app-import/src/main/java/eu/etaxonomy/cdm/io/mexico/MexicoBorhidiExcelImport.java

   1 /**
   2 * Copyright (C) 2016 EDIT
   3 * European Distributed Institute of Taxonomy
   4 * http://www.e-taxonomy.eu
   5 *
   6 * The contents of this file are subject to the Mozilla Public License Version 1.1
   7 * See LICENSE.TXT at the top of this package for the full license terms.
   8 */
   9 package eu.etaxonomy.cdm.io.mexico;
  10
  11 import java.net.URI;
  12 import java.net.URISyntaxException;
  13 import java.util.Arrays;
  14 import java.util.HashMap;
  15 import java.util.List;
  16 import java.util.Set;
  17 import java.util.UUID;
  18
  19 import org.apache.log4j.Logger;
  20 import org.springframework.stereotype.Component;
  21
  22 import eu.etaxonomy.cdm.common.CdmUtils;
  23 import eu.etaxonomy.cdm.common.UTF8;
  24 import eu.etaxonomy.cdm.model.common.DefinedTerm;
  25 import eu.etaxonomy.cdm.model.common.Extension;
  26 import eu.etaxonomy.cdm.model.common.ExtensionType;
  27 import eu.etaxonomy.cdm.model.common.Identifier;
  28 import eu.etaxonomy.cdm.model.common.TermVocabulary;
  29 import eu.etaxonomy.cdm.model.description.DescriptionElementBase;
  30 import eu.etaxonomy.cdm.model.description.Feature;
  31 import eu.etaxonomy.cdm.model.description.TaxonNameDescription;
  32 import eu.etaxonomy.cdm.model.description.TextData;
  33 import eu.etaxonomy.cdm.model.media.Media;
  34 import eu.etaxonomy.cdm.model.name.IBotanicalName;
  35 import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
  36 import eu.etaxonomy.cdm.model.name.Rank;
  37 import eu.etaxonomy.cdm.model.name.TaxonName;
  38 import eu.etaxonomy.cdm.model.name.TaxonNameFactory;
  39 import eu.etaxonomy.cdm.model.reference.Reference;
  40 import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
  41 import eu.etaxonomy.cdm.model.taxon.Classification;
  42 import eu.etaxonomy.cdm.model.taxon.Taxon;
  43 import eu.etaxonomy.cdm.model.taxon.TaxonNode;
  44 import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
  45
  46 /**
  47  * @author a.mueller
  48  * @since 16.06.2016
  49  *
  50  */
  51 @Component
  52 public class MexicoBorhidiExcelImport<CONFIG extends MexicoBorhidiImportConfigurator>
  53         extends SimpleExcelTaxonImport<CONFIG>{
  54     private static final Logger logger = Logger.getLogger(MexicoBorhidiExcelImport.class);
  55     private static final long serialVersionUID = -3607776356577606657L;
  56
  57     private  static List<String> expectedKeys= Arrays.asList(new String[]{
  58             "FullnameNoAuthors","OutputNameID","OutputFullNameWithAuthors","RefType"
  59             ,"OutputAbbreviatedTitle","OutputCollation","OutputVolume",
  60             "OutputIssue","OutputPage","OutputTitlePageYear","OutputYearPublished",
  61             "OutputBHLLink"});
  62
  63
  64     @Override
  65     protected void firstPass(SimpleExcelTaxonImportState<CONFIG> state) {
  66         String line = state.getCurrentLine() + ": ";
  67         HashMap<String, String> record = state.getOriginalRecord();
  68
  69         Set<String> keys = record.keySet();
  70         checkAllKeysExist(line, keys, expectedKeys);
  71
  72         if (record.get("FullnameNoAuthors") == null ){
  73             logger.warn("No FullnameNoAuthors given: " + line);
  74             return;
  75         }
  76
  77         //Name
  78         IBotanicalName speciesName = makeName(record, state);
  79
  80         //Taxon
  81         Reference sec = state.getConfig().getSecReference();
  82         Taxon taxon = Taxon.NewInstance(speciesName, sec);
  83         TaxonNode rubiaceae = getHighestNode(state);
  84
  85         taxon.addSource(makeOriginalSource(state));
  86
  87
  88         //make genus
  89         makeGenus(state, speciesName, sec, taxon, rubiaceae);
  90     }
  91
  92
  93     private Classification classification;
  94     private TaxonNode rubiaceaeNode;
  95     /**
  96      * @return
  97      */
  98     private TaxonNode getHighestNode(SimpleExcelTaxonImportState<CONFIG> state) {
  99         if (rubiaceaeNode == null){
 100             MexicoBorhidiImportConfigurator config = state.getConfig();
 101             classification = Classification.NewInstance(state.getConfig().getClassificationName());
 102             classification.setUuid(config.getClassificationUuid());
 103             classification.setReference(config.getSecReference());
 104             IBotanicalName nameRubiaceae = TaxonNameFactory.NewBotanicalInstance(Rank.FAMILY());
 105             nameRubiaceae.setGenusOrUninomial("Rubiaceae");
 106             Taxon rubiaceaeTaxon = Taxon.NewInstance(nameRubiaceae, classification.getReference());
 107             rubiaceaeNode = classification.addChildTaxon(rubiaceaeTaxon, null, null);
 108             getClassificationService().save(classification);
 109         }
 110         return rubiaceaeNode;
 111     }
 112
 113     /**
 114      * @param record
 115      * @param state
 116      * @return
 117      */
 118     private IBotanicalName makeName(HashMap<String, String> record, SimpleExcelTaxonImportState<CONFIG> state) {
 119         String line = state.getCurrentLine() + ": ";
 120
 121         String fullNameStr = getValue(record, "OutputFullNameWithAuthors");
 122 //        String volume = getValue(record, "OutputVolume");
 123 //        String issue = getValue(record, "OutputIssue");
 124 //        String page = getValue(record, "OutputPage");
 125         String titleYear = getValue(record, "OutputTitlePageYear");
 126         String publishedYear = getValue(record, "OutputYearPublished");
 127         String refAbbrevTitle = getValue(record, "OutputAbbreviatedTitle");
 128         String outputCollation = getValue(record, "OutputCollation");
 129         String refType = getValue(record, "RefType");
 130
 131
 132         TaxonName name = (TaxonName)nameParser.parseFullName(fullNameStr, NomenclaturalCode.ICNAFP, Rank.SPECIES());
 133         if (name.isProtectedTitleCache()){
 134             //for the 2 ined. names
 135             name = (TaxonName)nameParser.parseReferencedName(fullNameStr, NomenclaturalCode.ICNAFP, Rank.SPECIES());
 136         }
 137         if (name.isProtectedTitleCache()){
 138             logger.warn(line + "Name could not be parsed: " + fullNameStr );
 139         }else{
 140             replaceAuthorNamesAndNomRef(state, name);
 141         }
 142
 143         if (refAbbrevTitle != null){
 144             String[] volumeDetail = makeVolumeDetail(outputCollation);
 145             String detail;
 146             String volume = null;
 147             if (volumeDetail.length > 1){
 148                 volume = volumeDetail[0].trim();
 149                 detail = volumeDetail[1].trim();
 150             }else{
 151                 detail = volumeDetail[0].trim();
 152             }
 153
 154             refAbbrevTitle = refAbbrevTitle.trim();
 155             boolean isArticle = "A".equalsIgnoreCase(refType);
 156
 157             if (isArticle){
 158                 if (! "A".equalsIgnoreCase(refType)){
 159                     logger.warn(line + "RefType problem with article " + refType);
 160                 }
 161
 162                 Reference journal = state.getReference(refAbbrevTitle);
 163                 if (journal == null){
 164                     journal = ReferenceFactory.newJournal();
 165                     journal.setAbbrevTitle(refAbbrevTitle);
 166                     state.putReference(refAbbrevTitle, journal);
 167                     journal.addSource(makeOriginalSource(state));
 168
 169                 }
 170                 Reference article = ReferenceFactory.newArticle();
 171
 172
 173                 //            String detail = page;
 174                 name.setNomenclaturalMicroReference(detail);
 175
 176                 article.setVolume(CdmUtils.Ne(volume));
 177                 article.setInReference(journal);
 178
 179                 titleYear = (isBlank(publishedYear)? titleYear : UTF8.QUOT_DBL_LOW9 + titleYear + UTF8.QUOT_DBL_HIGH_REV9 + "[" + publishedYear + "]");
 180                 article.setDatePublished(TimePeriodParser.parseString(titleYear));
 181
 182                 article.setAuthorship(name.getCombinationAuthorship());
 183
 184                 Reference existingArticle = state.getReference(article.getTitleCache());
 185                 if (existingArticle != null){
 186                     name.setNomenclaturalReference(existingArticle);
 187                 }else{
 188                     name.setNomenclaturalReference(article);
 189                     state.putReference(article.getTitleCache(), article);
 190                     article.addSource(makeOriginalSource(state));
 191                 }
 192             }else{
 193                 if (! "B".equalsIgnoreCase(refType)){
 194                     logger.warn(line + "RefType problem with book" + refType);
 195                 }
 196
 197                 Reference book = ReferenceFactory.newBook();
 198                 book.setAbbrevTitle(refAbbrevTitle);
 199
 200                 //year
 201                 titleYear = (isBlank(publishedYear)? titleYear : UTF8.QUOT_DBL_LOW9 + titleYear + UTF8.QUOT_DBL_HIGH_REV9 + "[" + publishedYear + "]");
 202                 book.setDatePublished(TimePeriodParser.parseString(titleYear));
 203
 204                 book.setAuthorship(name.getCombinationAuthorship());
 205
 206                 //deduplicate
 207                 Reference existingBook = state.getReference(book.getTitleCache());
 208                 if (existingBook != null){
 209                     name.setNomenclaturalReference(existingBook);
 210                 }else{
 211                     name.setNomenclaturalReference(book);
 212                     state.putReference(book.getTitleCache(), book);
 213                 }
 214
 215                 book.setVolume(volume);
 216
 217                 //String detail = page;
 218                 name.setNomenclaturalMicroReference(detail);
 219             }
 220         }
 221
 222         addNomRefExtension(state, name);
 223
 224         //add protologue
 225         String bhlLink = record.get("OutputBHLLink");
 226         if (isNotBlank(bhlLink)){
 227             URI uri;
 228             try {
 229                 uri = new URI(bhlLink);
 230                 Media media = Media.NewInstance(uri, null, null, null);
 231                 TaxonNameDescription desc = TaxonNameDescription.NewInstance(name);
 232                 desc.setTitleCache("Protologue for " + name.getNameCache(), true);
 233                 DescriptionElementBase elem = TextData.NewInstance(Feature.PROTOLOGUE());
 234                 elem.addMedia(media);
 235                 desc.addElement(elem);
 236             } catch (URISyntaxException e) {
 237                 logger.warn(line + "URI could not be parsed: " + e.getMessage());
 238             }
 239         }
 240
 241         //add tropicos identifier
 242         String tropicosId = record.get("OutputNameID");
 243         if (isNotBlank(tropicosId)){
 244             String tropicosIdTypeLabel = "Tropicos Name Identifier";
 245             UUID uuid = DefinedTerm.uuidTropicosNameIdentifier;
 246             TermVocabulary<DefinedTerm> voc = null;  //for now it goes to user defined voc
 247             DefinedTerm identifierType = this.getIdentiferType(state, uuid, tropicosIdTypeLabel, tropicosIdTypeLabel, null, voc);
 248             Identifier<Taxon> identifier = Identifier.NewInstance(tropicosId, identifierType);
 249             name.addIdentifier(identifier);
 250         }
 251
 252         name.addSource(makeOriginalSource(state));
 253
 254
 255         return name;
 256     }
 257
 258     /**
 259      * @param outputCollation
 260      * @return
 261      */
 262     private String[] makeVolumeDetail(String outputCollation) {
 263         if (outputCollation == null){
 264             return new String[0];
 265         }else{
 266             String[] split = outputCollation.split(":");
 267             return split;
 268         }
 269     }
 270
 271     /**
 272      * @param state
 273      * @param referencedName
 274      */
 275     private void addNomRefExtension(SimpleExcelTaxonImportState<CONFIG> state, IBotanicalName name) {
 276         String newExtensionStr = name.getFullTitleCache() + " - BORHIDI";
 277         UUID uuidNomRefExtension = MexicoConabioTransformer.uuidNomRefExtension;
 278         for (Extension extension : name.getExtensions()){
 279             if (extension.getType().getUuid().equals(uuidNomRefExtension)){
 280                 extension.setValue(extension.getValue() + "\n" + newExtensionStr);
 281                 return;
 282             }
 283         }
 284         String label = "Nomenclatural reference in Sources";
 285         String abbrev = "Nom. ref. src.";
 286         ExtensionType extensionType = getExtensionType(state, uuidNomRefExtension, label, label, abbrev);
 287         Extension.NewInstance((TaxonName)name, newExtensionStr, extensionType);
 288     }
 289
 290
 291 }