Revision e6d7b501
Added by Andreas Müller almost 7 years ago
| cdmlib-model/src/main/java/eu/etaxonomy/cdm/strategy/parser/NonViralNameParserImpl.java | ||
|---|---|---|
| 27 | 27 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
| 28 | 28 |
import eu.etaxonomy.cdm.model.common.IParsable; |
| 29 | 29 |
import eu.etaxonomy.cdm.model.common.TimePeriod; |
| 30 |
import eu.etaxonomy.cdm.model.name.BotanicalName; |
|
| 31 |
import eu.etaxonomy.cdm.model.name.CultivarPlantName; |
|
| 32 | 30 |
import eu.etaxonomy.cdm.model.name.HybridRelationship; |
| 33 | 31 |
import eu.etaxonomy.cdm.model.name.HybridRelationshipType; |
| 32 |
import eu.etaxonomy.cdm.model.name.IBotanicalName; |
|
| 33 |
import eu.etaxonomy.cdm.model.name.ICultivarPlantName; |
|
| 34 | 34 |
import eu.etaxonomy.cdm.model.name.INonViralName; |
| 35 | 35 |
import eu.etaxonomy.cdm.model.name.IZoologicalName; |
| 36 | 36 |
import eu.etaxonomy.cdm.model.name.NomenclaturalCode; |
| 37 | 37 |
import eu.etaxonomy.cdm.model.name.NomenclaturalStatus; |
| 38 | 38 |
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType; |
| 39 |
import eu.etaxonomy.cdm.model.name.NonViralName; |
|
| 40 | 39 |
import eu.etaxonomy.cdm.model.name.Rank; |
| 41 | 40 |
import eu.etaxonomy.cdm.model.name.TaxonNameBase; |
| 42 | 41 |
import eu.etaxonomy.cdm.model.name.TaxonNameFactory; |
| 43 |
import eu.etaxonomy.cdm.model.name.ZoologicalName; |
|
| 44 | 42 |
import eu.etaxonomy.cdm.model.reference.IBook; |
| 45 | 43 |
import eu.etaxonomy.cdm.model.reference.IBookSection; |
| 46 | 44 |
import eu.etaxonomy.cdm.model.reference.INomenclaturalReference; |
| ... | ... | |
| 138 | 136 |
} |
| 139 | 137 |
|
| 140 | 138 |
@Override |
| 141 |
public NonViralName parseReferencedName(String fullReferenceString) {
|
|
| 139 |
public TaxonNameBase parseReferencedName(String fullReferenceString) {
|
|
| 142 | 140 |
return parseReferencedName(fullReferenceString, null, null); |
| 143 | 141 |
} |
| 144 | 142 |
|
| 145 | 143 |
@Override |
| 146 |
public NonViralName parseReferencedName(String fullReferenceString, NomenclaturalCode nomCode, Rank rank) {
|
|
| 144 |
public TaxonNameBase parseReferencedName(String fullReferenceString, NomenclaturalCode nomCode, Rank rank) {
|
|
| 147 | 145 |
if (fullReferenceString == null){
|
| 148 | 146 |
return null; |
| 149 | 147 |
}else{
|
| 150 |
INonViralName result = getNonViralNameInstance(fullReferenceString, nomCode, rank);
|
|
| 148 |
INonViralName result = getNonViralNameInstance(fullReferenceString, nomCode, rank);
|
|
| 151 | 149 |
parseReferencedName(result, fullReferenceString, rank, MAKE_EMPTY); |
| 152 |
return (NonViralName)result;
|
|
| 150 |
return TaxonNameBase.castAndDeproxy(result);
|
|
| 153 | 151 |
} |
| 154 | 152 |
} |
| 155 | 153 |
|
| ... | ... | |
| 177 | 175 |
*/ |
| 178 | 176 |
private String getCodeSpecificFullNameRegEx(INonViralName nameToBeFilledOrig){
|
| 179 | 177 |
INonViralName nameToBeFilled = CdmBase.deproxy(nameToBeFilledOrig); |
| 180 |
if (nameToBeFilled instanceof ZoologicalName){
|
|
| 178 |
if (nameToBeFilled.isZoological()){
|
|
| 181 | 179 |
return anyZooFullName; |
| 182 |
}else if (nameToBeFilled instanceof BotanicalName) {
|
|
| 180 |
}else if (nameToBeFilled.isBotanical()) {
|
|
| 183 | 181 |
return anyBotanicFullName; |
| 184 |
}else if (nameToBeFilled instanceof NonViralName) {
|
|
| 182 |
}else if (nameToBeFilled.isNonViral()) {
|
|
| 185 | 183 |
return anyBotanicFullName; //TODO ? |
| 186 | 184 |
}else{
|
| 187 | 185 |
logger.warn("nameToBeFilled class not supported ("+nameToBeFilled.getClass()+")");
|
| ... | ... | |
| 197 | 195 |
private String getCodeSpecificSimpleNameRegEx(INonViralName nameToBeFilled){
|
| 198 | 196 |
nameToBeFilled = CdmBase.deproxy(nameToBeFilled); |
| 199 | 197 |
|
| 200 |
if (nameToBeFilled instanceof ZoologicalName){
|
|
| 198 |
if (nameToBeFilled.isZoological()){
|
|
| 201 | 199 |
return anyZooName; |
| 202 |
}else if (nameToBeFilled instanceof NonViralName){
|
|
| 200 |
}else if (nameToBeFilled.isBotanical()) {
|
|
| 201 |
return anyBotanicName; |
|
| 202 |
}else if (nameToBeFilled.isNonViral()){
|
|
| 203 | 203 |
return anyZooName; //TODO ? |
| 204 |
}else if (nameToBeFilled instanceof BotanicalName) {
|
|
| 205 |
return anyBotanicName; |
|
| 206 | 204 |
}else{
|
| 207 | 205 |
logger.warn("nameToBeFilled class not supported ("+nameToBeFilled.getClass()+")");
|
| 208 | 206 |
return null; |
| ... | ... | |
| 329 | 327 |
nameToBeFilled.setProblemEnds(oldProblemEnds); |
| 330 | 328 |
|
| 331 | 329 |
//zoological new combinations should not have a nom. reference to be parsed |
| 332 |
if (nameToBeFilled.isInstanceOf(ZoologicalName.class)){
|
|
| 333 |
IZoologicalName zooName = CdmBase.deproxy(nameToBeFilled, ZoologicalName.class);
|
|
| 330 |
if (nameToBeFilled.isZoological()){
|
|
| 331 |
IZoologicalName zooName = (IZoologicalName)CdmBase.deproxy(nameToBeFilled);
|
|
| 334 | 332 |
//is name new combination? |
| 335 | 333 |
if (zooName.getBasionymAuthorship() != null || zooName.getOriginalPublicationYear() != null){
|
| 336 | 334 |
ParserProblem parserProblem = ParserProblem.NewCombinationHasPublication; |
| ... | ... | |
| 460 | 458 |
strReference = strReference.substring(0, strReference.length() - yearPart.length()); |
| 461 | 459 |
yearPart = yearPart.replaceFirst(pStart + yearSeperator, "").trim(); |
| 462 | 460 |
}else{
|
| 463 |
if (nameToBeFilled.isInstanceOf(ZoologicalName.class)){
|
|
| 464 |
IZoologicalName zooName = CdmBase.deproxy(nameToBeFilled, ZoologicalName.class);
|
|
| 461 |
if (nameToBeFilled.isZoological()){
|
|
| 462 |
IZoologicalName zooName = (IZoologicalName)CdmBase.deproxy(nameToBeFilled);
|
|
| 465 | 463 |
yearPart = String.valueOf(zooName.getPublicationYear()); |
| 466 | 464 |
//continue |
| 467 | 465 |
}else{
|
| ... | ... | |
| 1141 | 1139 |
*/ |
| 1142 | 1140 |
@Override |
| 1143 | 1141 |
public void parseAuthors(INonViralName nonViralNameOrig, String authorString) throws StringNotParsableException{
|
| 1144 |
INonViralName nonViralName = nonViralNameOrig;
|
|
| 1142 |
INonViralName nonViralName = CdmBase.deproxy(nonViralNameOrig);
|
|
| 1145 | 1143 |
TeamOrPersonBase<?>[] authors = new TeamOrPersonBase[4]; |
| 1146 | 1144 |
Integer[] years = new Integer[4]; |
| 1147 |
Class<? extends INonViralName> clazz = nonViralName.getClass();
|
|
| 1148 |
fullAuthors(authorString, authors, years, clazz);
|
|
| 1145 |
NomenclaturalCode code = nonViralName.getNameType();
|
|
| 1146 |
fullAuthors(authorString, authors, years, code);
|
|
| 1149 | 1147 |
nonViralName.setCombinationAuthorship(authors[0]); |
| 1150 | 1148 |
nonViralName.setExCombinationAuthorship(authors[1]); |
| 1151 | 1149 |
nonViralName.setBasionymAuthorship(authors[2]); |
| 1152 | 1150 |
nonViralName.setExBasionymAuthorship(authors[3]); |
| 1153 |
if (nonViralName instanceof ZoologicalName){
|
|
| 1154 |
IZoologicalName zooName = CdmBase.deproxy(nonViralName, ZoologicalName.class);
|
|
| 1151 |
if (nonViralName.isZoological()){
|
|
| 1152 |
IZoologicalName zooName = (IZoologicalName)nonViralName;
|
|
| 1155 | 1153 |
zooName.setPublicationYear(years[0]); |
| 1156 | 1154 |
zooName.setOriginalPublicationYear(years[2]); |
| 1157 | 1155 |
} |
| ... | ... | |
| 1166 | 1164 |
TeamOrPersonBase<?>[] authors = new TeamOrPersonBase[4]; |
| 1167 | 1165 |
Integer[] years = new Integer[4]; |
| 1168 | 1166 |
try {
|
| 1169 |
Class<? extends INonViralName> clazz = nameToBeFilled.getClass();
|
|
| 1170 |
fullAuthors(authorString, authors, years, clazz);
|
|
| 1167 |
NomenclaturalCode code = nameToBeFilled.getNameType();
|
|
| 1168 |
fullAuthors(authorString, authors, years, code);
|
|
| 1171 | 1169 |
} catch (StringNotParsableException e) {
|
| 1172 | 1170 |
nameToBeFilled.addParsingProblem(ParserProblem.UnparsableAuthorPart); |
| 1173 | 1171 |
nameToBeFilled.setTitleCache(fullNameString, true); |
| 1174 | 1172 |
// FIXME Quick fix, otherwise search would not deliver results for unparsable names |
| 1175 | 1173 |
nameToBeFilled.setNameCache(fullNameString, true); |
| 1176 | 1174 |
// END |
| 1177 |
logger.info("no applicable parsing rule could be found for \"" + fullNameString + "\"");;
|
|
| 1175 |
logger.info("no applicable parsing rule could be found for \"" + fullNameString + "\"");
|
|
| 1178 | 1176 |
} |
| 1179 | 1177 |
nameToBeFilled.setCombinationAuthorship(authors[0]); |
| 1180 | 1178 |
nameToBeFilled.setExCombinationAuthorship(authors[1]); |
| 1181 | 1179 |
nameToBeFilled.setBasionymAuthorship(authors[2]); |
| 1182 | 1180 |
nameToBeFilled.setExBasionymAuthorship(authors[3]); |
| 1183 |
if (nameToBeFilled instanceof ZoologicalName){
|
|
| 1184 |
IZoologicalName zooName = (ZoologicalName)nameToBeFilled; |
|
| 1181 |
if (nameToBeFilled.isZoological()){
|
|
| 1182 |
IZoologicalName zooName = (IZoologicalName)nameToBeFilled;
|
|
| 1185 | 1183 |
zooName.setPublicationYear(years[0]); |
| 1186 | 1184 |
zooName.setOriginalPublicationYear(years[2]); |
| 1187 | 1185 |
} |
| ... | ... | |
| 1194 | 1192 |
*/ |
| 1195 | 1193 |
private Rank guessUninomialRank(INonViralName nameToBeFilled, String uninomial) {
|
| 1196 | 1194 |
Rank result = Rank.GENUS(); |
| 1197 |
if (nameToBeFilled.isInstanceOf(BotanicalName.class)){
|
|
| 1195 |
if (nameToBeFilled.isBotanical()){
|
|
| 1198 | 1196 |
if (false){
|
| 1199 | 1197 |
// |
| 1200 | 1198 |
}else if (uninomial.endsWith("phyta") || uninomial.endsWith("mycota") ){ //plants, fungi
|
| ... | ... | |
| 1222 | 1220 |
}else if (uninomial.endsWith("ota")){
|
| 1223 | 1221 |
result = Rank.KINGDOM(); //TODO |
| 1224 | 1222 |
} |
| 1225 |
}else if (nameToBeFilled.isInstanceOf(ZoologicalName.class)){
|
|
| 1223 |
}else if (nameToBeFilled.isZoological()){
|
|
| 1226 | 1224 |
if (false){
|
| 1227 | 1225 |
// |
| 1228 | 1226 |
}else if (uninomial.endsWith("oideae")){
|
| ... | ... | |
| 1251 | 1249 |
* ExTeam[1], BasionymTeam[2], ExBasionymTeam[3] |
| 1252 | 1250 |
*/ |
| 1253 | 1251 |
protected void fullAuthors (String fullAuthorStringOrig, TeamOrPersonBase<?>[] authors, |
| 1254 |
Integer[] years, Class<? extends INonViralName> clazz)
|
|
| 1252 |
Integer[] years, NomenclaturalCode code)
|
|
| 1255 | 1253 |
throws StringNotParsableException{
|
| 1256 |
if (fullAuthorStringOrig == null || clazz == null){
|
|
| 1254 |
if (fullAuthorStringOrig == null || code == null){
|
|
| 1257 | 1255 |
return; |
| 1258 | 1256 |
} |
| 1259 | 1257 |
String fullAuthorString = fullAuthorStringOrig.trim(); |
| 1260 | 1258 |
|
| 1261 | 1259 |
//Botanic |
| 1262 |
if ( BotanicalName.class.isAssignableFrom(clazz) ){
|
|
| 1260 |
if ( code.isBotanical() ){
|
|
| 1263 | 1261 |
if (! fullBotanicAuthorStringPattern.matcher(fullAuthorString).matches() ){
|
| 1264 | 1262 |
throw new StringNotParsableException("fullAuthorString (" +fullAuthorString+") not parsable: ");
|
| 1265 | 1263 |
} |
| 1266 | 1264 |
} |
| 1267 | 1265 |
//Zoo |
| 1268 |
else if ( ZoologicalName.class.isAssignableFrom(clazz) ){
|
|
| 1266 |
else if ( code.isZoological() ){
|
|
| 1269 | 1267 |
if (! fullZooAuthorStringPattern.matcher(fullAuthorString).matches() ){
|
| 1270 | 1268 |
throw new StringNotParsableException("fullAuthorString (" +fullAuthorString+") not parsable: ");
|
| 1271 | 1269 |
} |
| 1272 | 1270 |
}else {
|
| 1273 | 1271 |
//TODO |
| 1274 |
logger.warn ("Full author String parsable only for defined BotanicalNames or ZoologicalNames but this is " + clazz.getSimpleName());
|
|
| 1272 |
logger.warn ("Full author String parsable only for defined BotanicalNames or ZoologicalNames but this is " + code.getMessage());
|
|
| 1275 | 1273 |
throw new StringNotParsableException("fullAuthorString (" +fullAuthorString+") not parsable: ");
|
| 1276 | 1274 |
} |
| 1277 | 1275 |
fullAuthorsChecked(fullAuthorString, authors, years); |
| ... | ... | |
| 1407 | 1405 |
|
| 1408 | 1406 |
// // Parsing of the given full name that has been identified as a cultivar already somwhere else. |
| 1409 | 1407 |
// // The ... cv. ... syntax is not covered here as it is not according the rules for naming cultivars. |
| 1410 |
public BotanicalName parseCultivar(String fullName) throws StringNotParsableException{
|
|
| 1411 |
CultivarPlantName result = null; |
|
| 1408 |
public IBotanicalName parseCultivar(String fullName) throws StringNotParsableException{
|
|
| 1409 |
ICultivarPlantName result = null;
|
|
| 1412 | 1410 |
String[] words = oWsPattern.split(fullName); |
| 1413 | 1411 |
|
| 1414 | 1412 |
/* --------------------------------------------------------------------------------- |
| ... | ... | |
| 1423 | 1421 |
String cultivarPart = cultivarMatcher.group(0).replace("'","").trim();
|
| 1424 | 1422 |
//OLD: String cultivarPart = cultivarRE.getParen(0).replace("'","").trim();
|
| 1425 | 1423 |
|
| 1426 |
result = (CultivarPlantName)parseFullName(namePart); |
|
| 1424 |
result = (ICultivarPlantName)parseFullName(namePart);
|
|
| 1427 | 1425 |
result.setCultivarName(cultivarPart); |
| 1428 | 1426 |
} |
| 1429 | 1427 |
}else if (fullName.indexOf(" cv.") != 0){
|
| ... | ... | |
| 1449 | 1447 |
//OLD: String namePart = fullName.substring(0, groupRE.getParenStart(0) - 0); |
| 1450 | 1448 |
|
| 1451 | 1449 |
String cultivarPart = words[words.length -1]; |
| 1452 |
result = (CultivarPlantName)parseFullName(namePart); |
|
| 1450 |
result = (ICultivarPlantName)parseFullName(namePart);
|
|
| 1453 | 1451 |
if (result != null){
|
| 1454 | 1452 |
result.setCultivarName(cultivarPart); |
| 1455 | 1453 |
|
| ... | ... | |
| 1468 | 1466 |
} |
| 1469 | 1467 |
|
| 1470 | 1468 |
|
| 1471 |
private void makeEmpty(INonViralName nameToBeFilled){
|
|
| 1469 |
private void makeEmpty(INonViralName name){
|
|
| 1470 |
TaxonNameBase<?,?> nameToBeFilled = TaxonNameBase.castAndDeproxy(name); |
|
| 1472 | 1471 |
nameToBeFilled.setRank(null); |
| 1473 | 1472 |
nameToBeFilled.setTitleCache(null, false); |
| 1474 | 1473 |
nameToBeFilled.setFullTitleCache(null, false); |
| ... | ... | |
| 1502 | 1501 |
nameToBeFilled.setBinomHybrid(false); |
| 1503 | 1502 |
nameToBeFilled.setTrinomHybrid(false); |
| 1504 | 1503 |
|
| 1505 |
if (nameToBeFilled.isInstanceOf(BotanicalName.class)){
|
|
| 1506 |
BotanicalName botanicalName = (BotanicalName)nameToBeFilled; |
|
| 1507 |
botanicalName.setAnamorphic(false); |
|
| 1508 |
} |
|
| 1504 |
nameToBeFilled.setAnamorphic(false); |
|
| 1509 | 1505 |
|
| 1510 |
if (nameToBeFilled.isInstanceOf(ZoologicalName.class)){
|
|
| 1511 |
IZoologicalName zoologicalName = (ZoologicalName)nameToBeFilled; |
|
| 1512 |
zoologicalName.setBreed(null); |
|
| 1513 |
zoologicalName.setOriginalPublicationYear(null); |
|
| 1514 |
} |
|
| 1506 |
nameToBeFilled.setBreed(null); |
|
| 1507 |
nameToBeFilled.setOriginalPublicationYear(null); |
|
| 1515 | 1508 |
|
| 1516 | 1509 |
//nom status handled in nom status parser, otherwise we loose additional information like reference etc. |
| 1517 | 1510 |
//hybrid relationships handled in hybrid formula and at end of fullNameParser |
| 1518 | 1511 |
} |
| 1519 |
|
|
| 1520 |
|
|
| 1521 |
|
|
| 1522 | 1512 |
} |
Also available in: Unified diff
ref #6368 remove TaxonNameBase subclasses