cdmlib-io/src/main/java/eu/etaxonomy/cdm/io/taxonx/TaxonXDescriptionImport.java

   1 /**
   2 * Copyright (C) 2007 EDIT
   3 * European Distributed Institute of Taxonomy
   4 * http://www.e-taxonomy.eu
   5 *
   6 * The contents of this file are subject to the Mozilla Public License Version 1.1
   7 * See LICENSE.TXT at the top of this package for the full license terms.
   8 */
   9
  10 package eu.etaxonomy.cdm.io.taxonx;
  11
  12 import java.util.HashMap;
  13 import java.util.Iterator;
  14 import java.util.List;
  15 import java.util.Map;
  16
  17 import org.apache.log4j.Logger;
  18 import org.jdom.Attribute;
  19 import org.jdom.Content;
  20 import org.jdom.Element;
  21 import org.jdom.Namespace;
  22 import org.jdom.Text;
  23 import org.springframework.stereotype.Component;
  24
  25 import eu.etaxonomy.cdm.api.application.CdmApplicationController;
  26 import eu.etaxonomy.cdm.api.service.ICommonService;
  27 import eu.etaxonomy.cdm.common.CdmUtils;
  28 import eu.etaxonomy.cdm.io.common.CdmImportBase;
  29 import eu.etaxonomy.cdm.io.common.IImportConfigurator;
  30 import eu.etaxonomy.cdm.model.common.Language;
  31 import eu.etaxonomy.cdm.model.common.OriginalSourceType;
  32 import eu.etaxonomy.cdm.model.common.TermVocabulary;
  33 import eu.etaxonomy.cdm.model.description.DescriptionElementBase;
  34 import eu.etaxonomy.cdm.model.description.Feature;
  35 import eu.etaxonomy.cdm.model.description.TaxonDescription;
  36 import eu.etaxonomy.cdm.model.description.TextData;
  37 import eu.etaxonomy.cdm.model.reference.Reference;
  38 import eu.etaxonomy.cdm.model.taxon.Taxon;
  39 import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException;
  40
  41
  42 /**
  43  * @author a.mueller
  44  * @created 29.07.2008
  45  */
  46 @Component
  47 public class TaxonXDescriptionImport
  48         extends CdmImportBase<TaxonXImportConfigurator, TaxonXImportState> {
  49
  50     private static final long serialVersionUID = 1957356490206741364L;
  51     private static final Logger logger = Logger.getLogger(TaxonXDescriptionImport.class);
  52
  53         public TaxonXDescriptionImport(){
  54                 super();
  55         }
  56
  57         @Override
  58     public boolean doCheck(TaxonXImportState state){
  59                 boolean result = true;
  60                 logger.warn("Checking for Facts not yet implemented");
  61                 //result &= checkArticlesWithoutJournal(bmiConfig);
  62
  63                 return result;
  64         }
  65
  66         public Map<Integer, Feature> fillFactCategories(IImportConfigurator config, CdmApplicationController cdmApp){
  67                 logger.warn("not yet implemented");
  68                 Map<Integer, Feature> featureMap = new HashMap<Integer, Feature>();
  69 //              IDescriptionService descriptionService = cdmApp.getDescriptionService();
  70 //              ITermService termService = cdmApp.getTermService();
  71 //
  72 //              Object source = config.getSource();
  73                 return featureMap;
  74         }
  75
  76         private String getDescriptionTitle(TaxonXImportState state){
  77                 String result = "Untitled";
  78                 Reference ref = state.getModsReference();
  79                 if (ref != null){
  80                         result = ref.getTitle();
  81                         if ( CdmUtils.isEmpty(result)){
  82                                 result = ref.getTitleCache();
  83                         }
  84                 }
  85                 return result;
  86         }
  87
  88         @Override
  89     public void doInvoke(TaxonXImportState state){
  90                 logger.debug("not yet fully implemented");
  91
  92                 TaxonXImportConfigurator txConfig = state.getConfig();
  93                 Element root = txConfig.getSourceRoot();
  94                 Namespace nsTaxonx = root.getNamespace();
  95
  96                 //Object source = config.getSource();
  97
  98                 logger.info("start make Descriptions ...");
  99
 100
 101                 //for testing only
 102                 Taxon taxon = getTaxon(txConfig);
 103                 if (taxon == null){
 104                         logger.warn("Taxon could not be found");
 105                         state.setUnsuccessfull();
 106                 }
 107
 108                 Reference modsReference = state.getModsReference();
 109                 if (modsReference == null){
 110                         modsReference = state.getConfig().getSourceReference();
 111                 }
 112
 113                 //unlazyDescription(txConfig, taxon);
 114                 TaxonDescription description = TaxonDescription.NewInstance();
 115                 description.setTitleCache(getDescriptionTitle(state), true);
 116                 if (modsReference != null){
 117                         description.addSource(OriginalSourceType.PrimaryTaxonomicSource, null, null, modsReference, null);
 118                 }
 119
 120                 Element elTaxonBody = root.getChild("taxonxBody", nsTaxonx);
 121                 Element elTreatment = elTaxonBody.getChild("treatment", nsTaxonx);
 122                 List<Element> elDivs = elTreatment.getChildren("div", nsTaxonx);
 123                 for (Element div : elDivs){
 124                         Attribute attrType = div.getAttribute("type", nsTaxonx);
 125                         String strType = attrType.getValue();
 126                         Feature feature = null;
 127                         try {
 128                                 feature = TaxonXTransformer.descriptionType2feature(strType);
 129                         } catch (UnknownCdmTypeException e) {
 130                                 feature = handleFeatureException(strType, e, txConfig);
 131                         }
 132                         String text = getText(div);
 133                         if (!"".equals(CdmUtils.Nz(text).trim())){
 134                                 // hibernate throws an exception when a string is longer than approx. 65500 chars.
 135                                 // for now we truncate any description text to 65500 characters.
 136                                 if(text.length() > 65500){
 137                                         text = text.substring(0, 65500) + "... [text truncated]";
 138                                         logger.warn("Truncation of text: description for taxon " + taxon.getTitleCache() + " was longer than 65500 characters.");
 139                                 }
 140
 141                                 DescriptionElementBase descriptionElement = TextData.NewInstance(text, Language.ENGLISH(), null);
 142                                 descriptionElement.setFeature(feature);
 143                                 description.addElement(descriptionElement);
 144
 145                                 //add reference
 146                                 if (modsReference != null){
 147                                         descriptionElement.addSource(OriginalSourceType.PrimaryTaxonomicSource, null, null, modsReference, null, null, null);
 148                                 }
 149                         }
 150
 151                 }
 152                 if (description.size() >0){
 153                         taxon.addDescription(description);
 154                         getTaxonService().save(taxon);
 155                 }
 156                 return;
 157         }
 158
 159         private Feature handleFeatureException(String strType, UnknownCdmTypeException e, TaxonXImportConfigurator txConfig){
 160                 Feature feature = null;
 161                 TermVocabulary<Feature> featureVoc = Feature.BIOLOGY_ECOLOGY().getVocabulary();
 162                 for (Feature oneFeature : featureVoc.getTerms()){
 163                         if (strType.equals(oneFeature.getLabel()) || strType.equals(oneFeature.getRepresentation(Language.DEFAULT() ).getText() )){
 164                                 feature = oneFeature;
 165                         }
 166                 }
 167
 168                 if (feature == null){
 169                         feature = Feature.NewInstance(strType, strType, null);
 170                         featureVoc.addTerm(feature);
 171                         getTermService().save(feature);
 172                         logger.warn(e.getMessage() + ". Feature was added to the feature vocabulary. " + getBracketSourceName(txConfig));
 173                 }
 174                 return feature;
 175         }
 176
 177
 178         private String getText(Element div){
 179                 String result = "";
 180                 Iterator<Content> it =div.getDescendants();
 181                 while (it.hasNext()){
 182                         Content next = it.next();
 183                         if (next instanceof Text){
 184                                 result += ((Text)next).getText();
 185                         }
 186                 }
 187                 return result;
 188         }
 189
 190         private Taxon getTaxon(TaxonXImportConfigurator config){
 191                 Taxon result;
 192 //              result =  Taxon.NewInstance(TaxonNameFactory.NewBotanicalInstance(null), null);
 193                 ICommonService commonService = getCommonService();
 194
 195                 String originalSourceId = config.getOriginalSourceId();
 196                 String namespace = config.getOriginalSourceTaxonNamespace();
 197                 result = (Taxon)commonService.getSourcedObjectByIdInSource(Taxon.class, originalSourceId , namespace);
 198                 if (result == null){
 199                         logger.warn("Taxon (id: " + originalSourceId + ", namespace: " + namespace + ") could not be found");
 200                 }
 201                 return result;
 202         }
 203
 204         private void unlazyDescription(TaxonXImportConfigurator config, Taxon taxon){
 205 //              logger.warn("Preliminary commented");  //used single Transaction for all import instead !
 206 //              TransactionStatus txStatus = config.getCdmAppController().startTransaction();
 207 //              ITaxonService taxonService = config.getCdmAppController().getTaxonService();
 208 //              taxonService.saveTaxon(taxon);
 209 //              taxon.getDescriptions().size();
 210 //              config.getCdmAppController().commitTransaction(txStatus);
 211         }
 212
 213         /* (non-Javadoc)
 214          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 215          */
 216         @Override
 217     protected boolean isIgnore(TaxonXImportState state){
 218                 return ! state.getConfig().isDoFacts();
 219         }
 220
 221         private String getBracketSourceName(TaxonXImportConfigurator config){
 222                 return "(" + config.getSourceNameString() + ")";
 223         }
 224
 225 }