Revision 484431ef
Added by Katja Luther over 12 years ago
cdmlib-io/src/main/java/eu/etaxonomy/cdm/io/markup/MarkupDocumentImport.java | ||
---|---|---|
1 | 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 |
*/
|
|
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 | 9 |
|
10 | 10 |
package eu.etaxonomy.cdm.io.markup; |
11 | 11 |
|
12 |
import java.net.MalformedURLException; |
|
13 |
import java.net.URL; |
|
12 | 14 |
import java.util.ArrayList; |
15 |
import java.util.Arrays; |
|
13 | 16 |
import java.util.HashMap; |
14 | 17 |
import java.util.HashSet; |
15 | 18 |
import java.util.LinkedList; |
... | ... | |
22 | 25 |
import java.util.regex.Pattern; |
23 | 26 |
|
24 | 27 |
import javax.xml.stream.FactoryConfigurationError; |
28 |
import javax.xml.stream.Location; |
|
25 | 29 |
import javax.xml.stream.XMLEventReader; |
26 | 30 |
import javax.xml.stream.XMLStreamException; |
31 |
import javax.xml.stream.events.Attribute; |
|
32 |
import javax.xml.stream.events.StartElement; |
|
27 | 33 |
import javax.xml.stream.events.XMLEvent; |
28 | 34 |
|
29 |
import org.apache.commons.lang.CharUtils; |
|
30 | 35 |
import org.apache.commons.lang.StringUtils; |
31 | 36 |
import org.apache.log4j.Logger; |
32 |
import org.jdom.Attribute; |
|
33 |
import org.jdom.Element; |
|
37 |
import org.springframework.beans.factory.annotation.Autowired; |
|
38 |
import org.springframework.security.access.PermissionEvaluator; |
|
39 |
import org.springframework.security.authentication.AuthenticationManager; |
|
40 |
import org.springframework.security.core.Authentication; |
|
34 | 41 |
import org.springframework.stereotype.Component; |
42 |
import org.springframework.transaction.TransactionStatus; |
|
35 | 43 |
|
44 |
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade; |
|
45 |
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade.DerivedUnitType; |
|
46 |
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacadeCacheStrategy; |
|
36 | 47 |
import eu.etaxonomy.cdm.common.CdmUtils; |
37 |
import eu.etaxonomy.cdm.common.ResultWrapper;
|
|
38 |
import eu.etaxonomy.cdm.common.XmlHelp;
|
|
48 |
import eu.etaxonomy.cdm.ext.geo.GeoServiceArea;
|
|
49 |
import eu.etaxonomy.cdm.ext.geo.IEditGeoService;
|
|
39 | 50 |
import eu.etaxonomy.cdm.io.common.ICdmIO; |
40 | 51 |
import eu.etaxonomy.cdm.io.common.mapping.UndefinedTransformerMethodException; |
41 | 52 |
import eu.etaxonomy.cdm.io.markup.UnmatchedLeads.UnmatchedLeadsKey; |
42 |
import eu.etaxonomy.cdm.model.agent.Person; |
|
53 |
import eu.etaxonomy.cdm.model.agent.AgentBase; |
|
54 |
import eu.etaxonomy.cdm.model.agent.INomenclaturalAuthor; |
|
43 | 55 |
import eu.etaxonomy.cdm.model.agent.Team; |
44 | 56 |
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase; |
45 | 57 |
import eu.etaxonomy.cdm.model.common.AnnotatableEntity; |
46 | 58 |
import eu.etaxonomy.cdm.model.common.Annotation; |
47 | 59 |
import eu.etaxonomy.cdm.model.common.AnnotationType; |
48 | 60 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
49 |
import eu.etaxonomy.cdm.model.common.Credit;
|
|
61 |
import eu.etaxonomy.cdm.model.common.Extension;
|
|
50 | 62 |
import eu.etaxonomy.cdm.model.common.ExtensionType; |
51 |
import eu.etaxonomy.cdm.model.common.ISourceable; |
|
52 | 63 |
import eu.etaxonomy.cdm.model.common.Language; |
53 |
import eu.etaxonomy.cdm.model.common.Marker; |
|
54 |
import eu.etaxonomy.cdm.model.common.MarkerType; |
|
64 |
import eu.etaxonomy.cdm.model.common.TermVocabulary; |
|
55 | 65 |
import eu.etaxonomy.cdm.model.common.TimePeriod; |
56 |
import eu.etaxonomy.cdm.model.description.CommonTaxonName; |
|
57 | 66 |
import eu.etaxonomy.cdm.model.description.DescriptionElementBase; |
67 |
import eu.etaxonomy.cdm.model.description.Distribution; |
|
58 | 68 |
import eu.etaxonomy.cdm.model.description.Feature; |
59 | 69 |
import eu.etaxonomy.cdm.model.description.KeyStatement; |
60 | 70 |
import eu.etaxonomy.cdm.model.description.PolytomousKey; |
61 | 71 |
import eu.etaxonomy.cdm.model.description.PolytomousKeyNode; |
72 |
import eu.etaxonomy.cdm.model.description.PresenceAbsenceTermBase; |
|
73 |
import eu.etaxonomy.cdm.model.description.PresenceTerm; |
|
62 | 74 |
import eu.etaxonomy.cdm.model.description.TaxonDescription; |
63 | 75 |
import eu.etaxonomy.cdm.model.description.TextData; |
64 |
import eu.etaxonomy.cdm.model.name.BotanicalName; |
|
76 |
import eu.etaxonomy.cdm.model.location.NamedArea; |
|
77 |
import eu.etaxonomy.cdm.model.location.NamedAreaLevel; |
|
78 |
import eu.etaxonomy.cdm.model.location.NamedAreaType; |
|
79 |
import eu.etaxonomy.cdm.model.media.IdentifiableMediaEntity; |
|
80 |
import eu.etaxonomy.cdm.model.media.Media; |
|
81 |
import eu.etaxonomy.cdm.model.name.CultivarPlantName; |
|
65 | 82 |
import eu.etaxonomy.cdm.model.name.HomotypicalGroup; |
66 |
import eu.etaxonomy.cdm.model.name.NameRelationshipType; |
|
67 |
import eu.etaxonomy.cdm.model.name.NameTypeDesignation; |
|
83 |
import eu.etaxonomy.cdm.model.name.NameTypeDesignationStatus; |
|
68 | 84 |
import eu.etaxonomy.cdm.model.name.NomenclaturalCode; |
85 |
import eu.etaxonomy.cdm.model.name.NomenclaturalStatus; |
|
86 |
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType; |
|
69 | 87 |
import eu.etaxonomy.cdm.model.name.NonViralName; |
70 | 88 |
import eu.etaxonomy.cdm.model.name.Rank; |
71 |
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignation; |
|
72 | 89 |
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignationStatus; |
73 | 90 |
import eu.etaxonomy.cdm.model.name.TaxonNameBase; |
74 |
import eu.etaxonomy.cdm.model.name.TypeDesignationBase; |
|
91 |
import eu.etaxonomy.cdm.model.occurrence.Collection; |
|
92 |
import eu.etaxonomy.cdm.model.occurrence.DerivedUnitBase; |
|
93 |
import eu.etaxonomy.cdm.model.occurrence.FieldObservation; |
|
75 | 94 |
import eu.etaxonomy.cdm.model.occurrence.Specimen; |
76 |
import eu.etaxonomy.cdm.model.reference.IBook;
|
|
95 |
import eu.etaxonomy.cdm.model.reference.IArticle;
|
|
77 | 96 |
import eu.etaxonomy.cdm.model.reference.IJournal; |
78 | 97 |
import eu.etaxonomy.cdm.model.reference.Reference; |
79 | 98 |
import eu.etaxonomy.cdm.model.reference.ReferenceFactory; |
... | ... | |
81 | 100 |
import eu.etaxonomy.cdm.model.taxon.Classification; |
82 | 101 |
import eu.etaxonomy.cdm.model.taxon.SynonymRelationshipType; |
83 | 102 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
84 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase; |
|
85 | 103 |
import eu.etaxonomy.cdm.model.taxon.TaxonNode; |
86 | 104 |
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException; |
105 |
import eu.etaxonomy.cdm.strategy.parser.NameTypeParser; |
|
87 | 106 |
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl; |
88 |
|
|
107 |
import eu.etaxonomy.cdm.strategy.parser.SpecimenTypeParser; |
|
108 |
import eu.etaxonomy.cdm.strategy.parser.SpecimenTypeParser.TypeInfo; |
|
89 | 109 |
|
90 | 110 |
/** |
91 | 111 |
* @author a.mueller |
92 |
* |
|
112 |
*
|
|
93 | 113 |
*/ |
94 | 114 |
@Component |
95 | 115 |
public class MarkupDocumentImport extends MarkupImportBase implements ICdmIO<MarkupImportState> { |
96 | 116 |
private static final Logger logger = Logger.getLogger(MarkupDocumentImport.class); |
97 | 117 |
|
98 |
private static int modCount = 30000; |
|
118 |
private static final boolean CREATE_NEW = true; |
|
119 |
private static final boolean IS_IMAGE_GALLERY = true; |
|
120 |
private static final boolean NO_IMAGE_GALLERY = false; |
|
121 |
|
|
122 |
|
|
123 |
private static final String ACCEPTED = "accepted"; |
|
124 |
private static final String ACCEPTED_NAME = "acceptedName"; |
|
125 |
private static final String ADDENDA = "addenda"; |
|
126 |
private static final String ALTERNATEPUBTITLE = "alternatepubtitle"; |
|
127 |
private static final String ALTERNATIVE_COLLECTION_TYPE_STATUS = "alternativeCollectionTypeStatus"; |
|
128 |
private static final String ALTERNATIVE_COLLECTOR = "alternativeCollector"; |
|
129 |
private static final String ALTERNATIVE_FIELD_NUM = "alternativeFieldNum"; |
|
130 |
private static final String ALTITUDE = "altitude"; |
|
131 |
private static final String ANNOTATION = "annotation"; |
|
132 |
private static final String AUTHOR = "author"; |
|
133 |
private static final String BIBLIOGRAPHY = "bibliography"; |
|
134 |
private static final String BIOGRAPHIES = "biographies"; |
|
135 |
private static final String BOLD = "bold"; |
|
136 |
private static final String BR = "br"; |
|
137 |
private static final String CHAR = "char"; |
|
138 |
private static final String CITATION = "citation"; |
|
139 |
private static final String COLLECTION_TYPE_STATUS = "collectionTypeStatus"; |
|
140 |
private static final String COLLECTOR = "collector"; |
|
141 |
private static final String COORDINATES = "coordinates"; |
|
142 |
private static final String COUPLET = "couplet"; |
|
143 |
private static final String DATES = "dates"; |
|
144 |
private static final String DEDICATION = "dedication"; |
|
145 |
private static final String DEFAULT_MEDIA_URL = "defaultMediaUrl"; |
|
146 |
private static final String DESTROYED = "destroyed"; |
|
147 |
private static final String DETAILS = "details"; |
|
148 |
private static final String DISTRIBUTION_LIST = "distributionList"; |
|
149 |
private static final String DISTRIBUTION_LOCALITY = "distributionLocality"; |
|
150 |
private static final String EDITION = "edition"; |
|
151 |
private static final String EDITORS = "editors"; |
|
152 |
private static final String FEATURE = "feature"; |
|
153 |
private static final String FIGURE = "figure"; |
|
154 |
private static final String FIGURE_LEGEND = "figureLegend"; |
|
155 |
private static final String FIGURE_PART = "figurePart"; |
|
156 |
private static final String FIGURE_REF = "figureRef"; |
|
157 |
private static final String FIGURE_TITLE = "figureTitle"; |
|
158 |
private static final String FOOTNOTE = "footnote"; |
|
159 |
private static final String FOOTNOTE_REF = "footnoteRef"; |
|
160 |
private static final String FOOTNOTE_STRING = "footnoteString"; |
|
161 |
private static final String FIELD_NUM = "fieldNum"; |
|
162 |
private static final String FREQUENCY = "frequency"; |
|
163 |
private static final String FULL_NAME = "fullName"; |
|
164 |
private static final String FULL_TYPE = "fullType"; |
|
165 |
private static final String GATHERING = "gathering"; |
|
166 |
private static final String HEADING = "heading"; |
|
167 |
private static final String HABITAT = "habitat"; |
|
168 |
private static final String HABITAT_LIST = "habitatList"; |
|
169 |
private static final String HOMONYM = "homonym"; |
|
170 |
private static final String HOMOTYPES = "homotypes"; |
|
171 |
private static final String ID = "id"; |
|
172 |
private static final String INFRANK = "infrank"; |
|
173 |
private static final String INFRAUT = "infraut"; |
|
174 |
private static final String INFRPARAUT = "infrparaut"; |
|
175 |
private static final String IS_SPOTCHARACTERS = "isSpotcharacters"; |
|
176 |
private static final String ISSUE = "issue"; |
|
177 |
private static final String ITALICS = "italics"; |
|
178 |
private static final String KEY = "key"; |
|
179 |
private static final String KEY_TITLE = "keyTitle"; |
|
180 |
private static final String KEYNOTES = "keynotes"; |
|
181 |
private static final String LIFE_CYCLE_PERIODS = "lifeCyclePeriods"; |
|
182 |
private static final String LOCALITY = "locality"; |
|
183 |
private static final String LOST = "lost"; |
|
184 |
private static final String META_DATA = "metaData"; |
|
185 |
private static final String NAME = "name"; |
|
186 |
private static final String NAME_TYPE = "nameType"; |
|
187 |
private static final String NOM = "nom"; |
|
188 |
private static final String NOMENCLATURE = "nomenclature"; |
|
189 |
private static final String NOT_FOUND = "notFound"; |
|
190 |
private static final String NOT_SEEN = "notSeen"; |
|
191 |
private static final String NOTES = "notes"; |
|
192 |
private static final String NUM = "num"; |
|
193 |
private static final String ORIGINAL_DETERMINATION = "originalDetermination"; |
|
194 |
private static final String PARAUT = "paraut"; |
|
195 |
private static final String PUBFULLNAME = "pubfullname"; |
|
196 |
private static final String PUBLICATION = "publication"; |
|
197 |
private static final String PUBNAME = "pubname"; |
|
198 |
private static final String PUBTITLE = "pubtitle"; |
|
199 |
private static final String PUBTYPE = "pubtype"; |
|
200 |
private static final String QUESTION = "question"; |
|
201 |
private static final String QUOTE = "quote"; |
|
202 |
private static final String RANK = "rank"; |
|
203 |
private static final String REF = "ref"; |
|
204 |
private static final String REF_NUM = "refNum"; |
|
205 |
private static final String REF_PART = "refPart"; |
|
206 |
private static final String REFERENCE = "reference"; |
|
207 |
private static final String REFERENCES = "references"; |
|
208 |
private static final String TAXON = "taxon"; |
|
209 |
private static final String TAXONTITLE = "taxontitle"; |
|
210 |
private static final String TAXONTYPE = "taxontype"; |
|
211 |
private static final String TEXT = "text"; |
|
212 |
private static final String TEXT_SECTION = "textSection"; |
|
213 |
private static final String TO_COUPLET = "toCouplet"; |
|
214 |
private static final String TO_KEY = "toKey"; |
|
215 |
private static final String TO_TAXON = "toTaxon"; |
|
216 |
private static final String TYPE = "type"; |
|
217 |
private static final String TYPE_STATUS = "typeStatus"; |
|
218 |
private static final String TREATMENT = "treatment"; |
|
219 |
private static final String SERIALS_ABBREVIATIONS = "serialsAbbreviations"; |
|
220 |
private static final String SPECIMEN_TYPE = "specimenType"; |
|
221 |
private static final String STATUS = "status"; |
|
222 |
private static final String STRING = "string"; |
|
223 |
private static final String SUB_HEADING = "subHeading"; |
|
224 |
private static final String SUB_COLLECTION = "subCollection"; |
|
225 |
private static final String SYNONYM = "synonym"; |
|
226 |
private static final String UNKNOWN = "unknown"; |
|
227 |
private static final String URL = "url"; |
|
228 |
private static final String USAGE = "usage"; |
|
229 |
private static final String VOLUME = "volume"; |
|
230 |
private static final String WRITER = "writer"; |
|
231 |
private static final String YEAR = "year"; |
|
232 |
|
|
99 | 233 |
private NonViralNameParserImpl parser = new NonViralNameParserImpl(); |
100 |
|
|
101 |
//TODO make part of state, but state is renewed when invoking the import a second time |
|
234 |
|
|
235 |
// TODO make part of state, but state is renewed when invoking the import a |
|
236 |
// second time |
|
102 | 237 |
private UnmatchedLeads unmatchedLeads; |
103 | 238 |
|
239 |
@Autowired |
|
240 |
private IEditGeoService editGeoService; |
|
104 | 241 |
|
105 |
public MarkupDocumentImport(){ |
|
242 |
// TODO remove preliminary |
|
243 |
@Autowired |
|
244 |
private AuthenticationManager authenticationManager; |
|
245 |
private Authentication authentication; |
|
246 |
private PermissionEvaluator permissionEvaluator; |
|
247 |
|
|
248 |
public MarkupDocumentImport() { |
|
106 | 249 |
super(); |
250 |
System.out.println("TODO remove preliminary authentication"); |
|
251 |
// UsernamePasswordAuthenticationToken token = new |
|
252 |
// UsernamePasswordAuthenticationToken("admin", "0000"); |
|
253 |
// authentication = authenticationManager.authenticate(token); |
|
254 |
// SecurityContext context = SecurityContextHolder.getContext(); |
|
255 |
// context.setAuthentication(authentication); |
|
256 |
// permissionEvaluator = new CdmPermissionEvaluator(); |
|
107 | 257 |
} |
108 |
|
|
109 |
|
|
258 |
|
|
110 | 259 |
@Override |
111 |
public boolean doCheck(MarkupImportState state){ |
|
260 |
public boolean doCheck(MarkupImportState state) {
|
|
112 | 261 |
state.setCheck(true); |
113 |
boolean result = doInvoke(state);
|
|
262 |
doInvoke(state); |
|
114 | 263 |
state.setCheck(false); |
115 |
return result;
|
|
264 |
return state.isSuccess();
|
|
116 | 265 |
} |
117 |
|
|
266 |
|
|
118 | 267 |
@Override |
119 |
public boolean doInvoke(MarkupImportState state){
|
|
268 |
public void doInvoke(MarkupImportState state) {
|
|
120 | 269 |
fireProgressEvent("Start import markup document", "Before start of document"); |
121 |
fireWarningEvent("Test a warning", "At start", 17); |
|
122 |
boolean success = false; |
|
123 | 270 |
|
124 | 271 |
Queue<CdmBase> outputStream = new LinkedList<CdmBase>(); |
125 |
|
|
126 |
//FIXME reset state |
|
272 |
|
|
273 |
TransactionStatus tx = startTransaction(); |
|
274 |
// FIXME reset state |
|
127 | 275 |
doAllTheOldOtherStuff(state); |
128 | 276 |
|
129 |
//START |
|
277 |
// START
|
|
130 | 278 |
try { |
131 |
//StAX |
|
132 |
XMLEventReader reader = getStaxReader(state);
|
|
279 |
// StAX
|
|
280 |
XMLEventReader reader = getStaxReader(state); |
|
133 | 281 |
state.setReader(reader); |
134 |
//start document |
|
135 |
if (! validateStartOfDocument(reader)){ |
|
136 |
return false; |
|
282 |
// start document |
|
283 |
if (!validateStartOfDocument(reader)) { |
|
284 |
state.setUnsuccessfull(); |
|
285 |
return; |
|
137 | 286 |
} |
287 |
|
|
288 |
// publication |
|
289 |
String elName = PUBLICATION; |
|
290 |
boolean hasPublication = false; |
|
138 | 291 |
|
139 |
//publication |
|
140 |
String elName = "publication"; |
|
141 | 292 |
while (reader.hasNext()) { |
142 |
if (isStartingElement(reader, elName)){ |
|
143 |
success &= handlePublication(reader, elName); |
|
144 |
}else{ |
|
293 |
XMLEvent nextEvent = reader.nextEvent(); |
|
294 |
if (isStartingElement(nextEvent, elName)) { |
|
295 |
handlePublication(state, reader, nextEvent, elName); |
|
296 |
hasPublication = true; |
|
297 |
} else if (nextEvent.isEndDocument()) { |
|
298 |
if (!hasPublication) { |
|
299 |
String message = "No publication root element found"; |
|
300 |
fireWarningEvent(message, nextEvent, 8); |
|
301 |
} |
|
302 |
// done |
|
303 |
} else { |
|
145 | 304 |
fireSchemaConflictEventExpectedStartTag(elName, reader); |
146 | 305 |
} |
147 | 306 |
} |
148 |
|
|
149 |
// //SAX |
|
150 |
// ImportHandlerBase handler= new PublicationHandler(this); |
|
151 |
// parseSAX(state, handler); |
|
152 |
|
|
307 |
commitTransaction(tx); |
|
308 |
|
|
309 |
// //SAX |
|
310 |
// ImportHandlerBase handler= new PublicationHandler(this); |
|
311 |
// parseSAX(state, handler); |
|
312 |
|
|
153 | 313 |
} catch (FactoryConfigurationError e1) { |
154 |
// TODO Auto-generated catch block |
|
155 | 314 |
fireWarningEvent("Some error occurred while setting up xml factory. Data can't be imported", "Start", 16); |
156 |
success = false;
|
|
315 |
state.setUnsuccessfull();
|
|
157 | 316 |
} catch (XMLStreamException e1) { |
158 |
// TODO Auto-generated catch block |
|
159 | 317 |
fireWarningEvent("An XMLStreamException occurred while parsing. Data can't be imported", "Start", 16); |
160 |
success = false; |
|
161 |
// } catch (ParserConfigurationException e) { |
|
162 |
// fireWarningEvent("A ParserConfigurationException occurred while parsing. Data can't be imported", "Start", 16); |
|
163 |
// } catch (SAXException e) { |
|
164 |
// fireWarningEvent("A SAXException occurred while parsing. Data can't be imported", "Start", 16); |
|
165 |
// } catch (IOException e) { |
|
166 |
// fireWarningEvent("An IO exception occurred while parsing. Data can't be imported", "Start", 16); |
|
318 |
state.setUnsuccessfull(); |
|
319 |
// } catch (ParserConfigurationException e) { |
|
320 |
// fireWarningEvent("A ParserConfigurationException occurred while parsing. Data can't be imported", |
|
321 |
// "Start", 16); |
|
322 |
// } catch (SAXException e) { |
|
323 |
// fireWarningEvent("A SAXException occurred while parsing. Data can't be imported", |
|
324 |
// "Start", 16); |
|
325 |
// } catch (IOException e) { |
|
326 |
// fireWarningEvent("An IO exception occurred while parsing. Data can't be imported", |
|
327 |
// "Start", 16); |
|
167 | 328 |
|
168 | 329 |
} |
169 | 330 |
|
170 |
|
|
171 |
return success; |
|
172 |
|
|
173 |
} |
|
174 |
|
|
175 |
|
|
176 |
private boolean handlePublication(XMLEventReader reader, String elName) throws XMLStreamException { |
|
177 |
boolean success = true; |
|
178 |
XMLEvent event = reader.nextEvent(); |
|
179 |
if (isEndingElement(event, elName)){ |
|
180 |
//TODO cardinality of alternative |
|
181 |
return success; |
|
182 |
}else if(isStartingElement(event, "metaData")){ |
|
183 |
|
|
184 |
}else if(isStartingElement(event, "treatment")){ |
|
185 |
|
|
186 |
}else if(isStartingElement(event, "biographies")){ |
|
187 |
|
|
188 |
}else if(isStartingElement(event, "references")){ |
|
189 |
|
|
190 |
}else if(isStartingElement(event, "metaData")){ |
|
191 |
|
|
192 |
}else if(isStartingElement(event, "metaData")){ |
|
193 |
|
|
194 |
}else if(isStartingElement(event, "metaData")){ |
|
195 |
|
|
196 |
} |
|
197 |
return success; |
|
198 |
} |
|
199 |
|
|
331 |
return; |
|
200 | 332 |
|
201 |
|
|
202 |
/** |
|
203 |
* This comes from the old version, needs to be checked on need |
|
204 |
* @param state |
|
205 |
*/ |
|
206 |
private void doAllTheOldOtherStuff(MarkupImportState state) { |
|
207 |
state.putTree(null, null); |
|
208 |
if (unmatchedLeads == null){ |
|
209 |
unmatchedLeads = UnmatchedLeads.NewInstance(); |
|
210 |
} |
|
211 |
state.setUnmatchedLeads(unmatchedLeads); |
|
212 |
|
|
213 |
// TransactionStatus tx = startTransaction(); |
|
214 |
unmatchedLeads.saveToSession(getPolytomousKeyNodeService()); |
|
215 |
|
|
216 |
|
|
217 |
//TODO generally do not store the reference object in the config |
|
218 |
Reference sourceReference = state.getConfig().getSourceReference(); |
|
219 |
getReferenceService().saveOrUpdate(sourceReference); |
|
220 | 333 |
} |
221 | 334 |
|
335 |
private void handlePublication(MarkupImportState state, |
|
336 |
XMLEventReader reader, XMLEvent currentEvent, String elName) |
|
337 |
throws XMLStreamException { |
|
222 | 338 |
|
223 |
private boolean doInvoke_old(MarkupImportState state){ |
|
224 |
Set<TaxonBase> taxaToSave = new HashSet<TaxonBase>(); |
|
225 |
ResultWrapper<Boolean> success = ResultWrapper.NewInstance(true); |
|
226 |
|
|
227 |
// Element elbody= getBodyElement(state.getConfig()); |
|
228 |
Element elbody = null; |
|
229 |
List<Element> elTaxonList = elbody.getChildren(); |
|
230 |
|
|
231 |
int i = 0; |
|
232 |
|
|
233 |
Set<String> unhandledTitleClassess = new HashSet<String>(); |
|
234 |
Set<String> unhandledNomeclatureChildren = new HashSet<String>(); |
|
235 |
Set<String> unhandledDescriptionChildren = new HashSet<String>(); |
|
236 |
|
|
237 |
Taxon lastTaxon = getLastTaxon(state); |
|
238 |
|
|
239 |
//for each taxon |
|
240 |
for (Element elTaxon : elTaxonList){ |
|
339 |
// attributes |
|
340 |
StartElement element = currentEvent.asStartElement(); |
|
341 |
Map<String, Attribute> attributes = getAttributes(element); |
|
342 |
handleUnexpectedAttributes(element.getLocation(), attributes, |
|
343 |
"noNamespaceSchemaLocation"); |
|
344 |
|
|
345 |
while (reader.hasNext()) { |
|
346 |
XMLEvent event = readNoWhitespace(reader); |
|
347 |
// TODO cardinality of alternative |
|
348 |
if (event.isEndElement()) { |
|
349 |
if (isEndingElement(event, elName)) { |
|
350 |
return; |
|
351 |
} else { |
|
352 |
if (isEndingElement(event, BIOGRAPHIES)) { |
|
353 |
// NOT YET IMPLEMENTED |
|
354 |
popUnimplemented(event.asEndElement()); |
|
355 |
} else if (isEndingElement(event, REFERENCES)) { |
|
356 |
// NOT YET IMPLEMENTED |
|
357 |
popUnimplemented(event.asEndElement()); |
|
358 |
} else if (isEndingElement(event, TEXT_SECTION)) { |
|
359 |
// NOT YET IMPLEMENTED |
|
360 |
popUnimplemented(event.asEndElement()); |
|
361 |
} else if (isEndingElement(event, ADDENDA)) { |
|
362 |
// NOT YET IMPLEMENTED |
|
363 |
popUnimplemented(event.asEndElement()); |
|
364 |
} else { |
|
365 |
handleUnexpectedElement(event); |
|
366 |
} |
|
367 |
} |
|
368 |
} else if (event.isStartElement()) { |
|
369 |
if (isStartingElement(event, META_DATA)) { |
|
370 |
handleMetaData(state, reader, event); |
|
371 |
} else if (isStartingElement(event, TREATMENT)) { |
|
372 |
handleTreatment(state, reader, event); |
|
373 |
} else if (isStartingElement(event, BIOGRAPHIES)) { |
|
374 |
handleNotYetImplementedElement(event); |
|
375 |
} else if (isStartingElement(event, REFERENCES)) { |
|
376 |
handleNotYetImplementedElement(event); |
|
377 |
} else if (isStartingElement(event, TEXT_SECTION)) { |
|
378 |
handleNotYetImplementedElement(event); |
|
379 |
} else if (isStartingElement(event, ADDENDA)) { |
|
380 |
handleNotYetImplementedElement(event); |
|
381 |
} else { |
|
382 |
handleUnexpectedStartElement(event); |
|
383 |
} |
|
384 |
} else { |
|
385 |
handleUnexpectedElement(event); |
|
386 |
} |
|
387 |
} |
|
388 |
throw new IllegalStateException("Publication has no ending element"); |
|
389 |
} |
|
390 |
|
|
391 |
private void handleMetaData(MarkupImportState state, XMLEventReader reader, |
|
392 |
XMLEvent parentEvent) throws XMLStreamException { |
|
393 |
checkNoAttributes(parentEvent); |
|
394 |
|
|
395 |
while (reader.hasNext()) { |
|
396 |
XMLEvent next = readNoWhitespace(reader); |
|
397 |
if (isMyEndingElement(next, parentEvent)) { |
|
398 |
return; |
|
399 |
} else if (isStartingElement(next, DEFAULT_MEDIA_URL)) { |
|
400 |
String baseUrl = getCData(state, reader, next); |
|
241 | 401 |
try { |
242 |
if ((i++ % modCount) == 0 && i > 1){ logger.info("Taxa handled: " + (i-1));} |
|
243 |
if (! elTaxon.getName().equalsIgnoreCase("taxon")){ |
|
244 |
logger.warn("body has element other than 'taxon'"); |
|
402 |
new URL(baseUrl); |
|
403 |
state.setBaseMediaUrl(baseUrl); |
|
404 |
} catch (MalformedURLException e) { |
|
405 |
String message = "defaultMediaUrl '%s' is not a valid URL"; |
|
406 |
message = String.format(message, baseUrl); |
|
407 |
fireWarningEvent(message, next, 8); |
|
245 | 408 |
} |
246 |
|
|
247 |
BotanicalName botanicalName = BotanicalName.NewInstance(Rank.SPECIES()); |
|
248 |
Taxon taxon = Taxon.NewInstance(botanicalName, state.getConfig().getSourceReference()); |
|
249 |
|
|
250 |
handleTaxonAttributes(elTaxon, taxon, state); |
|
251 |
|
|
252 |
|
|
253 |
List<Element> children = elTaxon.getChildren(); |
|
254 |
handleTaxonElement(state, unhandledTitleClassess, unhandledNomeclatureChildren, unhandledDescriptionChildren, taxon, children); |
|
255 |
handleTaxonRelation(state, taxon, lastTaxon); |
|
256 |
lastTaxon = taxon; |
|
257 |
taxaToSave.add(taxon); |
|
258 |
state.getConfig().setLastTaxonUuid(lastTaxon.getUuid()); |
|
259 |
|
|
260 |
} catch (Exception e) { |
|
261 |
logger.warn("Exception occurred in Sapindacea taxon import: " + e); |
|
262 |
e.printStackTrace(); |
|
409 |
} else { |
|
410 |
handleUnexpectedElement(next); |
|
263 | 411 |
} |
264 |
|
|
265 | 412 |
} |
266 |
|
|
267 |
System.out.println(state.getUnmatchedLeads().toString()); |
|
268 |
logger.warn("There are taxa with attributes 'excluded' and 'dubious'"); |
|
269 |
|
|
270 |
logger.info("Children for nomenclature are: " + unhandledNomeclatureChildren); |
|
271 |
logger.info("Children for description are: " + unhandledDescriptionChildren); |
|
272 |
logger.info("Children for homotypes are: " + unhandledHomotypeChildren); |
|
273 |
logger.info("Children for nom are: " + unhandledNomChildren); |
|
274 |
|
|
275 |
|
|
276 |
//invokeRelations(source, cdmApp, deleteAll, taxonMap, referenceMap); |
|
277 |
logger.info(i + " taxa handled. Saving ..."); |
|
278 |
getTaxonService().saveOrUpdate(taxaToSave); |
|
279 |
getFeatureTreeService().saveOrUpdateFeatureNodesAll(state.getFeatureNodesToSave()); |
|
280 |
state.getFeatureNodesToSave().clear(); |
|
281 |
// commitTransaction(tx); |
|
282 |
|
|
283 |
logger.info("end makeTaxa ..."); |
|
284 |
logger.info("start makeKey ..."); |
|
285 |
// invokeDoKey(state); |
|
286 |
logger.info("end makeKey ..."); |
|
287 |
|
|
288 |
return success.getValue(); |
|
289 |
} |
|
290 |
|
|
291 |
|
|
413 |
throw new IllegalStateException("MetaData has no ending element"); |
|
292 | 414 |
|
415 |
} |
|
293 | 416 |
|
294 |
private void handleTaxonAttributes(Element elTaxon, Taxon taxon, MarkupImportState state) { |
|
295 |
List<Attribute> attrList = elTaxon.getAttributes(); |
|
296 |
for (Attribute attr : attrList){ |
|
297 |
String attrName = attr.getName(); |
|
298 |
String attrValue = attr.getValue(); |
|
299 |
if ("class".equals(attrName)){ |
|
300 |
if (attrValue.equalsIgnoreCase("dubious") || attrValue.equalsIgnoreCase("DUBIOUS GENUS") || attrValue.equalsIgnoreCase("DOUBTFUL SPECIES") ){ |
|
301 |
taxon.setDoubtful(true); |
|
302 |
}else{ |
|
303 |
MarkerType markerType = getMarkerType(state, attrValue); |
|
304 |
if (markerType == null){ |
|
305 |
logger.warn("Class attribute value for taxon not yet supported: " + attrValue); |
|
306 |
}else{ |
|
307 |
taxon.addMarker(Marker.NewInstance(markerType, true)); |
|
417 |
private void handleTreatment(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException { |
|
418 |
checkNoAttributes(parentEvent); |
|
419 |
Taxon lastTaxon = null; |
|
420 |
while (reader.hasNext()) { |
|
421 |
XMLEvent next = readNoWhitespace(reader); |
|
422 |
if (isStartingElement(next, TAXON)) { |
|
423 |
Taxon thisTaxon = handleTaxon(state, reader, next.asStartElement()); |
|
424 |
doTaxonRelation(state, thisTaxon, lastTaxon, parentEvent.getLocation()); |
|
425 |
lastTaxon = thisTaxon; |
|
426 |
// TODO for imports spanning multiple documents ?? Still needed? |
|
427 |
state.getConfig().setLastTaxonUuid(lastTaxon.getUuid()); |
|
428 |
} else if (isMyEndingElement(next, parentEvent)) { |
|
429 |
Set<PolytomousKeyNode> keyNodesToSave = state.getPolytomousKeyNodesToSave(); |
|
430 |
//better save the key then the nodes |
|
431 |
Set<PolytomousKey> keySet = new HashSet<PolytomousKey>(); |
|
432 |
for (PolytomousKeyNode node : keyNodesToSave){ |
|
433 |
PolytomousKey key = node.getKey(); |
|
434 |
keySet.add(key); |
|
308 | 435 |
} |
436 |
save(keySet, state); |
|
437 |
//unmatched key leads |
|
438 |
UnmatchedLeads unmatched = state.getUnmatchedLeads(); |
|
439 |
if (unmatched.size() > 0){ |
|
440 |
String message = "The following key leads are unmatched: %s"; |
|
441 |
message = String.format(message, state.getUnmatchedLeads().toString()); |
|
442 |
fireWarningEvent(message, next, 6); |
|
309 | 443 |
} |
310 |
}else if ("num".equals(attrName)){ |
|
311 |
logger.warn("num not yet supported"); |
|
312 |
}else{ |
|
313 |
logger.warn("Attribute " + attrName + " not yet supported for element taxon"); |
|
444 |
// save(keyNodesToSave, state); |
|
445 |
|
|
446 |
return; |
|
447 |
} else { |
|
448 |
handleUnexpectedElement(next); |
|
314 | 449 |
} |
315 | 450 |
} |
316 |
|
|
451 |
return; |
|
317 | 452 |
} |
318 | 453 |
|
454 |
/** |
|
455 |
* @param taxon |
|
456 |
* @param lastTaxon |
|
457 |
*/ |
|
458 |
private void doTaxonRelation(MarkupImportState state, Taxon taxon, |
|
459 |
Taxon lastTaxon, Location dataLocation) { |
|
319 | 460 |
|
320 |
private Taxon getLastTaxon(MarkupImportState state) { |
|
321 |
if (state.getConfig().getLastTaxonUuid() == null){ |
|
322 |
return null; |
|
323 |
}else{ |
|
324 |
return (Taxon)getTaxonService().find(state.getConfig().getLastTaxonUuid()); |
|
325 |
} |
|
326 |
} |
|
327 |
|
|
328 |
|
|
329 |
// private void invokeDoKey(SapindaceaeImportState state) { |
|
330 |
// TransactionStatus tx = startTransaction(); |
|
331 |
// |
|
332 |
// Set<FeatureNode> nodesToSave = new HashSet<FeatureNode>(); |
|
333 |
// ITaxonService taxonService = getTaxonService(); |
|
334 |
// ResultWrapper<Boolean> success = ResultWrapper.NewInstance(true); |
|
335 |
// |
|
336 |
// Element elbody= getBodyElement(state.getConfig()); |
|
337 |
// List<Element> elTaxonList = elbody.getChildren(); |
|
338 |
// |
|
339 |
// int i = 0; |
|
340 |
// |
|
341 |
// //for each taxon |
|
342 |
// for (Element elTaxon : elTaxonList){ |
|
343 |
// if ((i++ % modCount) == 0 && i > 1){ logger.info("Taxa handled: " + (i-1));} |
|
344 |
// if (! elTaxon.getName().equalsIgnoreCase("taxon")){ |
|
345 |
// continue; |
|
346 |
// } |
|
347 |
// |
|
348 |
// List<Element> children = elTaxon.getChildren("key"); |
|
349 |
// for (Element element : children){ |
|
350 |
// handleKeys(state, element, null); |
|
351 |
// } |
|
352 |
// nodesToSave.add(taxon); |
|
353 |
// |
|
354 |
// } |
|
355 |
// |
|
356 |
// } |
|
461 |
Classification tree = makeTree(state, dataLocation); |
|
462 |
if (lastTaxon == null) { |
|
463 |
tree.addChildTaxon(taxon, null, null, null); |
|
464 |
return; |
|
465 |
} |
|
466 |
Rank thisRank = taxon.getName().getRank(); |
|
467 |
Rank lastRank = lastTaxon.getName().getRank(); |
|
468 |
if (lastTaxon.getTaxonNodes().size() > 0) { |
|
469 |
TaxonNode lastNode = lastTaxon.getTaxonNodes().iterator().next(); |
|
470 |
if (thisRank.isLower(lastRank)) { |
|
471 |
lastNode.addChildTaxon(taxon, null, null, null); |
|
472 |
fillMissingEpithetsForTaxa(lastTaxon, taxon); |
|
473 |
} else if (thisRank.equals(lastRank)) { |
|
474 |
TaxonNode parent = lastNode.getParent(); |
|
475 |
if (parent != null) { |
|
476 |
parent.addChildTaxon(taxon, null, null, null); |
|
477 |
fillMissingEpithetsForTaxa(parent.getTaxon(), taxon); |
|
478 |
} else { |
|
479 |
tree.addChildTaxon(taxon, null, null, null); |
|
480 |
} |
|
481 |
} else if (thisRank.isHigher(lastRank)) { |
|
482 |
doTaxonRelation(state, taxon, lastNode.getParent().getTaxon(), |
|
483 |
dataLocation); |
|
484 |
// TaxonNode parentNode = handleTaxonRelation(state, taxon, |
|
485 |
// lastNode.getParent().getTaxon()); |
|
486 |
// parentNode.addChildTaxon(taxon, null, null, null); |
|
487 |
} |
|
488 |
} else { |
|
357 | 489 |
|
490 |
String message = "Last taxon has no node"; |
|
491 |
fireWarningEvent(message, makeLocationStr(dataLocation), 6); |
|
492 |
} |
|
493 |
} |
|
358 | 494 |
|
359 |
// body/taxon/* |
|
360 |
private void handleTaxonElement(MarkupImportState state, Set<String> unhandledTitleClassess, Set<String> unhandledNomeclatureChildren, Set<String> unhandledDescriptionChildren, Taxon taxon, List<Element> children) { |
|
361 |
AnnotatableEntity lastEntity = null; |
|
362 |
for (Element element : children){ |
|
363 |
String elName = element.getName(); |
|
364 |
|
|
365 |
if (elName.equalsIgnoreCase("title")){ |
|
366 |
handleTitle(state, element, taxon, unhandledTitleClassess); |
|
367 |
lastEntity = null; |
|
368 |
}else if(elName.equalsIgnoreCase("nomenclature")){ |
|
369 |
handleNomenclature(state, element, taxon, unhandledNomeclatureChildren); |
|
370 |
lastEntity = null; |
|
371 |
}else if(elName.equalsIgnoreCase("description")){ |
|
372 |
handleDescription(state, element, taxon, unhandledDescriptionChildren); |
|
373 |
lastEntity = null; |
|
374 |
}else if(elName.equalsIgnoreCase("habitatecology")){ |
|
375 |
lastEntity = handleEcology(state, element, taxon); |
|
376 |
}else if(elName.equalsIgnoreCase("distribution")){ |
|
377 |
lastEntity = handleDistribution(state, element, taxon); |
|
378 |
}else if(elName.equalsIgnoreCase("uses")){ |
|
379 |
lastEntity = handleUses(state, element, taxon); |
|
380 |
}else if(elName.equalsIgnoreCase("notes")){ |
|
381 |
lastEntity = handleTaxonNotes(state, element, taxon); |
|
382 |
}else if(elName.equalsIgnoreCase("chromosomes")){ |
|
383 |
lastEntity = handleChromosomes(state, element, taxon); |
|
384 |
}else if(elName.equalsIgnoreCase("vernacularnames")){ |
|
385 |
handleVernaculars(state, element, taxon); |
|
386 |
}else if(elName.equalsIgnoreCase("key")){ |
|
387 |
lastEntity = handleKeys(state, element, taxon); |
|
388 |
}else if(elName.equalsIgnoreCase("references")){ |
|
389 |
handleReferences(state, element, taxon, lastEntity); |
|
390 |
lastEntity = null; |
|
391 |
}else if(elName.equalsIgnoreCase("taxon")){ |
|
392 |
logger.warn("A taxon should not be part of a taxon"); |
|
393 |
}else if(elName.equalsIgnoreCase("homotypes")){ |
|
394 |
logger.warn("Homotypes should be included in the nomenclature flag but is child of taxon [XPath: body/taxon/homotypes]"); |
|
395 |
}else{ |
|
396 |
logger.warn("Unexpected child for taxon: " + elName); |
|
495 |
/** |
|
496 |
* @param state |
|
497 |
* @param dataLocation |
|
498 |
* @return |
|
499 |
*/ |
|
500 |
private Classification makeTree(MarkupImportState state, Location dataLocation) { |
|
501 |
Classification result = state.getTree(null); |
|
502 |
if (result == null) { |
|
503 |
UUID uuid = state.getConfig().getClassificationUuid(); |
|
504 |
if (uuid == null) { |
|
505 |
String message = "No classification uuid is defined"; |
|
506 |
fireWarningEvent(message, makeLocationStr(dataLocation), 6); |
|
507 |
result = createNewClassification(state); |
|
508 |
} else { |
|
509 |
result = getClassificationService().find(uuid); |
|
510 |
if (result == null) { |
|
511 |
result = createNewClassification(state); |
|
512 |
result.setUuid(uuid); |
|
397 | 513 |
} |
398 | 514 |
} |
515 |
state.putTree(null, result); |
|
399 | 516 |
} |
400 |
|
|
401 |
|
|
402 |
private void handleVernaculars(MarkupImportState state, Element elVernacular, Taxon taxon) { |
|
403 |
verifyNoAttribute(elVernacular); |
|
404 |
verifyNoChildren(elVernacular, false); |
|
405 |
String value = elVernacular.getTextNormalize(); |
|
406 |
Feature feature = Feature.COMMON_NAME(); |
|
407 |
value = replaceStart(value, "Noms vernaculaires"); |
|
408 |
String[] dialects = value.split(";"); |
|
409 |
for (String singleDialect : dialects){ |
|
410 |
handleSingleDialect(taxon, singleDialect, feature, state); |
|
517 |
save(result, state); |
|
518 |
return result; |
|
411 | 519 |
} |
412 |
return; |
|
520 |
|
|
521 |
private Classification createNewClassification(MarkupImportState state) { |
|
522 |
Classification result; |
|
523 |
result = Classification.NewInstance(state.getConfig().getClassificationTitle()); |
|
524 |
state.putTree(null, result); |
|
525 |
return result; |
|
413 | 526 |
} |
414 | 527 |
|
528 |
private Taxon handleTaxon(MarkupImportState state, XMLEventReader reader, StartElement parentEvent) throws XMLStreamException { |
|
529 |
// TODO progress monitoring |
|
530 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
531 |
Taxon taxon = createTaxonAndName(state, attributes); |
|
532 |
state.setCurrentTaxon(taxon); |
|
415 | 533 |
|
416 |
private void handleSingleDialect(Taxon taxon, String singleDialect, Feature feature, MarkupImportState state) { |
|
417 |
singleDialect = singleDialect.trim(); |
|
418 |
TaxonDescription description = getDescription(taxon); |
|
419 |
String reDialect = "\\(dial\\.\\s.*\\)"; |
|
420 |
// String reDialect = "\\(.*\\)"; |
|
421 |
Pattern patDialect = Pattern.compile(reDialect); |
|
422 |
Matcher matcher = patDialect.matcher(singleDialect); |
|
423 |
if (matcher.find()){ |
|
424 |
String dialect = singleDialect.substring(matcher.start(), matcher.end()); |
|
425 |
dialect = dialect.replace("(dial. ", "").replace(")", ""); |
|
534 |
boolean hasTitle = false; |
|
535 |
boolean hasNomenclature = false; |
|
536 |
String taxonTitle = null; |
|
537 |
|
|
538 |
while (reader.hasNext()) { |
|
539 |
XMLEvent next = readNoWhitespace(reader); |
|
540 |
if (next.isEndElement()) { |
|
541 |
if (isMyEndingElement(next, parentEvent)) { |
|
542 |
checkMandatoryElement(hasTitle, parentEvent, TAXONTITLE); |
|
543 |
checkMandatoryElement(hasNomenclature, parentEvent, NOMENCLATURE); |
|
544 |
handleUnexpectedAttributes(parentEvent.getLocation(),attributes); |
|
426 | 545 |
|
427 |
Language language = null; |
|
546 |
makeKeyNodes(state, parentEvent, taxonTitle); |
|
547 |
state.setCurrentTaxon(null); |
|
548 |
state.setCurrentTaxonNum(null); |
|
549 |
save(taxon, state); |
|
550 |
return taxon; |
|
551 |
} else { |
|
552 |
if (isEndingElement(next, HEADING)) { |
|
553 |
// NOT YET IMPLEMENTED |
|
554 |
popUnimplemented(next.asEndElement()); |
|
555 |
} else if (isEndingElement(next, TEXT_SECTION)) { |
|
556 |
// NOT YET IMPLEMENTED |
|
557 |
popUnimplemented(next.asEndElement()); |
|
558 |
} else if (isEndingElement(next, REFERENCES)) { |
|
559 |
// NOT YET IMPLEMENTED |
|
560 |
popUnimplemented(next.asEndElement()); |
|
561 |
} else { |
|
562 |
handleUnexpectedEndElement(next.asEndElement()); |
|
563 |
} |
|
564 |
} |
|
565 |
} else if (next.isStartElement()) { |
|
566 |
if (isStartingElement(next, HEADING)) { |
|
567 |
handleNotYetImplementedElement(next); |
|
568 |
} else if (isStartingElement(next, TAXONTITLE)) { |
|
569 |
taxonTitle = handleTaxonTitle(state, reader, next); |
|
570 |
hasTitle = true; |
|
571 |
} else if (isStartingElement(next, WRITER)) { |
|
572 |
makeKeyWriter(state, reader, taxon, taxonTitle, next); |
|
573 |
} else if (isStartingElement(next, TEXT_SECTION)) { |
|
574 |
handleNotYetImplementedElement(next); |
|
575 |
} else if (isStartingElement(next, KEY)) { |
|
576 |
handleKey(state, reader, next); |
|
577 |
} else if (isStartingElement(next, NOMENCLATURE)) { |
|
578 |
handleNomenclature(state, reader, next); |
|
579 |
hasNomenclature = true; |
|
580 |
} else if (isStartingElement(next, FEATURE)) { |
|
581 |
handleFeature(state, reader, next); |
|
582 |
} else if (isStartingElement(next, NOTES)) { |
|
583 |
// TODO is this the correct way to handle notes? |
|
584 |
String note = handleNotes(state, reader, next); |
|
585 |
|
|
586 |
UUID notesUuid; |
|
428 | 587 |
try { |
429 |
language = this.getLanguage(state, state.getTransformer().getLanguageUuid(dialect), dialect, dialect, dialect); |
|
588 |
notesUuid = state.getTransformer().getFeatureUuid( |
|
589 |
"notes"); |
|
590 |
Feature feature = getFeature(state, notesUuid, "Notes", |
|
591 |
"Notes", "note", null); |
|
592 |
TextData textData = TextData.NewInstance(feature); |
|
593 |
textData.putText(Language.DEFAULT(), note); |
|
594 |
TaxonDescription description = getTaxonDescription( |
|
595 |
taxon, false, true); |
|
596 |
description.addElement(textData); |
|
430 | 597 |
} catch (UndefinedTransformerMethodException e) { |
431 |
logger.error(e.getMessage()); |
|
598 |
String message = "getFeatureUuid method not yet implemented"; |
|
599 |
fireWarningEvent(message, next, 8); |
|
432 | 600 |
} |
433 |
|
|
434 |
String commonNames = singleDialect.substring(0, matcher.start()); |
|
435 |
String[] splitNames = commonNames.split(","); |
|
436 |
for (String commonNameString : splitNames){ |
|
437 |
commonNameString = commonNameString.trim(); |
|
438 |
CommonTaxonName commonName = CommonTaxonName.NewInstance(commonNameString, language); |
|
439 |
description.addElement(commonName); |
|
601 |
} else if (isStartingElement(next, REFERENCES)) { |
|
602 |
handleNotYetImplementedElement(next); |
|
603 |
} else if (isStartingElement(next, FIGURE)) { |
|
604 |
handleFigure(state, reader, next); |
|
605 |
} else if (isStartingElement(next, FOOTNOTE)) { |
|
606 |
FootnoteDataHolder footnote = handleFootnote(state, reader, |
|
607 |
next); |
|
608 |
if (footnote.isRef()) { |
|
609 |
String message = "Ref footnote not implemented here"; |
|
610 |
fireWarningEvent(message, next, 4); |
|
611 |
} else { |
|
612 |
registerGivenFootnote(state, footnote); |
|
440 | 613 |
} |
441 |
}else{
|
|
442 |
logger.warn("No dialect match: " + singleDialect);
|
|
614 |
} else {
|
|
615 |
handleUnexpectedStartElement(next);
|
|
443 | 616 |
} |
617 |
} else { |
|
618 |
handleUnexpectedElement(next); |
|
444 | 619 |
} |
445 |
|
|
446 |
|
|
447 |
private void handleReferences(MarkupImportState state, Element elReferences, Taxon taxon, AnnotatableEntity lastEntity) { |
|
448 |
verifyNoAttribute(elReferences); |
|
449 |
verifyNoChildren(elReferences, true); |
|
450 |
String refString = elReferences.getTextNormalize(); |
|
451 |
if (lastEntity == null){ |
|
452 |
logger.warn("No last entity defined: " + refString); |
|
453 |
return; |
|
454 | 620 |
} |
455 |
|
|
456 |
Annotation annotation = Annotation.NewInstance(refString, AnnotationType.EDITORIAL(), Language.DEFAULT()); |
|
457 |
lastEntity.addAnnotation(annotation); |
|
621 |
throw new IllegalStateException("<Taxon> has no closing tag"); |
|
458 | 622 |
} |
459 | 623 |
|
460 |
|
|
461 |
private PolytomousKey handleKeys(MarkupImportState state, Element elKey, Taxon taxon) { |
|
462 |
UnmatchedLeads openKeys = state.getUnmatchedLeads(); |
|
463 |
|
|
464 |
//title |
|
465 |
String title = makeKeyTitle(elKey); |
|
466 |
|
|
467 |
//key |
|
468 |
PolytomousKey key = PolytomousKey.NewTitledInstance(title); |
|
624 |
private void makeKeyNodes(MarkupImportState state, XMLEvent event, String taxonTitle) { |
|
625 |
Taxon taxon = state.getCurrentTaxon(); |
|
626 |
String num = state.getCurrentTaxonNum(); |
|
469 | 627 |
|
470 |
//TODO add covered taxa etc.
|
|
471 |
verifyNoAttribute(elKey);
|
|
628 |
String nameString = CdmBase.deproxy(taxon.getName(), NonViralName.class).getNameCache();
|
|
629 |
// String nameString = taxonTitle;
|
|
472 | 630 |
|
473 |
//notes |
|
474 |
makeKeyNotes(elKey, key); |
|
631 |
//try to find matching lead nodes |
|
632 |
UnmatchedLeadsKey leadsKey = UnmatchedLeadsKey.NewInstance(num, nameString); |
|
633 |
Set<PolytomousKeyNode> matchingNodes = handleMatchingNodes(state, taxon, leadsKey); |
|
475 | 634 |
|
476 |
//keycouplets |
|
477 |
List<Element> keychoices = new ArrayList<Element>(); |
|
478 |
keychoices.addAll(elKey.getChildren("keycouplet")); |
|
479 |
keychoices.addAll(elKey.getChildren("keychoice")); |
|
480 |
|
|
481 |
|
|
482 |
for (Element elKeychoice : keychoices){ |
|
483 |
handleKeyChoices(state, openKeys, key, elKeychoice, taxon); |
|
484 |
elKey.removeContent(elKeychoice); |
|
635 |
if (num != null){//same without using the num |
|
636 |
UnmatchedLeadsKey noNumLeadsKey = UnmatchedLeadsKey.NewInstance("", nameString); |
|
637 |
Set<PolytomousKeyNode> noNumMatchingNodes = handleMatchingNodes(state, taxon, noNumLeadsKey); |
|
638 |
if(noNumMatchingNodes.size() > 0){ |
|
639 |
String message ="Taxon matches additional key node when not considering <num> attribute in taxontitle. This may be correct but may also indicate an error."; |
|
640 |
fireWarningEvent(message, event, 1); |
|
641 |
} |
|
485 | 642 |
} |
486 |
|
|
487 |
// |
|
488 |
verifyNoChildren(elKey); |
|
489 |
logger.info("Unmatched leads after key handling:" + openKeys.toString()); |
|
490 |
|
|
491 |
|
|
492 |
if (state.getConfig().isDoPrintKeys()){ |
|
493 |
key.print(System.err); |
|
643 |
//report missing match, if num exists |
|
644 |
if (matchingNodes.isEmpty() && num != null){ |
|
645 |
String message = "Taxon has <num> attribute in taxontitle but no matching key nodes exist: %s, Key: %s"; |
|
646 |
message = String.format(message, num, leadsKey.toString()); |
|
647 |
fireWarningEvent(message, event, 1); |
|
494 | 648 |
} |
495 |
getPolytomousKeyService().save(key); |
|
496 |
return key; |
|
497 |
} |
|
498 |
|
|
499 |
|
|
500 |
/** |
|
501 |
* @param state |
|
502 |
* @param elKey |
|
503 |
* @param openKeys |
|
504 |
* @param key |
|
505 |
* @param elKeychoice |
|
506 |
* @param taxon |
|
507 |
*/ |
|
508 |
private void handleKeyChoices(MarkupImportState state, UnmatchedLeads openKeys, PolytomousKey key, Element elKeychoice, Taxon taxon) { |
|
509 |
|
|
510 |
//char Attribute |
|
511 |
//TODO it's still unclear if char is a feature and needs to be a new attribute |
|
512 |
//or if it is handled as question. Therefore both cases are handled but feature |
|
513 |
//is finally not yet set |
|
514 |
KeyStatement question = handleKeychoiceChar(state, elKeychoice); |
|
515 |
Feature feature = handleKeychoiceCharAsFeature(state, elKeychoice); |
|
516 | 649 |
|
517 |
//lead |
|
518 |
List<PolytomousKeyNode> childNodes = handleKeychoiceLeads(state, key, elKeychoice, taxon, question, feature); |
|
519 |
|
|
520 |
//num -> match with unmatched leads |
|
521 |
handleKeychoiceNum(openKeys, key, elKeychoice, childNodes); |
|
522 |
|
|
523 |
//others |
|
524 |
verifyNoAttribute(elKeychoice); |
|
525 | 650 |
} |
526 |
|
|
527 |
|
|
528 |
/** |
|
529 |
* @param openKeys |
|
530 |
* @param key |
|
531 |
* @param elKeychoice |
|
532 |
* @param childNodes |
|
533 |
*/ |
|
534 |
private void handleKeychoiceNum(UnmatchedLeads openKeys, PolytomousKey key, Element elKeychoice, List<PolytomousKeyNode> childNodes) { |
|
535 |
Attribute numAttr = elKeychoice.getAttribute("num"); |
|
536 |
String num = CdmUtils.removeTrailingDot(numAttr == null? "":numAttr.getValue()); |
|
537 |
UnmatchedLeadsKey okk = UnmatchedLeadsKey.NewInstance(key, num); |
|
538 |
Set<PolytomousKeyNode> matchingNodes = openKeys.getNodes(okk); |
|
651 |
|
|
652 |
private Set<PolytomousKeyNode> handleMatchingNodes(MarkupImportState state, Taxon taxon, UnmatchedLeadsKey leadsKey) { |
|
653 |
Set<PolytomousKeyNode> matchingNodes = state.getUnmatchedLeads().getNodes(leadsKey); |
|
539 | 654 |
for (PolytomousKeyNode matchingNode : matchingNodes){ |
540 |
for (PolytomousKeyNode childNode : childNodes){ |
|
541 |
matchingNode.addChild(childNode); |
|
542 |
} |
|
543 |
openKeys.removeNode(okk, matchingNode); |
|
544 |
} |
|
545 |
if (matchingNodes.isEmpty()){ |
|
546 |
for (PolytomousKeyNode childNode : childNodes){ |
|
547 |
key.getRoot().addChild(childNode); |
|
548 |
} |
|
655 |
state.getUnmatchedLeads().removeNode(leadsKey, matchingNode); |
|
656 |
matchingNode.setTaxon(taxon); |
|
657 |
state.getPolytomousKeyNodesToSave().add(matchingNode); |
|
549 | 658 |
} |
550 |
|
|
551 |
elKeychoice.removeAttribute("num"); |
|
659 |
return matchingNodes; |
|
552 | 660 |
} |
553 | 661 |
|
662 |
private void handleKey(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException { |
|
663 |
// attributes |
|
664 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
665 |
String isSpotcharacters = getAndRemoveAttributeValue(attributes, IS_SPOTCHARACTERS); |
|
666 |
if (isNotBlank(isSpotcharacters) ) { |
|
667 |
//TODO isSpotcharacters |
|
668 |
String message = "Attribute isSpotcharacters not yet implemented for <key>"; |
|
669 |
fireWarningEvent(message, parentEvent, 4); |
|
670 |
} |
|
671 |
|
|
672 |
PolytomousKey key = PolytomousKey.NewInstance(); |
|
673 |
key.addTaxonomicScope(state.getCurrentTaxon()); |
|
674 |
state.setCurrentKey(key); |
|
675 |
|
|
676 |
boolean isFirstCouplet = true; |
|
677 |
while (reader.hasNext()) { |
|
678 |
XMLEvent next = readNoWhitespace(reader); |
|
679 |
if (isMyEndingElement(next, parentEvent)) { |
|
680 |
save(key, state); |
|
681 |
state.setCurrentKey(null); |
|
682 |
return; |
|
683 |
} else if (isEndingElement(next, KEYNOTES)){ |
|
684 |
popUnimplemented(next.asEndElement()); |
|
685 |
} else if (isStartingElement(next, KEY_TITLE)) { |
|
686 |
handleKeyTitle(state, reader, next); |
|
687 |
} else if (isStartingElement(next, KEYNOTES)) { |
|
688 |
//TODO |
|
689 |
handleNotYetImplementedElement(next); |
|
690 |
} else if (isStartingElement(next, COUPLET)) { |
|
691 |
PolytomousKeyNode node = null; |
|
692 |
if (isFirstCouplet){ |
|
693 |
node = key.getRoot(); |
|
694 |
isFirstCouplet = false; |
|
695 |
} |
|
696 |
handleCouplet(state, reader, next, node); |
|
697 |
} else { |
|
698 |
handleUnexpectedElement(next); |
|
699 |
} |
|
700 |
} |
|
701 |
throw new IllegalStateException("<key> has no closing tag"); |
|
702 |
} |
|
554 | 703 |
|
555 | 704 |
/** |
556 | 705 |
* @param state |
706 |
* @param reader |
|
557 | 707 |
* @param key |
558 |
* @param elKeychoice |
|
559 |
* @param taxon |
|
560 |
* @param feature |
|
561 |
* @return |
|
708 |
* @param next |
|
709 |
* @throws XMLStreamException |
|
562 | 710 |
*/ |
563 |
private List<PolytomousKeyNode> handleKeychoiceLeads( MarkupImportState state, PolytomousKey key, Element elKeychoice, Taxon taxon, KeyStatement question, Feature feature) {
|
|
564 |
List<PolytomousKeyNode> childNodes = new ArrayList<PolytomousKeyNode>();
|
|
565 |
List<Element> leads = elKeychoice.getChildren("lead");
|
|
566 |
for(Element elLead : leads){
|
|
567 |
PolytomousKeyNode childNode = handleLead(state, key, elLead, taxon, question, feature); |
|
568 |
childNodes.add(childNode);
|
|
569 |
}
|
|
570 |
return childNodes;
|
|
711 |
private void handleKeyTitle(MarkupImportState state, XMLEventReader reader, XMLEvent next) throws XMLStreamException {
|
|
712 |
PolytomousKey key = state.getCurrentKey();
|
|
713 |
String keyTitle = getCData(state, reader, next);
|
|
714 |
String standardTitles = "(?i)(Key\\sto\\sthe\\s(genera|species|varieties|forms))";
|
|
715 |
|
|
716 |
if (isNotBlank(keyTitle) ){
|
|
717 |
if (!state.getConfig().isReplaceStandardKeyTitles() || ! keyTitle.matches(standardTitles)){
|
|
718 |
key.setTitleCache(keyTitle, true);
|
|
571 | 719 |
} |
572 |
|
|
573 |
|
|
574 |
/** |
|
575 |
* @param state |
|
576 |
* @param elKeychoice |
|
577 |
* @return |
|
578 |
*/ |
|
579 |
private KeyStatement handleKeychoiceChar(MarkupImportState state, Element elKeychoice) { |
|
580 |
KeyStatement statement = null; |
|
581 |
Attribute charAttr = elKeychoice.getAttribute("char"); |
|
582 |
if (charAttr != null){ |
|
583 |
String charStr = charAttr.getValue(); |
|
584 |
if (StringUtils.isNotBlank(charStr)){ |
|
585 |
statement = KeyStatement.NewInstance(charStr); |
|
586 | 720 |
} |
587 |
elKeychoice.removeAttribute("char"); |
|
588 |
} |
|
589 |
return statement; |
|
590 |
} |
|
591 |
|
|
592 |
/** |
|
593 |
* @param state |
|
594 |
* @param elKeychoice |
|
595 |
* @return |
|
596 |
*/ |
|
597 |
private Feature handleKeychoiceCharAsFeature(MarkupImportState state, Element elKeychoice) { |
|
598 |
Feature feature = null; |
|
599 |
Attribute charAttr = elKeychoice.getAttribute("char"); |
|
600 |
if (charAttr != null){ |
|
601 |
String charStr = charAttr.getValue(); |
|
602 |
feature = getFeature(charStr, state); |
|
603 |
elKeychoice.removeAttribute("char"); |
|
604 | 721 |
} |
605 |
return feature; |
|
606 |
} |
|
607 |
|
|
608 | 722 |
|
609 |
private PolytomousKeyNode handleLead(MarkupImportState state, PolytomousKey key, Element elLead, Taxon taxon, KeyStatement question, Feature feature) { |
|
610 |
PolytomousKeyNode node = PolytomousKeyNode.NewInstance(); |
|
611 |
//TODO the char attribute in the keychoice is more a feature than a question |
|
612 |
//needs to be discussed on model side |
|
613 |
node.setQuestion(question); |
|
614 |
// node.setFeature(feature); |
|
615 |
|
|
616 |
//text |
|
617 |
String text = handleLeadText(elLead, node); |
|
618 |
|
|
619 |
//num |
|
620 |
handleLeadNum(elLead, text); |
|
723 |
private void handleCouplet(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, PolytomousKeyNode parentNode) throws XMLStreamException { |
|
724 |
String num = getOnlyAttribute(parentEvent, NUM, true); |
|
725 |
List<PolytomousKeyNode> childList = new ArrayList<PolytomousKeyNode>(); |
|
621 | 726 |
|
622 |
//goto |
|
623 |
handleLeadGoto(state, key, elLead, taxon, node); |
|
624 |
|
|
625 |
//others |
|
626 |
verifyNoAttribute(elLead); |
|
627 |
|
|
628 |
return node; |
|
727 |
while (reader.hasNext()) { |
|
728 |
XMLEvent next = readNoWhitespace(reader); |
|
729 |
if (isMyEndingElement(next, parentEvent)) { |
|
730 |
completeCouplet(state, parentEvent, parentNode, num, childList); |
|
731 |
return; |
|
732 |
} else if (isStartingElement(next, QUESTION)) { |
|
733 |
handleQuestion(state, reader, next, childList); |
|
734 |
} else if (isStartingElement(next, KEYNOTES)) { |
|
735 |
//TODO |
|
736 |
handleNotYetImplementedElement(next); |
|
737 |
} else if (isEndingElement(next, KEYNOTES)) { |
|
738 |
//TODO |
|
739 |
popUnimplemented(next.asEndElement()); |
|
740 |
} else { |
|
741 |
handleUnexpectedElement(next); |
|
629 | 742 |
} |
630 |
|
|
631 |
|
|
632 |
/** |
|
633 |
* @param elLead |
|
634 |
* @param node |
|
635 |
* @return |
|
636 |
*/ |
|
637 |
private String handleLeadText(Element elLead, PolytomousKeyNode node) { |
|
638 |
String text = elLead.getAttributeValue("text").trim(); |
|
639 |
if (StringUtils.isBlank(text)){ |
|
640 |
logger.warn("Empty text in lead"); |
|
641 |
} |
|
642 |
elLead.removeAttribute("text"); |
|
643 |
KeyStatement statement = KeyStatement.NewInstance(text); |
|
644 |
node.setStatement(statement); |
|
645 |
return text; |
|
743 |
} |
|
744 |
throw new IllegalStateException("<couplet> has no closing tag"); |
|
646 | 745 |
} |
647 | 746 |
|
648 |
|
|
649 | 747 |
/** |
650 | 748 |
* @param state |
651 |
* @param key
|
|
652 |
* @param elLead
|
|
653 |
* @param taxon
|
|
654 |
* @param node
|
|
749 |
* @param parentEvent
|
|
750 |
* @param parentNode
|
|
751 |
* @param num
|
|
752 |
* @param childList
|
|
655 | 753 |
*/ |
656 |
private void handleLeadGoto(MarkupImportState state, PolytomousKey key, Element elLead, Taxon taxon, PolytomousKeyNode node) { |
|
657 |
Attribute gotoAttr = elLead.getAttribute("goto"); |
|
658 |
if (gotoAttr != null){ |
|
659 |
String strGoto = gotoAttr.getValue().trim(); |
|
660 |
//create key |
|
661 |
UnmatchedLeadsKey gotoKey = null; |
|
662 |
if (isInternalNode(strGoto)){ |
|
663 |
gotoKey = UnmatchedLeadsKey.NewInstance(key, strGoto); |
|
664 |
}else{ |
|
665 |
String taxonKey = makeTaxonKey(strGoto, taxon); |
|
666 |
gotoKey = UnmatchedLeadsKey.NewInstance(taxonKey); |
|
667 |
} |
|
668 |
// |
|
669 |
UnmatchedLeads openKeys = state.getUnmatchedLeads(); |
|
670 |
if (gotoKey.isInnerLead()){ |
|
671 |
Set<PolytomousKeyNode> existingNodes = openKeys.getNodes(gotoKey); |
|
672 |
for (PolytomousKeyNode existingNode : existingNodes){ |
|
673 |
node.addChild(existingNode); |
|
674 |
} |
|
754 |
private void completeCouplet(MarkupImportState state, XMLEvent parentEvent, |
|
755 |
PolytomousKeyNode parentNode, String num, |
|
756 |
List<PolytomousKeyNode> childList) { |
|
757 |
if (parentNode != null){ |
|
758 |
for (PolytomousKeyNode childNode : childList){ |
|
759 |
parentNode.addChild(childNode); |
|
675 | 760 |
} |
676 |
openKeys.addKey(gotoKey, node); |
|
677 |
//remove attribute (need for consistency check) |
|
678 |
elLead.removeAttribute("goto"); |
|
679 |
}else{ |
|
680 |
logger.warn("lead has no goto attribute"); |
|
761 |
}else if (isNotBlank(num)){ |
|
762 |
UnmatchedLeadsKey unmatchedKey = UnmatchedLeadsKey.NewInstance(state.getCurrentKey(), num); |
|
763 |
Set<PolytomousKeyNode> nodes = state.getUnmatchedLeads().getNodes(unmatchedKey); |
|
764 |
for(PolytomousKeyNode nodeToMatch: nodes){ |
|
765 |
for (PolytomousKeyNode childNode : childList){ |
|
766 |
nodeToMatch.addChild(childNode); |
|
681 | 767 |
} |
768 |
state.getUnmatchedLeads().removeNode(unmatchedKey, nodeToMatch); |
|
682 | 769 |
} |
683 |
|
|
684 |
|
|
685 |
/** |
|
686 |
* @param elLead |
|
687 |
* @param text |
|
688 |
*/ |
|
689 |
private void handleLeadNum(Element elLead, String text) { |
|
690 |
Attribute numAttr = elLead.getAttribute("num"); |
|
691 |
if (numAttr != null){ |
|
692 |
//TODO num |
|
693 |
String num = numAttr.getValue(); |
|
694 |
elLead.removeAttribute("num"); |
|
695 | 770 |
}else{ |
696 |
logger.info("Keychoice has no num attribute: " + text); |
|
771 |
String message = "Parent num could not be matched. Please check if num (%s) is correct"; |
|
772 |
message = String.format(message, num); |
|
773 |
fireWarningEvent(message, parentEvent, 6); |
|
697 | 774 |
} |
698 | 775 |
} |
699 | 776 |
|
777 |
private void handleQuestion(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, List<PolytomousKeyNode> nodesList) throws XMLStreamException { |
|
778 |
// attributes |
|
779 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
780 |
//needed only for data lineage |
|
781 |
String questionNum = getAndRemoveRequiredAttributeValue(parentEvent, attributes, NUM); |
|
782 |
|
|
783 |
PolytomousKeyNode myNode = PolytomousKeyNode.NewInstance(); |
|
784 |
myNode.setKey(state.getCurrentKey()); //to avoid NPE while computing num in PolytomousKeyNode in case this node is not matched correctly with a parent |
|
785 |
nodesList.add(myNode); |
|
786 |
|
|
787 |
while (reader.hasNext()) { |
|
788 |
XMLEvent next = readNoWhitespace(reader); |
|
789 |
if (isMyEndingElement(next, parentEvent)) { |
|
790 |
return; |
|
791 |
} else if (isStartingElement(next, TEXT)) { |
|
792 |
String text = getCData(state, reader, next); |
|
793 |
KeyStatement statement = KeyStatement.NewInstance(text); |
|
794 |
myNode.setStatement(statement); |
|
795 |
} else if (isStartingElement(next, COUPLET)) { |
|
796 |
//TODO test |
|
797 |
handleCouplet(state, reader, next, myNode); |
|
798 |
} else if (isStartingElement(next, TO_COUPLET)) { |
|
799 |
handleToCouplet(state, reader, next, myNode); |
|
800 |
} else if (isStartingElement(next, TO_TAXON)) { |
|
801 |
handleToTaxon(state, reader, next, myNode); |
|
802 |
|
|
803 |
} else if (isStartingElement(next, TO_KEY)) { |
|
804 |
//TODO |
|
805 |
handleNotYetImplementedElement(next); |
|
806 |
} else if (isEndingElement(next, TO_KEY)){ |
|
807 |
//TODO |
|
808 |
popUnimplemented(next.asEndElement()); |
|
809 |
} else if (isStartingElement(next, KEYNOTES)) { |
|
810 |
//TODO |
|
811 |
handleNotYetImplementedElement(next); |
|
812 |
} else if (isEndingElement(next, KEYNOTES)){ |
|
813 |
//TODO |
|
814 |
popUnimplemented(next.asEndElement()); |
|
815 |
} else { |
|
816 |
handleUnexpectedElement(next); |
|
817 |
} |
|
818 |
} |
|
819 |
throw new IllegalStateException("<question> has no closing tag"); |
|
820 |
} |
|
821 |
|
|
822 |
private void handleToCouplet(MarkupImportState state, XMLEventReader reader, XMLEvent next, PolytomousKeyNode node) throws XMLStreamException { |
|
823 |
String num = getOnlyAttribute(next, NUM, true); |
|
824 |
String cData = getCData(state, reader, next, false); |
|
825 |
if (isNotBlank(cData) && ! cData.equals(num)){ |
|
826 |
String message = "CData ('%s') not handled in <toCouplet>"; |
|
827 |
message = String.format(message, cData); |
|
828 |
fireWarningEvent(message, next, 4); |
|
829 |
} |
|
830 |
UnmatchedLeadsKey unmatched = UnmatchedLeadsKey.NewInstance(state.getCurrentKey(), num); |
|
831 |
state.getUnmatchedLeads().addKey(unmatched, node); |
|
832 |
} |
|
700 | 833 |
|
834 |
private void handleToTaxon(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, PolytomousKeyNode node) throws XMLStreamException { |
|
835 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
836 |
String num = getAndRemoveAttributeValue(attributes, NUM); |
|
837 |
String taxonStr = getCData(state, reader, parentEvent, false); |
|
838 |
//TODO ? |
|
839 |
taxonStr = makeTaxonKey(taxonStr, state.getCurrentTaxon()); |
|
840 |
UnmatchedLeadsKey unmatched = UnmatchedLeadsKey.NewInstance(num, taxonStr); |
|
841 |
state.getUnmatchedLeads().addKey(unmatched, node); |
|
842 |
return; |
|
843 |
} |
|
844 |
|
|
701 | 845 |
private String makeTaxonKey(String strGoto, Taxon taxon) { |
702 | 846 |
String result = ""; |
703 | 847 |
if (strGoto == null){ |
704 | 848 |
return ""; |
705 | 849 |
} |
706 |
String strGenusName = CdmBase.deproxy(taxon.getName(), NonViralName.class).getGenusOrUninomial(); |
|
850 |
|
|
851 |
NonViralName<?> name = CdmBase.deproxy(taxon.getName(), NonViralName.class); |
|
852 |
String strGenusName = name.getGenusOrUninomial(); |
|
853 |
|
|
854 |
|
|
707 | 855 |
strGoto = strGoto.replaceAll("\\([^\\(\\)]*\\)", ""); //replace all brackets |
708 | 856 |
strGoto = strGoto.replaceAll("\\s+", " "); //replace multiple whitespaces by exactly one whitespace |
709 | 857 |
|
... | ... | |
714 | 862 |
if (isGenusAbbrev(single, strGenusName)){ |
715 | 863 |
split[i] = strGenusName; |
716 | 864 |
} |
717 |
// if (isInfraSpecificMarker(single)){ |
|
718 |
// String strSpeciesName = CdmBase.deproxy(taxon.getName(), NonViralName.class).getSpecificEpithet(); |
|
719 |
// split[i] = strGenusName + " " + strSpeciesName + " "; |
|
720 |
// } |
|
865 |
if (isInfraSpecificMarker(single)){ |
|
866 |
String strSpeciesEpi = name.getSpecificEpithet(); |
|
867 |
if (isBlank(result)){ |
|
868 |
result += strGenusName + " " + strSpeciesEpi; |
|
869 |
} |
|
870 |
} |
|
721 | 871 |
result = (result + " " + split[i]).trim(); |
722 | 872 |
} |
723 | 873 |
return result; |
724 | 874 |
} |
725 |
|
|
875 |
|
|
726 | 876 |
|
727 | 877 |
private boolean isInfraSpecificMarker(String single) { |
728 | 878 |
try { |
729 | 879 |
if (Rank.getRankByAbbreviation(single).isInfraSpecific()){ |
730 | 880 |
return true; |
881 |
}else{ |
|
882 |
return false; |
|
731 | 883 |
} |
732 | 884 |
} catch (UnknownCdmTypeException e) { |
733 | 885 |
return false; |
734 | 886 |
} |
735 |
return false; |
|
736 | 887 |
} |
737 |
|
|
738 |
|
|
888 |
|
|
739 | 889 |
private boolean isGenusAbbrev(String single, String strGenusName) { |
740 | 890 |
if (! single.matches("[A-Z]\\.?")) { |
741 | 891 |
return false; |
... | ... | |
746 | 896 |
} |
747 | 897 |
} |
748 | 898 |
|
749 |
|
|
750 |
private boolean isInternalNode(String strGoto) { |
|
751 |
return CdmUtils.isNumeric(strGoto); |
|
752 |
} |
|
753 |
|
|
754 |
|
|
755 |
private void makeKeyNotes(Element keyElement, PolytomousKey key) { |
|
756 |
Element elNotes = keyElement.getChild("notes"); |
|
757 |
if (elNotes != null){ |
|
758 |
keyElement.removeContent(elNotes); |
|
759 |
String notes = elNotes.getTextNormalize(); |
|
760 |
if (StringUtils.isNotBlank(notes)){ |
|
761 |
key.addAnnotation(Annotation.NewInstance(notes, AnnotationType.EDITORIAL(), Language.DEFAULT())); |
|
899 |
/** |
|
900 |
* @param state |
|
901 |
* @param reader |
|
902 |
* @param taxon |
|
903 |
* @param taxonTitle |
|
904 |
* @param next |
|
905 |
* @throws XMLStreamException |
|
906 |
*/ |
|
907 |
private void makeKeyWriter(MarkupImportState state, XMLEventReader reader, Taxon taxon, String taxonTitle, XMLEvent next) throws XMLStreamException { |
|
908 |
WriterDataHolder writer = handleWriter(state, reader, next); |
|
909 |
taxon.addExtension(writer.extension); |
|
910 |
// TODO what if taxonTitle comes later |
|
911 |
if (StringUtils.isNotBlank(taxonTitle) |
|
912 |
&& writer.extension != null) { |
|
913 |
Reference<?> sec = ReferenceFactory.newBookSection(); |
|
914 |
sec.setTitle(taxonTitle); |
|
915 |
TeamOrPersonBase<?> author = createAuthor(writer.writer); |
|
916 |
sec.setAuthorTeam(author); |
|
917 |
sec.setInReference(state.getConfig() |
|
918 |
.getSourceReference()); |
|
919 |
taxon.setSec(sec); |
|
920 |
registerFootnotes(state, sec, writer.footnotes); |
|
921 |
} else { |
|
922 |
String message = "No taxontitle exists for writer"; |
|
923 |
fireWarningEvent(message, next, 6); |
|
924 |
} |
|
925 |
} |
|
926 |
|
|
927 |
private String handleNotes(MarkupImportState state, XMLEventReader reader, |
|
928 |
XMLEvent parentEvent) throws XMLStreamException { |
|
929 |
checkNoAttributes(parentEvent); |
|
930 |
|
|
931 |
String text = ""; |
|
932 |
while (reader.hasNext()) { |
|
933 |
XMLEvent next = readNoWhitespace(reader); |
|
934 |
if (isMyEndingElement(next, parentEvent)) { |
|
935 |
return text; |
|
936 |
} else if (next.isEndElement()) { |
|
937 |
if (isEndingElement(next, HEADING)) { |
|
938 |
popUnimplemented(next.asEndElement()); |
|
939 |
} else if (isEndingElement(next, WRITER)) { |
|
940 |
popUnimplemented(next.asEndElement()); |
|
941 |
} else if (isEndingElement(next, NUM)) { |
|
942 |
popUnimplemented(next.asEndElement()); |
|
943 |
} else { |
|
944 |
handleUnexpectedEndElement(next.asEndElement()); |
|
945 |
} |
|
946 |
} else if (next.isStartElement()) { |
|
947 |
if (isStartingElement(next, HEADING)) { |
|
948 |
handleNotYetImplementedElement(next); |
|
949 |
} else if (isStartingElement(next, SUB_HEADING)) { |
|
950 |
String subheading = getCData(state, reader, next).trim(); |
|
951 |
if (! isNoteHeading(subheading)) { |
Also available in: Unified diff
trunk merged into branch