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.CdmIoBase;
  29 import eu.etaxonomy.cdm.io.common.ICdmIO;
  30 import eu.etaxonomy.cdm.io.common.IImportConfigurator;
  31 import eu.etaxonomy.cdm.model.common.Language;
  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.ReferenceBase;
  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  * @version 1.0
  46  */
  47 @Component
  48 public class TaxonXDescriptionImport extends CdmIoBase<TaxonXImportState> implements ICdmIO<TaxonXImportState> {
  49         private static final Logger logger = Logger.getLogger(TaxonXDescriptionImport.class);
  50
  51         public TaxonXDescriptionImport(){
  52                 super();
  53         }
  54
  55         public boolean doCheck(TaxonXImportState state){
  56                 boolean result = true;
  57                 logger.warn("Checking for Facts not yet implemented");
  58                 //result &= checkArticlesWithoutJournal(bmiConfig);
  59
  60                 return result;
  61         }
  62
  63         public Map<Integer, Feature> fillFactCategories(IImportConfigurator config, CdmApplicationController cdmApp){
  64                 logger.warn("not yet implemented");
  65                 Map<Integer, Feature> featureMap = new HashMap<Integer, Feature>();
  66 //              IDescriptionService descriptionService = cdmApp.getDescriptionService();
  67 //              ITermService termService = cdmApp.getTermService();
  68 //
  69 //              Object source = config.getSource();
  70                 return featureMap;
  71         }
  72
  73         private String getDescriptionTitle(TaxonXImportState state){
  74                 String result = "Untitled";
  75                 ReferenceBase<?> ref = state.getModsReference();
  76                 if (ref != null){
  77                         result = ref.getTitle();
  78                         if ( CdmUtils.isEmpty(result)){
  79                                 result = ref.getTitleCache();
  80                         }
  81                 }
  82                 return result;
  83         }
  84
  85         public boolean doInvoke(TaxonXImportState state){
  86                 logger.debug("not yet fully implemented");
  87
  88                 TaxonXImportConfigurator txConfig = state.getConfig();
  89                 Element root = txConfig.getSourceRoot();
  90                 Namespace nsTaxonx = root.getNamespace();
  91
  92                 //Object source = config.getSource();
  93
  94                 logger.info("start make Descriptions ...");
  95
  96
  97                 //for testing only
  98                 Taxon taxon = getTaxon(txConfig);
  99                 if (taxon == null){
 100                         logger.warn("Taxon could not be found");
 101                         return false;
 102                 }
 103                 //unlazyDescription(txConfig, taxon);
 104                 TaxonDescription description = TaxonDescription.NewInstance();
 105                 description.setTitleCache(getDescriptionTitle(state));
 106
 107                 Element elTaxonBody = root.getChild("taxonxBody", nsTaxonx);
 108                 Element elTreatment = elTaxonBody.getChild("treatment", nsTaxonx);
 109                 List<Element> elDivs = elTreatment.getChildren("div", nsTaxonx);
 110                 for (Element div : elDivs){
 111                         Attribute attrType = div.getAttribute("type", nsTaxonx);
 112                         String strType = attrType.getValue();
 113                         Feature feature = null;
 114                         try {
 115                                 feature = TaxonXTransformer.descriptionType2feature(strType);
 116                         } catch (UnknownCdmTypeException e) {
 117                                 feature = handleFeatureException(strType, e, txConfig);
 118                         }
 119                         String text = getText(div);
 120                         if (!"".equals(CdmUtils.Nz(text).trim())){
 121                                 // FIXME hibernate throws an exception when a string is longer than approx. 4000 chars.
 122                                 // for now we truncate any description text to 4000 characters.
 123                                 if(text.length() > 65500){
 124                                         text = text.substring(0, 65500) + "... [text truncated]";
 125                                         logger.warn("FIXME - Truncation of text occurred.");
 126                                 }
 127
 128                                 DescriptionElementBase descriptionElement = TextData.NewInstance(text, Language.ENGLISH(), null);
 129                                 descriptionElement.setFeature(feature);
 130                                 description.addElement(descriptionElement);
 131
 132                                 //add reference
 133                                 if (state.getModsReference() != null){
 134                                         descriptionElement.addSource(null, null, state.getModsReference(), null, null, null);
 135                                 }
 136                         }
 137
 138                 }
 139                 if (description.size() >0){
 140                         taxon.addDescription(description);
 141                         getTaxonService().save(taxon);
 142                 }
 143                 return true;
 144         }
 145
 146         private Feature handleFeatureException(String strType, UnknownCdmTypeException e, TaxonXImportConfigurator txConfig){
 147                 Feature feature = null;
 148                 TermVocabulary<Feature> featureVoc = Feature.BIOLOGY_ECOLOGY().getVocabulary();
 149                 for (Feature oneFeature : featureVoc.getTerms()){
 150                         if (strType.equals(oneFeature.getLabel()) || strType.equals(oneFeature.getRepresentation(Language.DEFAULT() ).getText() )){
 151                                 feature = oneFeature;
 152                         }
 153                 }
 154
 155                 if (feature == null){
 156                         feature = Feature.NewInstance(strType, strType, null);
 157                         featureVoc.addTerm(feature);
 158                         getTermService().save(feature);
 159                         logger.warn(e.getMessage() + ". Feature was added to the feature vocabulary. " + getBracketSourceName(txConfig));
 160                 }
 161                 return feature;
 162         }
 163
 164
 165         private String getText(Element div){
 166                 String result = "";
 167                 Iterator<Content> it =div.getDescendants();
 168                 while (it.hasNext()){
 169                         Content next = (Content)it.next();
 170                         if (next instanceof Text){
 171                                 result += ((Text)next).getText();
 172                         }
 173                 }
 174                 return result;
 175         }
 176
 177         private Taxon getTaxon(TaxonXImportConfigurator config){
 178                 Taxon result;
 179 //              result =  Taxon.NewInstance(BotanicalName.NewInstance(null), null);
 180                 ICommonService commonService = getCommonService();
 181
 182                 String originalSourceId = config.getOriginalSourceId();
 183                 String namespace = config.getOriginalSourceTaxonNamespace();
 184                 result = (Taxon)commonService.getSourcedObjectByIdInSource(Taxon.class, originalSourceId , namespace);
 185                 if (result == null){
 186                         logger.warn("Taxon (id: " + originalSourceId + ", namespace: " + namespace + ") could not be found");
 187                 }
 188                 return result;
 189         }
 190
 191         private void unlazyDescription(TaxonXImportConfigurator config, Taxon taxon){
 192 //              logger.warn("Preliminary commented");  //used single Transaction for all import instead !
 193 //              TransactionStatus txStatus = config.getCdmAppController().startTransaction();
 194 //              ITaxonService taxonService = config.getCdmAppController().getTaxonService();
 195 //              taxonService.saveTaxon(taxon);
 196 //              taxon.getDescriptions().size();
 197 //              config.getCdmAppController().commitTransaction(txStatus);
 198         }
 199
 200         /* (non-Javadoc)
 201          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 202          */
 203         protected boolean isIgnore(TaxonXImportState state){
 204                 return ! state.getConfig().isDoFacts();
 205         }
 206
 207         private String getBracketSourceName(TaxonXImportConfigurator config){
 208                 return "(" + config.getSourceNameString() + ")";
 209         }
 210
 211 }