]{"media"};
|
293
|
for(String param : ignoreCollectionParams){
|
294
|
matchStrategy.setMatchMode(param, MatchMode.IGNORE);
|
295
|
}
|
296
|
}
|
297
|
|
298
|
private static void addParsedIdentifiableEntityModes(IParsedMatchStrategy matchStrategy) throws MatchException {
|
299
|
|
300
|
addAnnotatableEntityModes(matchStrategy);
|
301
|
|
302
|
//TODO titleCache and protectedTitleCache
|
303
|
|
304
|
//TODO lsid should be handled with MATCH
|
305
|
String[] equalOrNullParams = new String[]{"lsid"};
|
306
|
for(String param : equalOrNullParams){
|
307
|
matchStrategy.setMatchMode(param, MatchMode.EQUAL_OR_FIRST_NULL);
|
308
|
}
|
309
|
String[] ignoreCollectionParams = new String[]{"credits","extensions","identifiers","links","rights"};
|
310
|
for(String param : ignoreCollectionParams){
|
311
|
matchStrategy.setMatchMode(param, MatchMode.IGNORE);
|
312
|
}
|
313
|
}
|
314
|
|
315
|
private static void addAnnotatableEntityModes(IParsedMatchStrategy matchStrategy) throws MatchException {
|
316
|
|
317
|
// //TODO titleCache and protectedTitleCache
|
318
|
//
|
319
|
// //TODO lsid should be handled with MATCH
|
320
|
// String[] equalOrNullParams = new String[]{"lsid"};
|
321
|
// for(String param : equalOrNullParams){
|
322
|
// matchStrategy.setMatchMode(param, MatchMode.EQUAL_OR_ONE_NULL);
|
323
|
// }
|
324
|
String[] ignoreCollectionParams = new String[]{"sources","annotations","markers"};
|
325
|
for(String param : ignoreCollectionParams){
|
326
|
matchStrategy.setMatchMode(param, MatchMode.IGNORE);
|
327
|
}
|
328
|
}
|
329
|
|
330
|
public static IParsedMatchStrategy NewParsedJournalInstance(){
|
331
|
|
332
|
try {
|
333
|
IParsedMatchStrategy parsedJournalMatchStrategy = NewParsedInstance(Reference.class);
|
334
|
addParsedReferenceMatchModes(parsedJournalMatchStrategy);
|
335
|
|
336
|
//journals must not have an author, better MatchMode would be MatchMode.NULL
|
337
|
parsedJournalMatchStrategy.setMatchMode("authorship", MatchMode.MATCH, NewParsedTeamOrPersonInstance());
|
338
|
//in-reference, parsed journal has no in-reference
|
339
|
parsedJournalMatchStrategy.setMatchMode("inReference", MatchMode.EQUAL_OR_FIRST_NULL);
|
340
|
parsedJournalMatchStrategy.setMatchMode("title", MatchMode.EQUAL_OR_FIRST_NULL); //full title not available in parsed journal
|
341
|
parsedJournalMatchStrategy.setMatchMode("abbrevTitle", MatchMode.EQUAL_REQUIRED);
|
342
|
|
343
|
//TODO caches
|
344
|
|
345
|
return parsedJournalMatchStrategy;
|
346
|
} catch (MatchException e) {
|
347
|
throw new RuntimeException("Exception when creating parsed journal match strategy.", e);
|
348
|
}
|
349
|
}
|
350
|
|
351
|
public static IParsedMatchStrategy NewParsedReferenceInstance(){
|
352
|
IParsedMatchStrategy refStrat = ParsedReferenceMatchStrategy.INSTANCE();
|
353
|
return refStrat;
|
354
|
}
|
355
|
|
356
|
/**
|
357
|
* Returns the best matching strategy for a given reference
|
358
|
* with a given reference type
|
359
|
* @return the matching strategy
|
360
|
* @throws NullPointerException if ref is <code>null</code>
|
361
|
*/
|
362
|
public static IMatchStrategy NewParsedReferenceInstance(Reference ref) {
|
363
|
ReferenceType type = ref.getType();
|
364
|
if (type.equals(ReferenceType.Article)){
|
365
|
return NewParsedArticleInstance();
|
366
|
}else if (type.equals(ReferenceType.Book)){
|
367
|
return NewParsedBookInstance();
|
368
|
}else if (type.equals(ReferenceType.BookSection)){
|
369
|
return NewParsedBookSectionInstance();
|
370
|
}else if (type.equals(ReferenceType.Journal)){
|
371
|
return NewParsedJournalInstance();
|
372
|
}else if (type.equals(ReferenceType.Generic)){
|
373
|
//TODO
|
374
|
return NewDefaultInstance(Reference.class);
|
375
|
}else{
|
376
|
//TODO
|
377
|
return NewDefaultInstance(Reference.class);
|
378
|
}
|
379
|
}
|
380
|
|
381
|
// public static IMatchStrategy NewParsedNameInstance(){
|
382
|
// IMatchStrategy result = new DefaultMatchStrategy(TaxonName.class);
|
383
|
// return result;
|
384
|
// }
|
385
|
}
|