cdmlib-model/src/main/java/eu/etaxonomy/cdm/strategy/parser/NonViralNameParserImplRegExBase.java

   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
  17 /**
  18  * This class is a base class that separates regex parts of the parser from methods
  19  * @author a.mueller
  20  *
  21  */
  22 public abstract class NonViralNameParserImplRegExBase  {
  23         @SuppressWarnings("unused")
  24         private static final Logger logger = Logger.getLogger(NonViralNameParserImplRegExBase.class);
  25
  26         // good intro: http://java.sun.com/docs/books/tutorial/essential/regex/index.html
  27
  28     //splitter
  29     protected static String epiSplitter = "(\\s+|\\(|\\))"; //( ' '+| '(' | ')' )
  30     protected static Pattern pattern = Pattern.compile(epiSplitter);
  31
  32         public static final String hybridSign = "\u00D7";
  33
  34     //some useful non-terminals
  35     protected static String pStart = "^";
  36     protected static String end = "$";
  37     protected static String anyEnd = ".*" + end;
  38     protected static String oWs = "\\s+"; //obligatory whitespaces
  39     protected static String fWs = "\\s*"; //facultative whitespcace
  40
  41     protected static String capitalWord = "\\p{javaUpperCase}\\p{javaLowerCase}*";
  42     protected static String nonCapitalWord = "\\p{javaLowerCase}+";
  43     protected static String word = "(" + capitalWord + "|" + nonCapitalWord + ")"; //word (capital or non-capital) with no '.' at the end
  44     protected static String uppercaseWord = "\\p{javaUpperCase}+";
  45
  46     protected static String capitalDotWord = capitalWord + "\\.?"; //capitalWord with facultativ '.' at the end
  47     protected static String nonCapitalDotWord = nonCapitalWord + "\\.?"; //nonCapitalWord with facultativ '.' at the end
  48     protected static String dotWord = "(" + capitalWord + "|" + nonCapitalWord + ")\\.?"; //word (capital or non-capital) with facultativ '.' at the end
  49     protected static String obligateDotWord = "(" + capitalWord + "|" + nonCapitalWord + ")\\.+"; //word (capital or non-capital) with obligate '.' at the end
  50
  51     //Words used in an epethiton for a TaxonName
  52     protected static String nonCapitalEpiWord = "[a-z\u00EF\\-]+";
  53     protected static String capitalEpiWord = "[A-Z]"+ nonCapitalEpiWord;
  54
  55
  56    //years
  57     protected static String month = "(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)";
  58     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
  59     protected static String yearPhrase = singleYear + "("+ fWs + "-" + fWs + singleYear + ")?" ;
  60                                                                 //+ "(" + month + ")?)" ;                 // optional month
  61
  62     //seperator
  63     protected static String yearSeperator = "\\." + oWs;
  64     protected static String detailSeparator = ":" + oWs;
  65     protected static String referenceSeparator1 = "," + oWs ;
  66     protected static String inReferenceSeparator = oWs + "in" + oWs;
  67     protected static String referenceSeperator = "(" + referenceSeparator1 +"|" + inReferenceSeparator + ")" ;
  68     protected static String referenceAuthorSeparator = ","+ oWs;
  69     protected static String volumeSeparator = oWs ;
  70     protected static String referenceEnd = "\\.";
  71
  72
  73     //status
  74     protected static String status = "";
  75
  76     //marker
  77     protected static String InfraGenusMarker = "(subgen.|subg.|sect.|subsect.|ser.|subser.|t.infgen.)";
  78     protected static String aggrOrGroupMarker = "(aggr.|agg.|group)";
  79     protected static String infraSpeciesMarker = "(subsp.|convar.|var.|subvar.|f.|subf.|f.spec.|tax." + fWs + "infrasp.)";
  80     protected static String oldInfraSpeciesMarker = "(prol.|proles|race|taxon|sublusus)";
  81
  82
  83     //AuthorString
  84     protected static String authorPart = "(" + "(d'|D'|L'|'t\\s)?" + capitalDotWord + "('" + nonCapitalDotWord + ")?" + "|da|de(n|l|\\sla)?)" ;
  85     protected static String author = "(" + authorPart + "(" + fWs + "|-)" + ")+" + "(f.|fil.|secundus)?";
  86     protected static String teamSplitter = fWs + "(&)" + fWs;
  87     protected static String authorTeam = fWs + "(" + author + teamSplitter + ")*" + author + "(" + teamSplitter + "al.)?" + fWs;
  88     protected static String exString = "(ex.?)";
  89     protected static String authorAndExTeam = "(" + authorTeam + oWs + exString + oWs + ")?" + authorTeam;
  90     protected static String basStart = "\\(";
  91     protected static String basEnd = "\\)";
  92     protected static String botanicBasionymAuthor = basStart + "(" + authorAndExTeam + ")" + basEnd;  // '(' and ')' is for evaluation with RE.paren(x)
  93     protected static String fullBotanicAuthorString = fWs + "((" + botanicBasionymAuthor +")?" + fWs + authorAndExTeam + "|" + botanicBasionymAuthor +")"+ fWs;
  94     protected static String facultFullBotanicAuthorString = "(" +  fullBotanicAuthorString + ")?" ;
  95
  96     //Zoo. Author
  97     //TODO does zoo author have ex-Author?
  98     protected static String zooAuthorYearSeperator = ",";
  99     protected static String zooAuthorAddidtion = fWs + zooAuthorYearSeperator + fWs + singleYear;
 100     protected static String zooAuthorTeam = authorTeam + zooAuthorAddidtion;
 101     protected static String zooBasionymAuthor = basStart + "(" + zooAuthorTeam + ")" + basEnd;
 102     protected static String fullZooAuthorString = fWs + "((" + zooBasionymAuthor +")?" + fWs + zooAuthorTeam + "|" + zooBasionymAuthor +")"+ fWs;
 103     protected static String facultFullZooAuthorString = "(" +  fullZooAuthorString + ")?" ;
 104
 105     protected static String facultFullAuthorString2 = "(" + facultFullBotanicAuthorString + "|" + facultFullZooAuthorString + ")";
 106
 107     protected static String basionymAuthor = "(" + botanicBasionymAuthor + "|" + zooBasionymAuthor+ ")";
 108     protected static String fullAuthorString = "(" + fullBotanicAuthorString + "|" + fullZooAuthorString+ ")";
 109
 110     //details
 111     //TODO still very simple
 112
 113
 114     protected static String nr2 = "\\d{1,2}";
 115     protected static String nr4 = "\\d{1,4}";
 116     protected static String nr5 = "\\d{1,5}";
 117
 118
 119     protected static String pPage = nr5 + "[a-z]?";
 120     protected static String pStrNo = "n\u00B0" + fWs + "(" + nr4 + ")";
 121
 122     protected static String pBracketNr = "\\[" + nr4 + "\\]";
 123     protected static String pFolBracket = "\\[fol\\." + fWs + "\\d{1,2}(-\\d{1,2})?\\]";
 124
 125     protected static String pStrTab = "tab\\." + fWs + nr4 + "(" + fWs + "(B|\u00DF|\\(\\d{1,3}\\)))?";
 126     protected static String pFig = "fig." + fWs + nr4 + "[a-z]?";
 127     protected static String pFigs = pFig + "(-" + nr4 + ")?";
 128     //static String pTabFig = pStrTab + "(," + fWs + pFigs + ")?";
 129     protected static String pTabFig = "(" + pStrTab + "|" + pFigs + ")";
 130
 131     //e.g.: p455; p.455; pp455-456; pp.455-456; pp.455,456; 455, 456; pages 456-457; pages 456,567
 132     protected static String pSinglePages = "(p\\.?)?" + fWs + pPage + "(," + pTabFig +")?";
 133     protected static String pMultiPages = "(pp\\.?|pages)?" + fWs + pPage + fWs + "(-|,)" +fWs + pPage ;
 134     //static String pPages = pPage + "(," + fWs + "(" + pPage + "|" + pTabFig + ")" + ")?";
 135     protected static String pPages = "(" + pSinglePages +"|" + pMultiPages +")";
 136
 137
 138     protected static String pCouv = "couv\\." + fWs + "\\d{1,3}";
 139
 140     protected static String pTabSpecial = "tab\\." + fWs + "(ad" + fWs + "\\d{1,3}|alphab)";
 141     protected static String pPageSpecial = nr4 + fWs + "(in obs|, Expl\\. Tab)";
 142     protected static String pSpecialGardDict = capitalWord + oWs + "n\u00B0" + oWs + "\\d{1,2}";
 143     //TODO
 144     // protected static String pSpecialDetail = "(in err|in tab|sine pag|add\\. & emend|Emend|""\\d{3}"" \\[\\d{3}\\])";
 145  // protected static String pSpecialDetail = "(in err|in tab|sine pag|add\\. & emend|Emend|""\\d{3}"" \\[\\d{3}\\])";
 146     protected static String pSpecialDetail = "(in err|in tab|sine pag|add\\.)";
 147
 148
 149 //    Const romI = "[Ii]{0,3}"
 150 //      Const romX = "[Xx]{0,3}"
 151 //      Const romC = "[Cc]{0,3}"
 152 //      Const romM = "[Mm]{0,3}"
 153 //    ' roman numbers
 154 //    ' !! includes empty string: ""
 155 //    romOne = "([Vv]?" & romI & or_ & "(IV|iv)" & or_ & "(IX|ix)" & ")"
 156 //    romTen = "([Ll]?" & romX & or_ & "(XL|xl)" & or_ & "(XC|xc)" & ")"
 157 //    romHun = "([Dd]?" & romC & or_ & "(CD|cd)" & or_ & "(CM|cm)" & ")"
 158 //    romNr = "(?=[MDCLXVImdclxvi])(((" & romM & ")?" & romHun & ")?" & romTen & ")?" & romOne
 159     protected static String pRomNr = "ljfweffaflas"; //TODO rom number have to be tested first
 160
 161     protected static String pDetailAlternatives = "(" + pPages + "|" + pPageSpecial + "|" + pStrNo + "|" + pBracketNr +
 162                         "|" + pTabFig + "|" + pTabSpecial + "|" + pFolBracket + "|" + pCouv + "|" + pRomNr + "|" +
 163                         pSpecialGardDict + "|" + pSpecialDetail + ")";
 164
 165     protected static String detail = pDetailAlternatives;
 166
 167     //reference
 168     protected static String volume = nr4 + "(\\("+ nr4  + "\\))?";
 169     //this line caused problem https://dev.e-taxonomy.eu/trac/ticket/1556 in its original form: "([\u005E:\\.]" + fWs + ")";
 170     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
 171 //  protected static String anySepChar = "([,\\-\\&\\.\\+\\']" + fWs + ")";
 172
 173     protected static int authorSeparatorMaxPosition = 4;  //Author may have a maximum of 4 words
 174     protected static String pTitleWordSeparator = "(\\."+ fWs+"|" + oWs + ")";
 175     protected static String pSeriesPart = ",?" + fWs + "[sS]er(\\.)?" + oWs + "\\d{1,2},?";
 176     protected static String referenceTitleFirstPart = "(" + word + pTitleWordSeparator + ")";
 177     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
 178     protected static String referenceTitleWithSepCharacters = "(((" + referenceTitle +"|\\(.+\\))"  + anySepChar + ")*" + referenceTitle + ")"; //,?
 179     //TODO test performance ??
 180     protected static String referenceTitleWithSepCharactersAndBrackets = referenceTitleWithSepCharacters + fWs + "(\\(" + referenceTitleWithSepCharacters + "\\)"+fWs+ ")?(" + referenceTitleWithSepCharacters +")?"  ;
 181
 182     protected static String referenceTitleWithoutAuthor = "(" + referenceTitleFirstPart + ")" + "{"+ (authorSeparatorMaxPosition -1) +",}" + dotWord +
 183                         anySepChar + referenceTitleWithSepCharactersAndBrackets ;   //separators exist and first separator appears at position authorSeparatorMaxPosition or later
 184
 185     protected static String editionSeparator = "(" + oWs + "|," + fWs + ")ed\\.?" + oWs;  //
 186     protected static String pEdition = nr2;
 187
 188     protected static String pVolPart = volumeSeparator +  volume;
 189     protected static String pEditionPart = editionSeparator +  pEdition;
 190     protected static String pEditionVolPart = editionSeparator +  pEdition + fWs + "," + volumeSeparator +  volume;
 191     protected static String pEditionVolAlternative = "(" + pEditionPart + "|" + pVolPart + "|" + pEditionVolPart + ")?";
 192
 193     protected static String pVolRefTitle = referenceTitle + "(" + pVolPart + ")?";
 194     protected static String softEditionVolRefTitle = referenceTitleWithSepCharactersAndBrackets + pEditionVolAlternative;
 195     protected static String softVolNoAuthorRefTitle = referenceTitleWithoutAuthor + "(" + volumeSeparator +  volume + ")?";
 196
 197     protected static String pBookReference = softEditionVolRefTitle;
 198     protected static String pBookSectionReference = authorTeam + referenceAuthorSeparator + softEditionVolRefTitle;
 199     protected static String pArticleReference = pVolRefTitle  ;
 200     protected static String pSoftArticleReference = softVolNoAuthorRefTitle  ;
 201
 202
 203     protected static String pReferenceSineDetail = "(" + pArticleReference + "|" + pBookSectionReference + "|" + pBookReference + ")";
 204
 205
 206     protected static String pReference = pReferenceSineDetail + detailSeparator + detail +
 207                                         yearSeperator + yearPhrase + "(" + referenceEnd + ")?";
 208
 209     //static String strictBook = referenc
 210
 211
 212
 213     protected static Pattern referencePattern = Pattern.compile(pReference);
 214     protected static Pattern referenceSineDetailPattern = Pattern.compile(pReferenceSineDetail);
 215
 216     protected static String pNomStatusNom = "nom\\." + fWs + "(superfl\\.|nud\\.|illeg\\.|inval\\.|cons\\.|alternativ\\.|subnud.|"+
 217                                         "rej\\.|rej\\."+ fWs + "prop\\.|provis\\.)";
 218     protected static String pNomStatusOrthVar = "orth\\." + fWs + "var\\.";
 219     protected static String pNomStatus = "(" + pNomStatusNom + "|" + pNomStatusOrthVar +  ")";
 220     protected static String pNomStatusPhrase1 = "," + fWs + pNomStatus;
 221     protected static String pNomStatusPhrase2 = "\\[" + fWs + pNomStatus + "\\]";
 222
 223     protected static String pNomStatusPhrase = "(?:" + pNomStatusPhrase1 + "|" + pNomStatusPhrase2 + ")";
 224
 225 // Soraya
 226 //opus utique oppr.
 227 //pro syn.
 228 //provisional synonym
 229 //fossil name
 230
 231
 232
 233     //cultivars and hybrids
 234     protected static String cultivar = oWs + "'..+'"; //Achtung mit Hochkomma in AuthorNamen
 235     protected static String cultivarMarker = oWs + "(cv.|')";
 236     protected static String hybridPart = "((x|X)" + oWs + "|"+hybridSign+"|notho)";
 237     protected static String hybridFull = "(" +oWs +"|"+ pStart +")" + hybridPart;
 238
 239
 240     //  Name String
 241     protected static String genusOrSupraGenus = "("+hybridFull+")?" + capitalEpiWord;
 242     protected static String infraGenus = capitalEpiWord + oWs + InfraGenusMarker + oWs + capitalEpiWord;
 243     protected static String aggrOrGroup = capitalEpiWord + oWs + nonCapitalEpiWord + oWs + aggrOrGroupMarker;
 244     protected static String species = genusOrSupraGenus + oWs + "("+hybridPart+")?" + nonCapitalEpiWord;
 245     protected static String infraSpecies = species + oWs + infraSpeciesMarker + oWs + "("+hybridPart+")?" + nonCapitalEpiWord;
 246     protected static String oldInfraSpecies = capitalEpiWord + oWs +  nonCapitalEpiWord + oWs + oldInfraSpeciesMarker + oWs + nonCapitalEpiWord;
 247     protected static String autonym = capitalEpiWord + oWs + "(" + nonCapitalEpiWord +")" + oWs + fullBotanicAuthorString +  oWs + infraSpeciesMarker + oWs + "\\1";  //2-nd word and last word are the same
 248     //autonym pattern used within anyBotanicalFullName pattern
 249     protected static String autonym2 = capitalEpiWord + oWs + "(" + nonCapitalEpiWord +")" + oWs + fullBotanicAuthorString +  oWs + infraSpeciesMarker + oWs + "\\2";  //2-nd word and last word are the same
 250
 251     protected static String anyBotanicName = "(" + genusOrSupraGenus + "|" + infraGenus + "|" + aggrOrGroup + "|" + species + "|" +
 252                                         infraSpecies + "|" + infraSpecies + "|" + oldInfraSpecies + "|" + autonym   + ")+";
 253     protected static String anyZooName = "(" + genusOrSupraGenus + "|" + infraGenus + "|" + aggrOrGroup + "|" + species + "|" +
 254                                         infraSpecies + "|" + infraSpecies + "|" + oldInfraSpecies + ")+";
 255     protected static String anyBotanicFullName = "(" + autonym2 + "|" + anyBotanicName + oWs + fullBotanicAuthorString + ")"  ;
 256     protected static String anyZooFullName = anyZooName + oWs + fullZooAuthorString ;
 257     protected static String anyFullName = "(" + anyBotanicFullName + "|" + anyZooFullName + ")";
 258
 259     //Pattern
 260     protected static Pattern oWsPattern = Pattern.compile(oWs);
 261     protected static Pattern teamSplitterPattern = Pattern.compile(teamSplitter);
 262     protected static Pattern cultivarPattern = Pattern.compile(cultivar);
 263     protected static Pattern cultivarMarkerPattern = Pattern.compile(cultivarMarker);
 264     protected static Pattern hybridPattern = Pattern.compile(hybridFull);
 265
 266     protected static Pattern genusOrSupraGenusPattern = Pattern.compile(pStart + genusOrSupraGenus + facultFullAuthorString2 + end);
 267     protected static Pattern infraGenusPattern = Pattern.compile(pStart + infraGenus + facultFullAuthorString2 + end);
 268     protected static Pattern aggrOrGroupPattern = Pattern.compile(pStart + aggrOrGroup + fWs + end); //aggr. or group has no author string
 269     protected static Pattern speciesPattern = Pattern.compile(pStart + species + facultFullAuthorString2 + end);
 270     protected static Pattern infraSpeciesPattern = Pattern.compile(pStart + infraSpecies + facultFullAuthorString2 + end);
 271     protected static Pattern oldInfraSpeciesPattern = Pattern.compile(pStart + oldInfraSpecies + facultFullAuthorString2 + end);
 272     protected static Pattern autonymPattern = Pattern.compile(pStart + autonym + fWs + end);
 273
 274     protected static Pattern botanicBasionymPattern = Pattern.compile(botanicBasionymAuthor);
 275     protected static Pattern zooBasionymPattern = Pattern.compile(zooBasionymAuthor);
 276     protected static Pattern basionymPattern = Pattern.compile(basionymAuthor);
 277
 278     protected static Pattern zooAuthorPattern = Pattern.compile(zooAuthorTeam);
 279     protected static Pattern zooAuthorAddidtionPattern = Pattern.compile(zooAuthorAddidtion);
 280
 281     protected static Pattern exAuthorPattern = Pattern.compile(oWs + exString);
 282
 283     protected static Pattern fullBotanicAuthorStringPattern = Pattern.compile(fullBotanicAuthorString);
 284     protected static Pattern fullZooAuthorStringPattern = Pattern.compile(fullZooAuthorString);
 285     protected static Pattern fullAuthorStringPattern = Pattern.compile(fullAuthorString);
 286
 287     protected static Pattern anyBotanicFullNamePattern = Pattern.compile(anyBotanicFullName);
 288     protected static Pattern anyZooFullNamePattern = Pattern.compile(anyZooFullName);
 289
 290
 291 }