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.strategy.cache.reference;
|
10
|
|
11
|
import org.apache.commons.lang3.StringUtils;
|
12
|
import org.apache.log4j.Logger;
|
13
|
|
14
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
15
|
import eu.etaxonomy.cdm.common.UTF8;
|
16
|
import eu.etaxonomy.cdm.model.reference.IJournal;
|
17
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
18
|
import eu.etaxonomy.cdm.model.reference.ReferenceType;
|
19
|
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImplRegExBase;
|
20
|
|
21
|
/**
|
22
|
* @author a.mueller
|
23
|
* @since 25.05.2016
|
24
|
*/
|
25
|
public class TitleWithoutYearAndAuthorHelper {
|
26
|
private static final Logger logger = Logger.getLogger(TitleWithoutYearAndAuthorHelper.class);
|
27
|
|
28
|
//article
|
29
|
private static final String prefixArticleReferenceJounal = UTF8.EN_DASH + " ";
|
30
|
private static final String prefixSeriesArticle = "ser.";
|
31
|
|
32
|
//book
|
33
|
private static final String prefixBookEdition = "ed.";
|
34
|
private static final String prefixBookSeries = ", ser.";
|
35
|
|
36
|
//generic
|
37
|
private static final String prefixEditionGeneric = "ed.";
|
38
|
private static final String prefixSeriesGeneric = "ser.";
|
39
|
|
40
|
//common
|
41
|
private static final String blank = " ";
|
42
|
private static final String comma = ",";
|
43
|
|
44
|
|
45
|
// *************************Main METHODS ***********************************/
|
46
|
|
47
|
public static String getTitleWithoutYearAndAuthor(Reference ref, boolean isAbbrev, boolean isNomRef){
|
48
|
ReferenceType type = ref.getType();
|
49
|
if (! ReferenceDefaultCacheStrategy.isNomRef(type)){
|
50
|
//NOTE: this method is now called always for generating the reference abbrevTitleCache, therefore it is also called for non-nomenclatural reference.
|
51
|
// logger.warn("getTitleWithoutYearAndAuthor should not be required"
|
52
|
// + " for reference type " + type.getLabel() +
|
53
|
// " and does not exist. Use Generic getTitleWithoutYearAndAuthorGeneric instead");
|
54
|
return getTitleWithoutYearAndAuthorGeneric(ref, isAbbrev);
|
55
|
}else if (type == ReferenceType.Article){
|
56
|
return getTitleWithoutYearAndAuthorArticle(ref, isAbbrev, isNomRef);
|
57
|
}else if(type == ReferenceType.Book){
|
58
|
return getTitleWithoutYearAndAuthorBook(ref, isAbbrev);
|
59
|
}else if(type == ReferenceType.CdDvd){
|
60
|
return getTitleWithoutYearAndAuthorCdDvd(ref, isAbbrev);
|
61
|
}else if(type == ReferenceType.Generic){
|
62
|
return getTitleWithoutYearAndAuthorGeneric(ref, isAbbrev);
|
63
|
}else if (type == ReferenceType.WebPage) {
|
64
|
return getTitleWithoutYearAndAuthorWebPage(ref, isAbbrev);
|
65
|
}else if (type == ReferenceType.Thesis) {
|
66
|
return getTitleWithoutYearAndAuthorThesis(ref, isAbbrev);
|
67
|
}else if(type == ReferenceType.BookSection){
|
68
|
return getTitleWithoutYearAndAuthorBookSection(ref, isAbbrev);
|
69
|
}else if (type == ReferenceType.Section || type == ReferenceType.BookSection ){
|
70
|
// not needed in Section
|
71
|
logger.warn("Questionable procedure call. Procedure not implemented because not needed. ");
|
72
|
return null;
|
73
|
}else{
|
74
|
//FIXME
|
75
|
return null;
|
76
|
}
|
77
|
}
|
78
|
|
79
|
private static String getTitleWithoutYearAndAuthorArticle(Reference article, boolean isAbbrev, boolean isNomRef){
|
80
|
if (article == null){
|
81
|
return null;
|
82
|
}
|
83
|
IJournal journal = article.getInReference();
|
84
|
|
85
|
String journalTitel;
|
86
|
if (journal != null){
|
87
|
journalTitel = CdmUtils.getPreferredNonEmptyString(journal.getTitle(), journal.getAbbrevTitle(), isAbbrev, true);
|
88
|
}else{
|
89
|
journalTitel = ReferenceDefaultCacheStrategy.UNDEFINED_JOURNAL;
|
90
|
}
|
91
|
|
92
|
String series = Nz(article.getSeriesPart()).trim();
|
93
|
String volume = Nz(article.getVolume()).trim();
|
94
|
|
95
|
boolean needsComma = false;
|
96
|
|
97
|
//inJournal
|
98
|
String result = isNomRef? "in ": prefixArticleReferenceJounal;
|
99
|
|
100
|
//titelAbbrev
|
101
|
if (isNotBlank(journalTitel)){
|
102
|
result = result + journalTitel;
|
103
|
needsComma = computeNeedsCommaArticle(result, volume, series);
|
104
|
if (! needsComma){
|
105
|
result = result + blank;
|
106
|
}
|
107
|
}
|
108
|
|
109
|
//series and vol.
|
110
|
result = getSeriesAndVolPartArticle(series, volume, needsComma, result);
|
111
|
|
112
|
//delete "."
|
113
|
while (result.endsWith(".")){
|
114
|
result = CdmUtils.removeTrailingDots(result);
|
115
|
}
|
116
|
|
117
|
String articleTitle = CdmUtils.getPreferredNonEmptyString(article.getTitle(), article.getAbbrevTitle(), isAbbrev, true);
|
118
|
if (!isNomRef && isNotBlank(articleTitle)){
|
119
|
articleTitle = CdmUtils.addTrailingDotIfNotExists(articleTitle);
|
120
|
result = articleTitle + " " + result;
|
121
|
}
|
122
|
return result.trim();
|
123
|
}
|
124
|
|
125
|
private static String getTitleWithoutYearAndAuthorBook(Reference ref, boolean isAbbrev){
|
126
|
if (ref == null){
|
127
|
return null;
|
128
|
}
|
129
|
//TODO
|
130
|
String title = CdmUtils.getPreferredNonEmptyString(ref.getTitle(), ref.getAbbrevTitle(), isAbbrev, true);
|
131
|
String edition = Nz(ref.getEdition()).trim();
|
132
|
String series = ""; //TODO ref.getSeries(); //SeriesPart is handled later
|
133
|
String refSeriesPart = Nz(ref.getSeriesPart());
|
134
|
String volume = CdmUtils.Nz(ref.getVolume()).trim();
|
135
|
String refYear = ""; //TODO nomenclaturalReference.getYear();
|
136
|
|
137
|
String nomRefCache = "";
|
138
|
Integer len;
|
139
|
String lastChar = "";
|
140
|
String character =".";
|
141
|
len = title.length();
|
142
|
if (len > 0){
|
143
|
lastChar = title.substring(len-1, len);
|
144
|
}
|
145
|
//lastCharIsDouble = f_core_CompareStrings(RIGHT(@TitelAbbrev,1),character);
|
146
|
boolean lastCharIsDouble = title.equals(character);
|
147
|
|
148
|
if(lastCharIsDouble && edition.length() == 0 && refSeriesPart.length() == 0 && volume.length() == 0 && refYear.length() > 0 ){
|
149
|
title = title.substring(1, len-1); // SUBSTRING(@TitelAbbrev,1,@LEN-1)
|
150
|
}
|
151
|
|
152
|
boolean needsComma = false;
|
153
|
//titelAbbrev
|
154
|
if (!"".equals(title) ){
|
155
|
String postfix = isNotBlank(edition + refSeriesPart) ? "" : blank;
|
156
|
nomRefCache = title + postfix;
|
157
|
}
|
158
|
//edition
|
159
|
String editionPart = "";
|
160
|
if (isNotBlank(edition)){
|
161
|
editionPart = edition;
|
162
|
if (isNumeric(edition)){
|
163
|
editionPart = prefixBookEdition + blank + editionPart;
|
164
|
}
|
165
|
needsComma = true;
|
166
|
}
|
167
|
nomRefCache = CdmUtils.concat(", ", nomRefCache, editionPart);
|
168
|
|
169
|
//inSeries
|
170
|
String seriesPart = "";
|
171
|
if (isNotBlank(refSeriesPart)){
|
172
|
seriesPart = refSeriesPart;
|
173
|
if (isNumeric(refSeriesPart)){
|
174
|
seriesPart = prefixBookSeries + blank + seriesPart;
|
175
|
}else if (needsComma){
|
176
|
seriesPart = comma + seriesPart;
|
177
|
}else if (seriesPart.matches("^\\p{L}.*")){ //
|
178
|
//for characters we expect a space
|
179
|
seriesPart = " " + seriesPart;
|
180
|
}
|
181
|
needsComma = true;
|
182
|
}
|
183
|
nomRefCache += seriesPart;
|
184
|
|
185
|
//volume Part
|
186
|
String volumePart = "";
|
187
|
if (isNotBlank(volume)){
|
188
|
volumePart = volume;
|
189
|
if (needsComma){
|
190
|
volumePart = comma + blank + volumePart;
|
191
|
}else{
|
192
|
volumePart = "" + volumePart;
|
193
|
}
|
194
|
//needsComma = false;
|
195
|
}
|
196
|
nomRefCache += volumePart;
|
197
|
|
198
|
//delete .
|
199
|
nomRefCache = CdmUtils.removeTrailingDots(nomRefCache);
|
200
|
|
201
|
return nomRefCache.trim();
|
202
|
}
|
203
|
|
204
|
private static String getTitleWithoutYearAndAuthorBookSection(Reference ref, boolean isAbbrev){
|
205
|
if (ref == null){
|
206
|
return null;
|
207
|
}
|
208
|
String title = CdmUtils.getPreferredNonEmptyString(ref.getTitle(), ref.getAbbrevTitle(), isAbbrev, true);
|
209
|
title = CdmUtils.removeTrailingDots(title);
|
210
|
|
211
|
return title.trim();
|
212
|
}
|
213
|
|
214
|
public static String getTitleWithoutYearAndAuthorGeneric(Reference genericReference, boolean isAbbrev){
|
215
|
if (genericReference == null){
|
216
|
return null;
|
217
|
}
|
218
|
//TODO
|
219
|
String titel = CdmUtils.getPreferredNonEmptyString(genericReference.getTitle(), genericReference.getAbbrevTitle(), isAbbrev, true);
|
220
|
String edition = CdmUtils.Nz(genericReference.getEdition());
|
221
|
//TODO
|
222
|
String series = CdmUtils.Nz(genericReference.getSeriesPart()).trim(); //nomenclaturalReference.getSeries();
|
223
|
String volume = CdmUtils.Nz(genericReference.getVolume()).trim();
|
224
|
|
225
|
String result = "";
|
226
|
boolean lastCharIsDouble;
|
227
|
Integer len;
|
228
|
String lastChar ="";
|
229
|
String character =".";
|
230
|
len = titel.length();
|
231
|
if (len > 0){lastChar = titel.substring(len-1, len);}
|
232
|
//lastCharIsDouble = f_core_CompareStrings(RIGHT(@TitelAbbrev,1),character);
|
233
|
lastCharIsDouble = titel.equals(character);
|
234
|
|
235
|
// if(lastCharIsDouble && edition.length() == 0 && series.length() == 0 && volume.length() == 0 && refYear.length() > 0 ){
|
236
|
// titelAbbrev = titelAbbrev.substring(1, len-1); // SUBSTRING(@TitelAbbrev,1,@LEN-1)
|
237
|
// }
|
238
|
|
239
|
boolean needsComma = false;
|
240
|
//titelAbbrev
|
241
|
if (titel.length() > 0 ){
|
242
|
String postfix = isNotBlank(edition) ? "" : blank;
|
243
|
result = titel + postfix;
|
244
|
}
|
245
|
//edition
|
246
|
String editionPart = "";
|
247
|
if (isNotBlank(edition)){
|
248
|
editionPart = edition;
|
249
|
if (edition.matches(NonViralNameParserImplRegExBase.pEdition)){
|
250
|
editionPart = prefixEditionGeneric + blank + editionPart;
|
251
|
}
|
252
|
needsComma = true;
|
253
|
}
|
254
|
result = CdmUtils.concat(", ", result, editionPart);
|
255
|
|
256
|
//inSeries
|
257
|
String seriesPart = "";
|
258
|
if (isNotBlank(series)){
|
259
|
seriesPart = series;
|
260
|
if (isNumeric(series)){
|
261
|
seriesPart = prefixSeriesGeneric + blank + seriesPart;
|
262
|
}
|
263
|
if (needsComma){
|
264
|
seriesPart = comma + seriesPart;
|
265
|
}
|
266
|
needsComma = true;
|
267
|
}
|
268
|
result += seriesPart;
|
269
|
|
270
|
|
271
|
//volume Part
|
272
|
String volumePart = "";
|
273
|
if (!"".equals(volume)){
|
274
|
volumePart = volume;
|
275
|
if (needsComma){
|
276
|
volumePart = comma + blank + volumePart;
|
277
|
}
|
278
|
//needsComma = false;
|
279
|
}
|
280
|
result += volumePart;
|
281
|
|
282
|
//delete . //TODO needed? Creates problems e.g. if vol ends with dot, like vol="3, Suppl.", this is not handled correctly
|
283
|
while (result.endsWith(".")){
|
284
|
result = result.substring(0, result.length()-1);
|
285
|
}
|
286
|
|
287
|
return result.trim();
|
288
|
}
|
289
|
|
290
|
private static String getTitleWithoutYearAndAuthorCdDvd(Reference ref, boolean isAbbrev){
|
291
|
if (ref == null){
|
292
|
return null;
|
293
|
}
|
294
|
String nomRefCache = "";
|
295
|
//TODO
|
296
|
String titel = CdmUtils.getPreferredNonEmptyString(ref.getTitle(), ref.getAbbrevTitle(), isAbbrev, true);
|
297
|
// String publisher = CdmUtils.Nz(nomenclaturalReference.getPublisher());
|
298
|
|
299
|
boolean needsComma = false;
|
300
|
//titelAbbrev
|
301
|
if (titel.length() > 0){
|
302
|
nomRefCache = titel + blank;
|
303
|
}
|
304
|
// //publisher
|
305
|
// String publisherPart = "";
|
306
|
// if (!"".equals(publisher)){
|
307
|
// publisherPart = publisher;
|
308
|
// needsComma = true;
|
309
|
// }
|
310
|
// nomRefCache += publisherPart;
|
311
|
|
312
|
//delete .
|
313
|
while (nomRefCache.endsWith(".")){
|
314
|
nomRefCache = nomRefCache.substring(0, nomRefCache.length()-1);
|
315
|
}
|
316
|
return nomRefCache.trim();
|
317
|
}
|
318
|
|
319
|
private static String getTitleWithoutYearAndAuthorThesis(Reference ref, boolean isAbbrev) {
|
320
|
//FIXME this is only a very fast copy and paste from "Generic". Must still be cleaned !
|
321
|
|
322
|
if (ref == null){
|
323
|
return null;
|
324
|
}
|
325
|
|
326
|
//titelAbbrev
|
327
|
//TODO
|
328
|
String titelAbbrev = CdmUtils.getPreferredNonEmptyString(ref.getTitle(), ref.getAbbrevTitle(), isAbbrev, true);
|
329
|
|
330
|
//titelAbbrev
|
331
|
String nomRefCache = titelAbbrev + blank;
|
332
|
|
333
|
//delete .
|
334
|
while (nomRefCache.endsWith(".")){
|
335
|
nomRefCache = nomRefCache.substring(0, nomRefCache.length()-1);
|
336
|
}
|
337
|
|
338
|
return nomRefCache.trim();
|
339
|
}
|
340
|
|
341
|
private static String getTitleWithoutYearAndAuthorWebPage(Reference ref, boolean isAbbrev) {
|
342
|
//FIXME this is only a very fast copy and paste from "Generic". Must still be cleaned !
|
343
|
|
344
|
if (ref == null){
|
345
|
return null;
|
346
|
}
|
347
|
|
348
|
//titleAbbrev
|
349
|
//TODO
|
350
|
String titleAbbrev = CdmUtils.getPreferredNonEmptyString(ref.getTitle(), ref.getAbbrevTitle(), isAbbrev, true);
|
351
|
if (isBlank(titleAbbrev) && ref.getUri() != null){
|
352
|
titleAbbrev = ref.getUri().toString();
|
353
|
}
|
354
|
|
355
|
//titelAbbrev
|
356
|
String nomRefCache = titleAbbrev + blank;
|
357
|
|
358
|
//delete .
|
359
|
while (nomRefCache.endsWith(".")){
|
360
|
nomRefCache = nomRefCache.substring(0, nomRefCache.length()-1);
|
361
|
}
|
362
|
|
363
|
return nomRefCache.trim();
|
364
|
}
|
365
|
|
366
|
//**************************** HELPER ********************************/
|
367
|
|
368
|
public static boolean computeNeedsCommaArticle(String nomRefCache, String volume, String series) {
|
369
|
|
370
|
nomRefCache = nomRefCache.toLowerCase();
|
371
|
int serIndex = nomRefCache.indexOf(" ser. ");
|
372
|
int sectIndex = nomRefCache.indexOf(" sect. ");
|
373
|
int abtIndex = nomRefCache.indexOf(" abt. ");
|
374
|
int index = Math.max(Math.max(serIndex, sectIndex), abtIndex);
|
375
|
int commaIndex = nomRefCache.indexOf(",", index);
|
376
|
if (index > -1 && commaIndex == -1 && isNotBlank(volume)){
|
377
|
return true;
|
378
|
}else if (isNotBlank(series)){
|
379
|
return true;
|
380
|
}else{
|
381
|
return false;
|
382
|
}
|
383
|
}
|
384
|
|
385
|
public static String getSeriesAndVolPartArticle(String series, String volume,
|
386
|
boolean needsComma, String nomRefCache) {
|
387
|
//inSeries
|
388
|
String seriesPart = "";
|
389
|
if (isNotBlank(series)){
|
390
|
seriesPart = series;
|
391
|
if (CdmUtils.isNumeric(series)){
|
392
|
seriesPart = prefixSeriesArticle + blank + seriesPart;
|
393
|
}
|
394
|
// if (needsComma){
|
395
|
seriesPart = comma + blank + seriesPart;
|
396
|
// }
|
397
|
needsComma = true;
|
398
|
}
|
399
|
nomRefCache += seriesPart;
|
400
|
|
401
|
//volume Part
|
402
|
String volumePart = "";
|
403
|
if (!"".equals(volume)){
|
404
|
volumePart = volume;
|
405
|
if (needsComma){
|
406
|
volumePart = comma + blank + volumePart;
|
407
|
}
|
408
|
//needsComma = false;
|
409
|
}
|
410
|
nomRefCache += volumePart;
|
411
|
return nomRefCache;
|
412
|
}
|
413
|
|
414
|
// ****************** COMMON **********************************************/
|
415
|
|
416
|
/**
|
417
|
* Null safe string. Returns the given string if it is not <code>null</code>.
|
418
|
* Empty string otherwise.
|
419
|
* @see CdmUtils#Nz(String)
|
420
|
* @return the null-safe string
|
421
|
*/
|
422
|
private static String Nz(String str){
|
423
|
return CdmUtils.Nz(str);
|
424
|
}
|
425
|
|
426
|
/**
|
427
|
* Checks if a string is not blank.
|
428
|
* @see StringUtils#isNotBlank(String)
|
429
|
*/
|
430
|
private static boolean isNotBlank(String str){
|
431
|
return StringUtils.isNotBlank(str);
|
432
|
}
|
433
|
|
434
|
/**
|
435
|
* Checks if a string is blank.
|
436
|
* @see StringUtils#isNotBlank(String)
|
437
|
*/
|
438
|
private static boolean isBlank(String str){
|
439
|
return StringUtils.isBlank(str);
|
440
|
}
|
441
|
|
442
|
private static boolean isNumeric(String string){
|
443
|
if (string == null){
|
444
|
return false;
|
445
|
}
|
446
|
try {
|
447
|
Double.valueOf(string);
|
448
|
return true;
|
449
|
} catch (NumberFormatException e) {
|
450
|
return false;
|
451
|
}
|
452
|
}
|
453
|
}
|