cdm-pesi/src/main/java/eu/etaxonomy/cdm/io/pesi/faunaEuropaea/FaunaEuropaeaAuthorImport.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 package eu.etaxonomy.cdm.io.pesi.faunaEuropaea;
  10
  11 import java.sql.ResultSet;
  12 import java.sql.SQLException;
  13 import java.util.Collection;
  14 import java.util.Map;
  15
  16 import org.apache.log4j.Logger;
  17 import org.springframework.stereotype.Component;
  18 import org.springframework.transaction.TransactionStatus;
  19
  20 import eu.etaxonomy.cdm.io.common.ICdmIO;
  21 import eu.etaxonomy.cdm.io.common.ImportHelper;
  22 import eu.etaxonomy.cdm.io.common.MapWrapper;
  23 import eu.etaxonomy.cdm.io.common.Source;
  24 import eu.etaxonomy.cdm.model.agent.Person;
  25 import eu.etaxonomy.cdm.model.agent.Team;
  26 import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
  27 import eu.etaxonomy.cdm.model.common.CdmBase;
  28
  29
  30 /**
  31  * @author a.babadshanjan
  32  * @created 12.05.2009
  33  * @version 1.0
  34  */
  35 @Component
  36 public class FaunaEuropaeaAuthorImport extends FaunaEuropaeaImportBase {
  37         private static final Logger logger = Logger.getLogger(FaunaEuropaeaAuthorImport.class);
  38
  39         private static int modCount = 1000;
  40         private final static String authorSeparator = ", ";
  41         private final static String lastAuthorSeparator = " & ";
  42
  43
  44         /* (non-Javadoc)
  45          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#doCheck(eu.etaxonomy.cdm.io.common.IImportConfigurator)
  46          */
  47         @Override
  48         protected boolean doCheck(FaunaEuropaeaImportState state){
  49                 boolean result = true;
  50                 logger.warn("No checking for Authors not implemented");
  51
  52                 return result;
  53         }
  54
  55         /* (non-Javadoc)
  56          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#doInvoke(eu.etaxonomy.cdm.io.common.IImportConfigurator, eu.etaxonomy.cdm.api.application.CdmApplicationController, java.util.Map)
  57          */
  58         @Override
  59         protected void doInvoke(FaunaEuropaeaImportState state){
  60                 /*
  61                 logger.warn("Start author doInvoke");
  62                 ProfilerController.memorySnapshot();
  63                 */
  64                 if (!state.getConfig().isDoAuthors()){
  65                         return;
  66                 }
  67                 Map<String, MapWrapper<? extends CdmBase>> stores = state.getStores();
  68                 MapWrapper<TeamOrPersonBase<?>> authorStore = (MapWrapper<TeamOrPersonBase<?>>)stores.get(ICdmIO.TEAM_STORE);
  69                 TransactionStatus txStatus = null;
  70
  71                 FaunaEuropaeaImportConfigurator fauEuConfig = state.getConfig();
  72                 Source source = fauEuConfig.getSource();
  73
  74                 String namespace = "AuthorTeam";
  75
  76                 if(logger.isInfoEnabled()) { logger.info("Start making Authors..."); }
  77
  78                 try {
  79
  80                         String strQuery =
  81                                 " SELECT *  " +
  82                                 " FROM author " ;
  83                         ResultSet rs = source.getResultSet(strQuery) ;
  84
  85                         int i = 0;
  86                         while (rs.next()) {
  87
  88                                 if ((i++ % modCount) == 0 && i!= 1 ) {
  89                                         if(logger.isDebugEnabled()) {
  90                                                 logger.debug("Authors retrieved: " + (i-1));
  91                                         }
  92                                 }
  93
  94                                 int authorId = rs.getInt("aut_id");
  95                                 String authorName = rs.getString("aut_name");
  96
  97                                 String auctWithNecRegEx = "\\bauct\\b\\.?.*\\bnec\\b\\.?.*";
  98                                 String necAuctRegEx = "\\bnec\\b\\.?.*\\bauct\\b\\.?.*";
  99
 100                                 boolean auctWithNecFound = expressionMatches(auctWithNecRegEx, authorName);
 101                                 boolean necAuctFound = expressionMatches(necAuctRegEx, authorName);
 102                                 if (auctWithNecFound){
 103                                         logger.debug("authorName before auct nec string is removed" + authorName);
 104                                         authorName = authorName.substring(expressionEnd("nec\\.?", authorName)+1, authorName.length());
 105                                         logger.debug("authorName after auct nec string is removed" + authorName);
 106                                 }
 107
 108                                 if (necAuctFound){
 109                                         logger.debug("authorName before nec auct string is removed" + authorName);
 110                                         authorName = authorName.substring(0, authorName.indexOf("nec")-1);
 111                                         logger.debug("authorName before nec auct string is removed" + authorName);
 112                                 }
 113                                 TeamOrPersonBase<?> author = null;
 114
 115                                 try {
 116                                     author = parseAuthorString(authorName);
 117                                     ImportHelper.setOriginalSource(author, fauEuConfig.getSourceReference(), authorId, namespace);
 118
 119                                         if (!authorStore.containsId(authorId)) {
 120                                                 authorStore.put(authorId, author);
 121                                                 if (logger.isDebugEnabled()) { logger.debug("Stored author (" + authorId + ") " + authorName); }
 122                                         } else {
 123                                                 logger.warn("Not imported author with duplicated aut_id (" + authorId + ") " + authorName);
 124                                         }
 125                                 } catch (Exception e) {
 126                                         logger.warn("An exception occurred when creating author with id " + authorId + ". Author could not be saved.");
 127                                 }
 128                         }
 129
 130                         if(logger.isInfoEnabled()) { logger.info("Saving authors ..."); }
 131
 132                         txStatus = startTransaction();
 133
 134                         // save authors
 135                         getAgentService().save((Collection)authorStore.objects());
 136
 137                         commitTransaction(txStatus);
 138
 139                         if(logger.isInfoEnabled()) { logger.info("End making authors ..."); }
 140
 141                         return;
 142
 143                 } catch (SQLException e) {
 144                         logger.error("SQLException:" +  e);
 145                         state.setUnsuccessfull();
 146                 }
 147         }
 148
 149         /* (non-Javadoc)
 150          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 151          */
 152         @Override
 153     protected boolean isIgnore(FaunaEuropaeaImportState state){
 154                 return ! state.getConfig().isDoAuthors();
 155         }
 156
 157         public static TeamOrPersonBase<?> parseAuthorString(String authorName){
 158             TeamOrPersonBase<?> author = null;
 159             String[] teamMembers = authorName.split(authorSeparator);
 160         String lastMember;
 161         String[] lastMembers;
 162         Person teamMember;
 163         if (teamMembers.length>1){
 164             lastMember = teamMembers[teamMembers.length -1];
 165             lastMembers = lastMember.split(lastAuthorSeparator);
 166             teamMembers[teamMembers.length -1] = "";
 167             author = Team.NewInstance();
 168             for(String member:teamMembers){
 169                 if (!member.equals("")){
 170                     teamMember = Person.NewInstance();
 171                     teamMember.setTitleCache(member, true);
 172                    ((Team)author).addTeamMember(teamMember);
 173                 }
 174             }
 175             if (lastMembers != null){
 176                 for(String member:lastMembers){
 177                    teamMember = Person.NewInstance();
 178                    teamMember.setTitleCache(member, true);
 179                    ((Team)author).addTeamMember(teamMember);
 180                 }
 181             }
 182
 183         } else {
 184             teamMembers = authorName.split(lastAuthorSeparator);
 185             if (teamMembers.length>1){
 186                 author = Team.NewInstance();
 187                 for(String member:teamMembers){
 188                   teamMember = Person.NewInstance();
 189                   teamMember.setTitleCache(member, true);
 190                   ((Team)author).addTeamMember(teamMember);
 191
 192                 }
 193             }else{
 194                 author = Person.NewInstance();
 195                 author.setTitleCache(authorName, true);
 196             }
 197         }
 198         author.getTitleCache();
 199         return author;
 200         }
 201
 202 }