app-import/src/main/java/eu/etaxonomy/cdm/io/edaphobase/EdaphobaseReferenceImport.java

   1 // $Id$
   2 /**
   3 * Copyright (C) 2015 EDIT
   4 * European Distributed Institute of Taxonomy
   5 * http://www.e-taxonomy.eu
   6 *
   7 * The contents of this file are subject to the Mozilla Public License Version 1.1
   8 * See LICENSE.TXT at the top of this package for the full license terms.
   9 */
  10 package eu.etaxonomy.cdm.io.edaphobase;
  11
  12 import java.sql.ResultSet;
  13 import java.sql.SQLException;
  14 import java.util.HashMap;
  15 import java.util.HashSet;
  16 import java.util.Map;
  17 import java.util.Set;
  18
  19 import org.apache.commons.lang3.StringUtils;
  20 import org.apache.log4j.Logger;
  21 import org.springframework.stereotype.Component;
  22
  23 import eu.etaxonomy.cdm.common.DOI;
  24 import eu.etaxonomy.cdm.io.common.IPartitionedIO;
  25 import eu.etaxonomy.cdm.io.common.ImportHelper;
  26 import eu.etaxonomy.cdm.io.common.ResultSetPartitioner;
  27 import eu.etaxonomy.cdm.model.common.CdmBase;
  28 import eu.etaxonomy.cdm.model.common.TimePeriod;
  29 import eu.etaxonomy.cdm.model.reference.Reference;
  30 import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
  31
  32 /**
  33  * @author a.mueller
  34  * @date 18.12.2015
  35  *
  36  */
  37 @Component
  38 public class EdaphobaseReferenceImport extends EdaphobaseImportBase {
  39     private static final long serialVersionUID = 6895687693249076160L;
  40
  41     @SuppressWarnings("unused")
  42     private static final Logger logger = Logger.getLogger(EdaphobaseReferenceImport.class);
  43
  44     private static final String tableName = "lit_document";
  45
  46     private static final String pluralString = "documents";
  47 //
  48 //    private static final Object AUTHOR_NAMESPACE = "tax_author_name";
  49
  50     /**
  51      * @param tableName
  52      * @param pluralString
  53      */
  54     public EdaphobaseReferenceImport() {
  55         super(tableName, pluralString);
  56     }
  57
  58     @Override
  59     protected String getIdQuery(EdaphobaseImportState state) {
  60         return    " SELECT DISTINCT document_id "
  61                 + " FROM lit_document ld INNER JOIN tax_taxon t ON t.tax_document = ld.document_id "
  62                 + " UNION "
  63                 + " SELECT DISTINCT pd.document_id "
  64                 + " FROM lit_document ld INNER JOIN tax_taxon t ON t.tax_document = ld.document_id "
  65                 + " INNER JOIN lit_document pd ON pd.document_id = ld.parent_document_fk_document_id "
  66                 + " ORDER BY document_id ";
  67     }
  68
  69     @Override
  70     protected String getRecordQuery(EdaphobaseImportConfigurator config) {
  71         String result = " SELECT * "
  72                 + " FROM lit_document ld "
  73                 + " WHERE ld.document_id IN (@IDSET)";
  74         result = result.replace("@IDSET", IPartitionedIO.ID_LIST_TOKEN);
  75         return result;
  76     }
  77
  78     @Override
  79     public boolean doPartition(ResultSetPartitioner partitioner, EdaphobaseImportState state) {
  80         ResultSet rs = partitioner.getResultSet();
  81         Set<Reference> referencesToSave = new HashSet<>();
  82         try {
  83             while (rs.next()){
  84
  85                 handleSingleReference(state, rs, referencesToSave);
  86
  87             }
  88         } catch (SQLException e) {
  89             // TODO Auto-generated catch block
  90             e.printStackTrace();
  91         }
  92
  93         getReferenceService().saveOrUpdate(referencesToSave);
  94
  95         return true;
  96     }
  97
  98     /**
  99      * @param state
 100      * @param rs
 101      * @param referencesToSave
 102      * @throws SQLException
 103      */
 104     private void handleSingleReference(EdaphobaseImportState state, ResultSet rs, Set<Reference> referencesToSave) throws SQLException {
 105         Integer id = nullSafeInt(rs, "document_id");
 106         String dtype = rs.getString("dtype");
 107         String issue = rs.getString("issue");
 108         String orderer = rs.getString("orderer");
 109         String place = rs.getString("place");
 110         Integer pageFrom = nullSafeInt(rs, "page_from");
 111         Integer pageTo = nullSafeInt(rs, "page_to");
 112         String subtitle = rs.getString("subtitle");
 113         Integer year = nullSafeInt(rs, "year");
 114         String isbn = rs.getString("isbn");
 115         //refers_to_literature
 116         //refers_to_collection
 117         //refers_to_observation
 118         String remark = rs.getString("remark");
 119         String volume = rs.getString("volume");
 120         //abbreviation (no record)
 121         String title = rs.getString("title");
 122         String issn = rs.getString("issn");
 123         //circulation //2 records
 124         String keywords = rs.getString("keywords");
 125         String abstractt = rs.getString("abstract");
 126         String parallel_title = rs.getString("parallel_title");
 127         //language_fk_language_id
 128         //document_type_fk_document_type_id
 129         //editor_fk_person_id
 130         Integer editorFk = nullSafeInt(rs, "editor_fk_person_id");
 131
 132 //        Integer parentFk = nullSafeInt(rs, "parent_document_fk_document_id");
 133         //publisher_fk_publisher_id
 134         //deleted
 135         //chapter_no
 136         //versionfield
 137         String doi = rs.getString("doi");
 138         String displayString = rs.getString("display_string");
 139         //aquisistion_date, aquisition_type, adoption_date, ex_colletion, barcode_prefix, barcode_org_prefix
 140         //barcode_type, collection_status, barcode, typus_form,
 141
 142         //taxon_for_scope, taxon_is_scope
 143         //language_fk, document_type_backup
 144
 145         Integer documentType = nullSafeInt(rs, "document_type");
 146         //normalized_title, normalized_abk_official_remark
 147
 148         Reference<?> ref = makeReferenceType(documentType, dtype);
 149         ref.setTitle(title);
 150         ref.setPlacePublished(place);
 151         ref.setIssn(issn);
 152         ref.setIsbn(isbn);
 153         if (pageFrom != null || pageTo != null){
 154             String pageStr = pageFrom == null ? "" : String.valueOf(pageFrom);
 155             pageStr = pageTo == null ? pageStr : "-" + pageTo;
 156             ref.setPages(pageStr);
 157         }
 158         if (year != null){
 159             ref.setDatePublished(TimePeriod.NewInstance(year));
 160         }
 161         ref.setVolume(volume);
 162         ref.setReferenceAbstract(abstractt);
 163         if (StringUtils.isNotBlank(doi)){
 164             try {
 165                 String doiStr = doi;
 166                 if (doiStr.startsWith("dx.doi.org/")){
 167                     doiStr = doiStr.substring(11);
 168                 }
 169                 ref.setDoi(DOI.fromString(doiStr));
 170             } catch (IllegalArgumentException e) {
 171                 logger.warn("DOI could not be parsed: " + doi);
 172             }
 173         }
 174         ref.setEdition(issue);
 175
 176         //id
 177         ImportHelper.setOriginalSource(ref, state.getTransactionalSourceReference(), id, REFERENCE_NAMESPACE);
 178
 179         referencesToSave.add(ref);
 180     }
 181
 182
 183     /**
 184      * @param documentType
 185      * @return
 186      */
 187     private Reference<?> makeReferenceType(Integer documentType, String dtype) {
 188         if (documentType == 11914){
 189             return ReferenceFactory.newArticle();
 190         } else if (documentType == 11916){
 191             return ReferenceFactory.newBook();
 192         } else if (documentType == 11915){
 193             return ReferenceFactory.newPrintSeries();
 194         } else if (documentType == 11913){
 195             return ReferenceFactory.newJournal();
 196         } else if (documentType == 11917){
 197             return ReferenceFactory.newBookSection();
 198         } else if (documentType == 11912 || documentType == 11919 || documentType == 11924 ){
 199             Reference<?> ref = ReferenceFactory.newGeneric();
 200             return ref;
 201         } else {
 202             throw new RuntimeException("DocumentType not yet supported: " + documentType + ", " + dtype);
 203         }
 204     }
 205
 206     @Override
 207     public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs,
 208             EdaphobaseImportState state) {
 209         Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<>();
 210 //        Map<String, TeamOrPersonBase<?>> authorMap = new HashMap<>();
 211 //        Set<String> authorSet = new HashSet<>();
 212 //        try {
 213 //            while (rs.next()){
 214 //                String authorStr = rs.getString("tax_author_name");
 215 //                authorSet.add(authorStr);
 216 //            }
 217 //        } catch (SQLException e) {
 218 //            e.printStackTrace();
 219 //        }
 220 //
 221 //        //Authors
 222 //        Set<UUID> uuidSet = new HashSet<>();
 223 //        for (String authorStr : authorSet){
 224 //            UUID uuid = state.getAuthorUuid(authorStr);
 225 //            uuidSet.add(uuid);
 226 //        }
 227 //        List<TeamOrPersonBase<?>> authors = (List)getAgentService().find(uuidSet);
 228 //        Map<UUID, TeamOrPersonBase<?>> authorUuidMap = new HashMap<>();
 229 //        for (TeamOrPersonBase<?> author : authors){
 230 //            authorUuidMap.put(author.getUuid(), author);
 231 //        }
 232 //
 233 //        for (String authorStr : authorSet){
 234 //            UUID uuid = state.getAuthorUuid(authorStr);
 235 //            TeamOrPersonBase<?> author = authorUuidMap.get(uuid);
 236 //            authorMap.put(authorStr, author);
 237 //        }
 238 //        result.put(AUTHOR_NAMESPACE, authorMap);
 239
 240         return result;
 241     }
 242
 243
 244
 245     @Override
 246     protected boolean doCheck(EdaphobaseImportState state) {
 247         return false;
 248     }
 249
 250     @Override
 251     protected boolean isIgnore(EdaphobaseImportState state) {
 252         return ! state.getConfig().isDoReferences();
 253     }
 254
 255 }