Project

General

Profile

Download (22.2 KB) Statistics
| Branch: | Tag: | Revision:
1
/**
2
* Copyright (C) 2009 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
package eu.etaxonomy.cdm.strategy.parser;
11

    
12
import java.util.regex.Pattern;
13

    
14
import org.apache.log4j.Logger;
15

    
16
import eu.etaxonomy.cdm.common.UTF8;
17

    
18

    
19
/**
20
 * This class is a base class that separates regex parts of the parser from methods
21
 * @author a.mueller
22
 *
23
 */
24
public abstract class NonViralNameParserImplRegExBase  {
25
	@SuppressWarnings("unused")
26
	private static final Logger logger = Logger.getLogger(NonViralNameParserImplRegExBase.class);
27

    
28
	// good intro: http://java.sun.com/docs/books/tutorial/essential/regex/index.html
29

    
30
    //splitter
31
    protected static String epiSplitter = "(\\s+|\\(|\\))"; //( ' '+| '(' | ')' )
32
    protected static Pattern pattern = Pattern.compile(epiSplitter);
33

    
34
	public static final String hybridSign = UTF8.HYBRID.toString();  //  "\u00D7";
35

    
36
    //some useful non-terminals
37
    protected static String pStart = "^";
38
    protected static String end = "$";
39
    protected static String anyEnd = ".*" + end;
40
    protected static String oWs = "\\s+"; //obligatory whitespaces
41
    protected static String fWs = "\\s*"; //facultative whitespcace
42
    protected static String dotSpaceOrBoth = "((?<=\\.)|\\s+)+";
43

    
44
    public static String capitalWord = "\\p{javaUpperCase}\\p{javaLowerCase}*";
45
    protected static String capital2LetterWord = "\\p{javaUpperCase}\\p{javaLowerCase}+";
46
    protected static String nonCapitalWord = "\\p{javaLowerCase}+";
47
    protected static String word = "(" + capitalWord + "|" + nonCapitalWord + ")"; //word (capital or non-capital) with no '.' at the end
48
    protected static String uppercaseWord = "\\p{javaUpperCase}{2,}";
49
    protected static String apostrophWord = word + "('\\p{javaLowerCase}*)?";
50

    
51
    protected static String capitalDotWord = capitalWord + "\\.?"; //capitalWord with facultativ '.' at the end
52
    protected static String capital2charDotWord = "(" + capital2LetterWord + "\\.?|\\p{javaUpperCase}\\.)"; //capitalWord with facultativ '.' but minimum 2 characters (single capital word like 'L' is not allowed
53
    protected static String twoCapitalDotWord = "\\p{javaUpperCase}{2}\\.";   //e.g. NY.
54

    
55
    protected static String nonCapitalDotWord = nonCapitalWord + "\\.?"; //nonCapitalWord with facultativ '.' at the end
56
    protected static String dotWord = "(" + capitalWord + "|" + nonCapitalWord + ")\\.?"; //word (capital or non-capital) with facultativ '.' at the end
57
    protected static String obligateDotWord = "(" + capitalWord + "|" + nonCapitalWord + ")\\.+"; //word (capital or non-capital) with obligate '.' at the end
58

    
59
    //Words used in an epethiton for a TaxonName
60
    protected static String nonCapitalEpiWord = "[a-z\u00EF\u00EB\u00F6\\-]+";   //a-z + diaeresis for ieo
61
    protected static String capitalEpiWord = "[A-Z]"+ nonCapitalEpiWord;
62

    
63

    
64
   //years
65
    protected static String month = "(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)";
66
    protected static String singleYear = "\\b" + "(?:17|18|19|20)" + "\\d{2}" + "\\b";                      // word boundary followed by either 17,18,19, or 20 (not captured) followed by 2 digits
67
    protected static String correctYearPhrase = singleYear + "("+ fWs + "-" + fWs + singleYear + ")?" ;
68
    								//+ "(" + month + ")?)" ;                 // optional month
69
    //!! also used by TimePeriodParser
70
    public static String incorrectYearPhrase = "(\"" + correctYearPhrase + "\"|" + correctYearPhrase + "|"
71
            + UTF8.QUOT_DBL_LOW9 + correctYearPhrase + UTF8.QUOT_DBL_HIGH_REV9 + ")"
72
			+ fWs + "\\[" + singleYear + "\\]"  ;
73
    protected static String yearPhrase = "(" + correctYearPhrase + "|" + incorrectYearPhrase + ")";
74

    
75
    protected static String yearSeperator = "\\." + oWs;
76
    protected static String detailSeparator = ":" + oWs;
77
    protected static String referenceSeparator1 = "," + oWs ;
78
    protected static String inReferenceSeparator = oWs + "in" + oWs;
79
    protected static String referenceSeperator = "(" + referenceSeparator1 +"|" + inReferenceSeparator + ")" ;
80
    protected static String referenceAuthorSeparator = ","+ oWs;
81
    protected static String volumeSeparator = oWs ;
82
    protected static String referenceEnd = "\\.";
83

    
84

    
85
    //status
86
    protected static String status = "";
87

    
88
    //marker
89
    protected static String InfraGenusMarker = "(n|notho)?(subg(en)?\\.|sect\\.|subsect\\.|ser\\.|subser\\.|t\\.infgen\\.|\\[unranked\\])";
90
    protected static String aggrOrGroupMarker = "(aggr\\.|agg\\.|group)";
91
    protected static String infraSpeciesMarkerNoNotho = "(subsp\\.|convar\\.|var\\.|subvar\\.|f\\.|forma|subf\\.|f\\.\\ssp\\.|f\\.spec\\.|f\\.sp\\.|\\[unranked\\]|tax\\." + fWs + "infrasp\\.)";
92
    protected static String infraSpeciesMarker = "(n|notho)?" + infraSpeciesMarkerNoNotho;
93
    protected static String oldInfraSpeciesMarker = "(prol\\.|proles|race|taxon|sublusus)";
94

    
95

    
96
    //AuthorString
97
    protected static String qm = "[" + UTF8.QUOT_SINGLE_RIGHT + UTF8.ACUTE_ACCENT + "'`]";
98
    protected static String authorPart = "(" + "([OdDL]"+qm+"|"+ qm + "t\\s?|ten\\s||l[ae]\\s|zur\\s)?" + "(" + capital2charDotWord + "|DC\\.)I?" + "(" + qm + nonCapitalDotWord + ")?" + "|[vV][ao]n(\\sder)?|da|du|-e|de(n|l|\\sla)?)" ;
99
    protected static String author = "((" + authorPart + "(" + fWs + "|-)" + ")+" + "(f(il)?\\.|secundus|jun\\.|ter|bis)?|Man in "+qm+"t Veld|Sant"+qm+"Anna)" ;
100
    protected static String finalTeamSplitter = "(" + fWs + "(&)" + fWs + "|" + oWs + "et" + oWs + ")";
101
    protected static String notFinalTeamSplitter = "(?:" + fWs + "," + fWs + "|" + finalTeamSplitter + ")";
102
    protected static String authorTeam = fWs + "((?>" + author + notFinalTeamSplitter + ")*" + author + finalTeamSplitter + ")?(?:"  + author + "|al\\.)" +  fWs;
103
    protected static String exString = "(ex\\.?)";
104
    protected static String authorAndExTeam = "(" + authorTeam + oWs + exString + oWs + ")?" + authorTeam;
105
    protected static String basStart = "\\(";
106
    protected static String basEnd = "\\)";
107
    protected static String botanicBasionymAuthor = basStart + "(" + authorAndExTeam + ")" + basEnd;  // '(' and ')' is for evaluation with RE.paren(x)
108
    protected static String fullBotanicAuthorString = fWs + "((" + botanicBasionymAuthor +")?" + fWs + authorAndExTeam + "|" + botanicBasionymAuthor +")"+ fWs;
109
    protected static String facultFullBotanicAuthorString = "(" +  fullBotanicAuthorString + ")?" ;
110

    
111
    //Zoo. Author
112
    //TODO does zoo author have ex-Author?
113
    protected static String zooAuthorYearSeperator = "(,|\\s)";
114
    protected static String zooAuthorAddidtion = fWs + zooAuthorYearSeperator + fWs + singleYear;
115
    protected static String zooAuthorTeam = authorTeam + zooAuthorAddidtion;
116
    protected static String zooBasionymAuthor = basStart + "(" + zooAuthorTeam + ")" + basEnd;
117
    protected static String fullZooAuthorString = fWs + "((" + zooBasionymAuthor +")?" + fWs + zooAuthorTeam + "|" + zooBasionymAuthor +")"+ fWs;
118
    protected static String facultFullZooAuthorString = "(" +  fullZooAuthorString + ")?" ;
119

    
120
    protected static String facultFullAuthorString2 = "(" + facultFullBotanicAuthorString + "|" + facultFullZooAuthorString + ")";
121

    
122
    protected static String basionymAuthor = "(" + botanicBasionymAuthor + "|" + zooBasionymAuthor+ ")";
123
    protected static String fullAuthorString = "(" + fullBotanicAuthorString + "|" + fullZooAuthorString+ ")";
124

    
125
    //details
126
    //TODO still not all parsed
127

    
128
    protected static String nr2 = "\\d{1,2}";
129
    protected static String nr4 = "\\d{1,4}";
130
    protected static String nr5 = "\\d{1,5}";
131

    
132

    
133
    protected static String pPage = nr5 + "[a-zA-Z]?";
134
    protected static String pStrNo = "n\u00B0" + fWs + "(" + nr4 + ")";
135

    
136
    protected static String pBracketNr = "\\[" + nr4 + "\\]";
137
    protected static String pFolBracket = "\\[fol\\." + fWs + "\\d{1,2}(-\\d{1,2})?\\]";  //maybe merge with pTabFigPlate (see below)
138

    
139

    
140
    protected static String pRangeSep = "[-\u2013]";
141
    protected static String pRangeSepCo = "[-\u2013,]";
142

    
143
    protected static String pTabFigPlateStart = "([tT](abs?)?|[fF](igs?)?|[pP]l?s?)(\\.|\\s|$)";   //$ for only 'f'
144
    protected static String pAbcNr = "[a-zA-Z\u00DF]";
145
    protected static String pTabFigPlateNumber = "(" + nr4 + "|" + pAbcNr + "|" + nr4 + fWs + pAbcNr + ")" + "("+ pRangeSepCo + fWs + pAbcNr + ")?";
146
    protected static String pTabFigPlateNumbers = "(" + pTabFigPlateNumber + "(" + pRangeSepCo + fWs + pTabFigPlateNumber + ")?)";
147

    
148
    protected static String pTabFigPlate = pTabFigPlateStart + fWs + pTabFigPlateNumbers + "?";
149
    protected static String pTabFigPl = pTabFigPlate;
150

    
151
    //e.g.: p455; p.455; pp455-456; pp.455-456; pp.455,456; 455, 456; pages 456-457; pages 456,567
152
    protected static String pSinglePages = "(p\\.?)?" + fWs + pPage + "(," + pTabFigPl +"){0,2}";
153
    protected static String pMultiPages = "(pp\\.?|pages)?" + fWs + pPage + fWs + pRangeSepCo +fWs + pPage ;
154
    //static String pPages = pPage + "(," + fWs + "(" + pPage + "|" + pTabFig + ")" + ")?";
155
    protected static String pPages = "(" + pSinglePages +"|" + pMultiPages +")";
156
    protected static String pPagesTabFig = pPages +"([,\\.]" + fWs + pTabFigPl + "){1,2}";
157

    
158

    
159

    
160
    protected static String pCouv = "couv\\." + fWs + "\\d{1,3}";
161

    
162
    protected static String pTabSpecial = "tab\\." + fWs + "(ad" + fWs + "\\d{1,3}|alphab)";
163
    protected static String pPageSpecial = nr4 + fWs + "(in obs|, Expl\\. Tab)";
164
    protected static String pSpecialGardDict = capitalWord + oWs + "n\u00B0" + oWs + "\\d{1,2}";
165
    //TODO
166
    // protected static String pSpecialDetail = "(in err|in tab|sine pag|add\\. & emend|Emend|""\\d{3}"" \\[\\d{3}\\])";
167
 // protected static String pSpecialDetail = "(in err|in tab|sine pag|add\\. & emend|Emend|""\\d{3}"" \\[\\d{3}\\])";
168
    protected static String pSpecialDetail = "(in err|in tab|sine pag|add\\.)";
169

    
170

    
171
//    Const romI = "[Ii]{0,3}"
172
//    	Const romX = "[Xx]{0,3}"
173
//    	Const romC = "[Cc]{0,3}"
174
//    	Const romM = "[Mm]{0,3}"
175
//    ' roman numbers
176
//    ' !! includes empty string: ""
177
//    romOne = "([Vv]?" & romI & or_ & "(IV|iv)" & or_ & "(IX|ix)" & ")"
178
//    romTen = "([Ll]?" & romX & or_ & "(XL|xl)" & or_ & "(XC|xc)" & ")"
179
//    romHun = "([Dd]?" & romC & or_ & "(CD|cd)" & or_ & "(CM|cm)" & ")"
180
//    romNr = "(?=[MDCLXVImdclxvi])(((" & romM & ")?" & romHun & ")?" & romTen & ")?" & romOne
181
    protected static String pRomNr = "ljfweffaflas"; //TODO rom number have to be tested first
182

    
183
//    "(,\\s*" + pTabFigPl + ")?" +
184
    protected static String pDetailAlternatives = "(" + pPages + "|" + pPageSpecial + "|" + pStrNo + "|" + pBracketNr +
185
    			"|" + pTabFigPl + "(,\\s*" + pTabFigPl + ")?" + "|" + pTabSpecial + "|" + pFolBracket + "|" + pCouv + "|" + pRomNr + "|" +
186
    			pSpecialGardDict + "|" + pSpecialDetail + "|" + pPagesTabFig + ")";
187

    
188
    protected static String detail = pDetailAlternatives;
189

    
190
    //reference
191
    protected static String volume = nr4 + "[a-z]?" + fWs + "(\\("+ nr4 + "[a-z]?"  + "([-\u2013]" + nr4 + ")?\\))?" + "(\\(((Suppl|Beibl|App|Beil|Misc|Vorabdr)\\.|Heft\\s*\\d{1,4})\\))?";
192
    //this line caused problem https://dev.e-taxonomy.eu/trac/ticket/1556 in its original form: "([\u005E:\\.]" + fWs + ")";
193
    protected static String anySepChar = "([\u005E:a-zA-Z]" + fWs + ")"; //all characters except for the detail separator, a stricter version would be [,\\-\\&] and some other characters
194
//  protected static String anySepChar = "([,\\-\\&\\.\\+\\']" + fWs + ")";
195

    
196
    protected static int authorSeparatorMaxPosition = 4;  //Author may have a maximum of 4 words
197
    protected static String pTitleWordSeparator = "(\\."+ fWs+"|" + oWs + "|\\.?[-\u2013])";
198
    protected static String pSeriesPart = ",?" + fWs + "[sS]er(\\.)?" + oWs + "\\d{1,2},?";
199

    
200
    protected static String referenceTitleFirstPart = "(" + apostrophWord + pTitleWordSeparator + "|" + twoCapitalDotWord + fWs + ")";
201
    protected static String referenceTitle = referenceTitleFirstPart + "*" + "("+ dotWord + "|" + uppercaseWord + "|" + pSeriesPart + ")";  //reference title may have words seperated by whitespace or dot. The last word may not have a whitespace at the end. There must be at least one word
202
    protected static String referenceTitleWithSepCharacters = "(((" + referenceTitle +"|\\(.+\\))"  + anySepChar + ")*" + referenceTitle + ")"; //,?
203
    //TODO test performance ??
204
    protected static String referenceTitleWithSepCharactersAndBrackets = referenceTitleWithSepCharacters + fWs + "(\\(" + referenceTitleWithSepCharacters + "\\)"+fWs+ ")?(" + referenceTitleWithSepCharacters +")?"  ;
205

    
206
    protected static String referenceTitleWithoutAuthor = "(" + referenceTitleFirstPart + ")" + "{"+ (authorSeparatorMaxPosition -1) +",}" + dotWord +
207
    			anySepChar + referenceTitleWithSepCharactersAndBrackets ;   //separators exist and first separator appears at position authorSeparatorMaxPosition or later
208
    protected static String referenceTitleWithPlaceBracket = referenceTitle + "(" + oWs + "\\(" + capitalWord + "(" + oWs + capitalWord + ")?" + "\\))?" ;
209

    
210
    protected static String editionSeparator = "(" + oWs + "|," + fWs + ")ed\\.?" + oWs;  //
211
    protected static String pEdition = nr2;
212

    
213
    protected static String pVolPart = volumeSeparator +  volume;
214
    protected static String pEditionPart = editionSeparator +  pEdition;
215
    protected static String pEditionVolPart = editionSeparator +  pEdition + fWs + "," + volumeSeparator +  volume;
216
    protected static String pEditionVolAlternative = "(" + pEditionPart + "|" + pVolPart + "|" + pEditionVolPart + ")?";
217

    
218
//    protected static String pVolRefTitle = referenceTitle + "(" + pVolPart + ")?";
219
    protected static String pVolRefTitle = referenceTitleWithPlaceBracket + "(" + pVolPart + ")?";
220
    protected static String softEditionVolRefTitle = referenceTitleWithSepCharactersAndBrackets + pEditionVolAlternative;
221
    protected static String softVolNoAuthorRefTitle = referenceTitleWithoutAuthor + "(" + volumeSeparator +  volume + ")?";
222

    
223
    protected static String pBookReference = softEditionVolRefTitle;
224
    protected static String pBookSectionReference = authorTeam + referenceAuthorSeparator + softEditionVolRefTitle;
225
    protected static String pArticleReference = pVolRefTitle;
226
    protected static String pSoftArticleReference = softVolNoAuthorRefTitle;
227

    
228
    protected static String pReferenceSineDetail = "(" + pArticleReference + "|" + pBookSectionReference + "|" + pBookReference + ")";
229

    
230
    protected static String pReference = pReferenceSineDetail + detailSeparator + detail +
231
					yearSeperator + yearPhrase + "(" + referenceEnd + ")?";
232

    
233
    //static String strictBook = referenc
234

    
235
    protected static Pattern referencePattern = Pattern.compile(pReference);
236
    protected static Pattern referenceSineDetailPattern = Pattern.compile(pReferenceSineDetail);
237

    
238
    protected static String pNomStatusNom =
239
            "nom\\." + fWs + "(ambig\\.|dub\\.|confus\\.|superfl\\.|nud\\.|illeg\\.|inval\\.|cons\\.(\\s*(prop|des)\\.)?|altern(ativ)?\\.|subnud\\.|nov\\.|legit\\.|sanct\\.|valid|"+
240
    			"rej\\.("+ fWs + "prop\\.)?|provis\\.|utique"+fWs+"rej\\.("+fWs+"prop\\.)?|orth\\."+fWs+"cons\\.("+fWs+"prop\\.)?)";
241
    protected static String pNomStatusOrthVar = "orth\\." + fWs + "(var\\.|rej\\.)";
242
    protected static String pNomStatusComb = "comb\\." + fWs + "(inval\\.|illeg\\.|nov\\.)";
243
    protected static String pNomStatusOpus = "opus\\." + fWs + "utique" + fWs + "oppr\\.";
244
    protected static String pNomStatusIned = "ined\\.";
245

    
246

    
247
    protected static String pNomStatus = "(" + pNomStatusNom + "|" + pNomStatusOrthVar + "|" +pNomStatusComb + "|" + pNomStatusOpus + "|" + pNomStatusIned + ")";
248
    protected static String pNomStatusPhrase1 = "," + fWs + pNomStatus;
249
    protected static String pNomStatusPhrase2 = "\\[" + fWs + pNomStatus + "\\]";
250

    
251
    protected static String pNomStatusPhrase = "(?:" + pNomStatusPhrase1 + "|" + pNomStatusPhrase2 + ")";
252

    
253
// Soraya
254
//opus utique oppr.
255
//pro syn.
256
//provisional synonym
257
//fossil name
258

    
259

    
260
    //cultivars and hybrids
261
    protected static String cultivar = oWs + "'..+'"; //Achtung mit Hochkomma in AuthorNamen
262
    protected static String cultivarMarker = oWs + "(cv\\.|')";
263
    protected static String notho = "notho";
264
    protected static String hybridPart = "([xX]" + oWs + "|"+hybridSign+"|"+notho+")";
265
    protected static String noNothoHybridPart = "([xX]" + oWs + "|"+hybridSign+")";
266
    protected static String hybridFull = "(" +oWs +"|"+ pStart +")" + noNothoHybridPart;  //for some reason infraspecific notho ranks do not parse if notho is allowed as uninomial prefix.
267
    protected static String hybridFormularSeparator = oWs + "[" + hybridSign + "xX]" + oWs;
268

    
269

    
270
    //  Name String
271
    protected static String genusOrSupraGenus = "("+hybridFull+")?" + capitalEpiWord;
272
    protected static String infraGenus = capitalEpiWord + oWs + InfraGenusMarker + oWs + capitalEpiWord;
273
    protected static String aggrOrGroup = capitalEpiWord + oWs + nonCapitalEpiWord + oWs + aggrOrGroupMarker;
274
    protected static String spNov = "sp\\.(\\s*nov\\.)?(\\s*\\d{1,2})?";
275
    protected static String specificEpi = "(" + nonCapitalEpiWord + "|" + spNov + ")";
276
    protected static String species = genusOrSupraGenus + oWs + "("+hybridPart+")?" + specificEpi;
277
    protected static String speciesWithInfraGen = genusOrSupraGenus + oWs + "\\(" + capitalEpiWord + "\\)" + oWs + specificEpi;
278

    
279
    protected static String infraSpecies = species + oWs + infraSpeciesMarker + oWs + "("+hybridPart+")?" + nonCapitalEpiWord;
280
    protected static String zooInfraSpecies = species + oWs + "(" + infraSpeciesMarker + oWs +")?" + "("+hybridPart+")?" + nonCapitalEpiWord;
281
    protected static String oldInfraSpecies = capitalEpiWord + oWs +  nonCapitalEpiWord + oWs + oldInfraSpeciesMarker + oWs + nonCapitalEpiWord;
282
    protected static String autonym = capitalEpiWord + oWs + "(" + nonCapitalEpiWord +")" + oWs + fullBotanicAuthorString +  oWs + infraSpeciesMarker + oWs + "\\1";  //2-nd word and last word are the same
283
    //autonym pattern used within anyBotanicalFullName pattern
284
    protected static String autonym2 = capitalEpiWord + oWs + "(" + nonCapitalEpiWord +")" + oWs + fullBotanicAuthorString +  oWs + infraSpeciesMarker + oWs + "\\2";  //2-nd word and last word are the same
285

    
286

    
287
    protected static String anyBotanicName = "(" + genusOrSupraGenus + "|" + infraGenus + "|" + aggrOrGroup + "|" + species + "|" +
288
                    speciesWithInfraGen + "|" + infraSpecies + "|" + oldInfraSpecies + "|" + autonym   + ")+";
289
    protected static String anyZooName = "(" + genusOrSupraGenus + "|" + infraGenus + "|" + aggrOrGroup + "|" + species + "|" +
290
                    speciesWithInfraGen + "|" +zooInfraSpecies + "|" +  oldInfraSpecies + ")+";
291
    protected static String anyBotanicFullName = "(" + autonym2 + "|" + anyBotanicName + oWs + fullBotanicAuthorString + ")"  ;
292
    protected static String anyZooFullName = anyZooName + oWs + fullZooAuthorString ;
293
    protected static String anyFullName = "(" + anyBotanicFullName + "|" + anyZooFullName + ")";
294
    protected static String abbrevHybridGenus = "([A-Z](\\.\\s*|\\s+))";
295
    protected static String abbrevHybridSecondPartWithSpecies = abbrevHybridGenus + "?" + nonCapitalEpiWord + "(" + oWs + infraSpeciesMarkerNoNotho + oWs + nonCapitalEpiWord + ")?";  //#5983 first step but still to strict
296
    protected static String abbrevHybridSecondPartOnlyInfraSpecies = infraSpeciesMarkerNoNotho + oWs + nonCapitalEpiWord;
297
    protected static String abbrevHybridSecondPart = "(" + abbrevHybridSecondPartWithSpecies + "|" + abbrevHybridSecondPartOnlyInfraSpecies + ")";
298

    
299
    protected static String hybridSecondPart = "(" + anyFullName  + "|" +  anyBotanicName + "|" + anyZooName + "|" + abbrevHybridSecondPart + ")";
300
    protected static String hybridFullName = "(" + anyFullName  + "|" +  anyBotanicName + "|" + anyZooName + ")" + hybridFormularSeparator + hybridSecondPart ;
301

    
302
    //Pattern
303
    protected static Pattern oWsPattern = Pattern.compile(oWs);
304
    protected static Pattern finalTeamSplitterPattern = Pattern.compile(finalTeamSplitter);
305
    protected static Pattern cultivarPattern = Pattern.compile(cultivar);
306
    protected static Pattern cultivarMarkerPattern = Pattern.compile(cultivarMarker);
307

    
308
    protected static Pattern genusOrSupraGenusPattern = Pattern.compile(pStart + genusOrSupraGenus + facultFullAuthorString2 + end);
309
    protected static Pattern infraGenusPattern = Pattern.compile(pStart + infraGenus + facultFullAuthorString2 + end);
310
    protected static Pattern aggrOrGroupPattern = Pattern.compile(pStart + aggrOrGroup + fWs + end); //aggr. or group has no author string
311
    protected static Pattern speciesPattern = Pattern.compile(pStart + species + facultFullAuthorString2 + end);
312
    protected static Pattern speciesWithInfraGenPattern = Pattern.compile(pStart + speciesWithInfraGen + facultFullAuthorString2 + end);
313
    protected static Pattern infraSpeciesPattern = Pattern.compile(pStart + infraSpecies + facultFullAuthorString2 + end);
314
    protected static Pattern zooInfraSpeciesPattern = Pattern.compile(pStart + zooInfraSpecies + facultFullAuthorString2 + end);
315
    protected static Pattern oldInfraSpeciesPattern = Pattern.compile(pStart + oldInfraSpecies + facultFullAuthorString2 + end);
316
    protected static Pattern autonymPattern = Pattern.compile(pStart + autonym + fWs + end);
317
    protected static Pattern hybridFormulaPattern = Pattern.compile(pStart + hybridFullName + fWs + end);
318

    
319

    
320
    protected static Pattern botanicBasionymPattern = Pattern.compile(botanicBasionymAuthor);
321
    protected static Pattern zooBasionymPattern = Pattern.compile(zooBasionymAuthor);
322
    protected static Pattern basionymPattern = Pattern.compile(basionymAuthor);
323

    
324
    protected static Pattern zooAuthorPattern = Pattern.compile(zooAuthorTeam);
325
    protected static Pattern zooAuthorAddidtionPattern = Pattern.compile(zooAuthorAddidtion);
326

    
327
    protected static Pattern exAuthorPattern = Pattern.compile(oWs + exString);
328

    
329
    protected static Pattern fullBotanicAuthorStringPattern = Pattern.compile(fullBotanicAuthorString);
330
    protected static Pattern fullZooAuthorStringPattern = Pattern.compile(fullZooAuthorString);
331
    protected static Pattern fullAuthorStringPattern = Pattern.compile(fullAuthorString);
332

    
333
    protected static Pattern anyBotanicFullNamePattern = Pattern.compile(anyBotanicFullName);
334
    protected static Pattern anyZooFullNamePattern = Pattern.compile(anyZooFullName);
335

    
336
    protected static Pattern spNovPattern = Pattern.compile(spNov);
337

    
338

    
339
}
(4-4/7)