app-import/src/main/java/eu/etaxonomy/cdm/app/wp6/diptera/DipteraCollectionImport.java

   1 // $Id$
   2 /**
   3 * Copyright (C) 2007 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.app.wp6.diptera;
  11
  12 import java.io.File;
  13 import java.io.FileInputStream;
  14 import java.io.InputStream;
  15 import java.io.InputStreamReader;
  16 import java.util.ArrayList;
  17 import java.util.HashMap;
  18 import java.util.List;
  19 import java.util.Map;
  20
  21 import org.apache.log4j.Logger;
  22 import org.springframework.transaction.TransactionStatus;
  23
  24 import au.com.bytecode.opencsv.CSVReader;
  25 import eu.etaxonomy.cdm.api.application.CdmApplicationController;
  26 import eu.etaxonomy.cdm.app.common.CdmDestinations;
  27 import eu.etaxonomy.cdm.common.CdmUtils;
  28 import eu.etaxonomy.cdm.database.DbSchemaValidation;
  29 import eu.etaxonomy.cdm.database.ICdmDataSource;
  30 import eu.etaxonomy.cdm.model.agent.Institution;
  31 import eu.etaxonomy.cdm.model.occurrence.Collection;
  32 import eu.etaxonomy.cdm.model.occurrence.Specimen;
  33 import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationBase;
  34
  35 /**
  36  * @author a.mueller
  37  * @date 07.04.2010
  38  *
  39  */
  40 public class DipteraCollectionImport {
  41         private static final Logger logger = Logger.getLogger(DipteraCollectionImport.class);
  42
  43         public static final File acronymsFile = new File("src/main/resources/collections/Acronyms.tab");
  44         //datasource for use from local main()
  45         static final ICdmDataSource cdmDestination = CdmDestinations.cdm_edit_diptera_a();
  46
  47
  48         public boolean invoke(ICdmDataSource dataSource) {
  49                 CdmApplicationController cdmApp = CdmApplicationController.NewInstance(dataSource, DbSchemaValidation.VALIDATE);
  50
  51                 //create collections
  52                 TransactionStatus tx = cdmApp.startTransaction();
  53                 Map<String, Collection> colletionMap = createCollections(cdmApp);
  54
  55                 //add collections to specimen
  56                 addCollectionsToSpecimen(cdmApp, colletionMap);
  57                 cdmApp.commitTransaction(tx);
  58
  59                 return true;
  60
  61         }
  62
  63
  64         /**
  65          * @param cdmApp
  66          * @param colletionMap
  67          */
  68         private void addCollectionsToSpecimen(CdmApplicationController cdmApp, Map<String, Collection> colletionMap) {
  69                 List<SpecimenOrObservationBase> specimens = (cdmApp.getOccurrenceService().list(Specimen.class, null, null, null, null));
  70                 for (SpecimenOrObservationBase specOrObservBase : specimens){
  71                         if (specOrObservBase instanceof Specimen){
  72                                 handleSingleSpecimen((Specimen)specOrObservBase, colletionMap);
  73                         }else{
  74                                 logger.warn("There are specimenOrObservationBase objects which are not of class Specimen. This is probably an error.");
  75                         }
  76                 }
  77                 cdmApp.getOccurrenceService().save(specimens);
  78         }
  79
  80
  81         /**
  82          * @param specimen
  83          * @param colletionMap
  84          */
  85         private void handleSingleSpecimen(Specimen specimen, Map<String, Collection> collectionMap) {
  86                 String titleCache = specimen.getTitleCache();
  87                 String collectionCode = getCollectionCode(titleCache);
  88                 if (CdmUtils.isEmpty(collectionCode)){
  89                         logger.warn("Collection code is empty for: " + titleCache);
  90                 }else{
  91                         Collection collection = collectionMap.get(collectionCode);
  92                         if (collection != null){
  93                                 specimen.setCollection(collection);
  94                         }else{
  95                                 logger.warn("Collection not found for code: " +  collectionCode + "; titleCache: " +  titleCache);
  96                         }
  97                 }
  98         }
  99
 100
 101         /**
 102          * @param titleCache
 103          * @return
 104          */
 105         private String getCollectionCode(String titleCache) {
 106                 String result = titleCache.trim();
 107                 result = replaceBracket(result);
 108                 result = replaceLastFullStop(result);
 109                 result = replaceLastQuestionMark(result);
 110                 result = parseLastUpperCase(result);
 111                 return result;
 112         }
 113
 114
 115         /**
 116          * @param result
 117          * @return
 118          */
 119         private String parseLastUpperCase(String string) {
 120                 String result = "";
 121                 String tmpString = string;
 122                 int pos = tmpString.lastIndexOf(" ");
 123                 if (pos>-1){
 124                         tmpString = tmpString.substring(pos+1);
 125                 }
 126                 while (tmpString.length() > 0){
 127                         int len = tmpString.length();
 128                         char lastChar = tmpString.charAt(len-1);
 129                         if (Character.isUpperCase( lastChar)){
 130                                 result = lastChar + result;
 131                         }else{
 132                                 if (result.length() > 0){
 133                                         logger.warn("Collection code is not space separated: " + string);
 134                                 }
 135                                 break;
 136                         }
 137                         //remove last character
 138                         tmpString = tmpString.substring(0, tmpString.length()-1);
 139                 }
 140                 return result;
 141         }
 142
 143
 144
 145         /**
 146          * @param result
 147          * @return
 148          */
 149         private String replaceLastQuestionMark(String string) {
 150                 if (string.endsWith("?")){
 151                         string = string.substring(0,string.length()-1).trim();
 152                 }
 153                 return string;
 154         }
 155
 156         /**
 157          * @param result
 158          * @return
 159          */
 160         private String replaceLastFullStop(String string) {
 161                 if (string.endsWith(".")){
 162                         string = string.substring(0,string.length()-1).trim();
 163                 }
 164                 return string;
 165         }
 166
 167
 168         /**
 169          * @param result
 170          * @return
 171          */
 172         private String replaceBracket(String string) {
 173                 if (string.endsWith("]")){
 174                         int pos  = string.indexOf("[");
 175                         if (pos >0){
 176                                 string = string.substring(0, pos).trim();
 177                         }else{
 178                                 logger.warn("Closing bracket has no opening bracket in: " + string);
 179                         }
 180                 }
 181                 return string;
 182         }
 183
 184
 185         /**
 186          * @param cdmApp
 187          */
 188         private Map<String, Collection> createCollections(CdmApplicationController cdmApp) {
 189                 Map<String, Collection> collectionMap = new HashMap<String, Collection>();
 190                 List<String[]> lines = getLines();
 191                 for (String[] line:lines){
 192                         Collection collection = makeLine(line);
 193                         collectionMap.put(collection.getCode(), collection);
 194                 }
 195                 cdmApp.getCollectionService().save(collectionMap.values());
 196 //                      for (Collection collection: collectionMap.values()){
 197 //                              System.out.println(collection.getTitleCache());
 198 //                      }
 199                 return collectionMap;
 200         }
 201
 202
 203         private Collection makeLine(String[] line) {
 204                 String code = line[0];
 205                 String instituteName = line[1];
 206                 String lowerInstitutionName = line[2];
 207                 String higherInstitutionName = line[3];
 208                 String location = line[4];
 209                 String country = line[5];
 210                 //create objects
 211                 Collection collection = Collection.NewInstance();
 212                 collection.setCode(code);
 213                 Institution institution = Institution.NewInstance();
 214                 institution.setCode(code);
 215
 216                 institution.setName(instituteName);
 217
 218                 if (CdmUtils.isNotEmpty(lowerInstitutionName)){
 219                         Institution lowerInstitution = Institution.NewInstance();
 220                         lowerInstitution.setName(lowerInstitutionName);
 221                         lowerInstitution.setIsPartOf(institution);
 222                 }
 223
 224                 if (CdmUtils.isNotEmpty(higherInstitutionName)){
 225                         Institution higherInstitution = Institution.NewInstance();
 226                         higherInstitution.setName(higherInstitutionName);
 227                         institution.setIsPartOf(higherInstitution);
 228                 }
 229
 230                 collection.setInstitute(institution);
 231                 String locationAndCountry = CdmUtils.concat("/", location, country);
 232                 collection.setTownOrLocation(locationAndCountry);
 233
 234                 String titleCache = CdmUtils.concat(", ", new String[]{instituteName, lowerInstitutionName, higherInstitutionName, location, country});
 235                 collection.setTitleCache(titleCache, true);
 236
 237                 return collection;
 238         }
 239
 240
 241
 242
 243         private List<String[]> getLines() {
 244                 List<String[]> result = new ArrayList<String[]>();
 245
 246                 try {
 247                         InputStream inStream = new FileInputStream(acronymsFile);
 248                         InputStreamReader inputStreamReader = new InputStreamReader(inStream, "UTF8");
 249                         CSVReader reader = new CSVReader(inputStreamReader, '\t');
 250                         String [] nextLine = reader.readNext();
 251
 252
 253                         while ((nextLine = reader.readNext()) != null) {
 254                                 if (nextLine.length == 0){
 255                                         continue;
 256                                 }
 257                                 result.add(nextLine);
 258                         }
 259                         return result;
 260                 } catch (Exception e) {
 261                         logger.error(e + " " + e.getCause() + " " + e.getMessage());
 262                         for(StackTraceElement ste : e.getStackTrace()) {
 263                                 logger.error(ste);
 264                         }
 265                         throw new RuntimeException(e);
 266                 }
 267         }
 268
 269
 270
 271
 272
 273         /**
 274          * @param args
 275          */
 276         public static void main(String[] args) {
 277                 try {
 278                         DipteraCollectionImport collectionImport = new DipteraCollectionImport();
 279                         collectionImport.invoke(cdmDestination);
 280 //                      String titleCache = "Peru. Mouth of Rio Pachitea. ST 2R SMT. [fig. of male abdomen]";
 281 //                      String collectionCode = collectionImport.getCollectionCode(titleCache);
 282 //                      System.out.println(collectionCode);
 283                 } catch (Exception e) {
 284                         e.printStackTrace();
 285                         System.exit(-1);
 286                 }
 287         }
 288
 289 }