cdmlib-ext/src/main/java/eu/etaxonomy/cdm/ext/openurl/MobotOpenUrlResponseSchemaAdapter.java

   1 /**
   2 * Copyright (C) 2009 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.ext.openurl;
  10
  11 import java.io.IOException;
  12 import java.io.InputStream;
  13 import java.io.InputStreamReader;
  14 import java.io.Reader;
  15 import java.net.URISyntaxException;
  16 import java.util.ArrayList;
  17 import java.util.List;
  18
  19 import javax.xml.parsers.ParserConfigurationException;
  20 import javax.xml.parsers.SAXParser;
  21 import javax.xml.parsers.SAXParserFactory;
  22
  23 import org.xml.sax.Attributes;
  24 import org.xml.sax.InputSource;
  25 import org.xml.sax.SAXException;
  26 import org.xml.sax.helpers.DefaultHandler;
  27
  28 import eu.etaxonomy.cdm.common.URI;
  29 import eu.etaxonomy.cdm.ext.common.SchemaAdapterBase;
  30 import eu.etaxonomy.cdm.model.agent.Person;
  31 import eu.etaxonomy.cdm.model.agent.Team;
  32 import eu.etaxonomy.cdm.model.common.VerbatimTimePeriod;
  33 import eu.etaxonomy.cdm.model.reference.Reference;
  34
  35 /**
  36  * @author a.kohlbecker
  37  * @since 25.08.2010
  38  */
  39 public class MobotOpenUrlResponseSchemaAdapter extends SchemaAdapterBase<Reference>{
  40
  41         static URI identifier = null;
  42
  43         @Override
  44         public URI getIdentifier() {
  45                 return identifier;
  46         }
  47
  48         @Override
  49         public String getShortName() {
  50                 return "MOBOT.OpenUrl.Utilities.OpenUrlResponse";
  51         }
  52
  53         @Override
  54         public List<Reference> getCmdEntities(InputStream inputStream) throws IOException {
  55
  56                 SAXParserFactory factory = SAXParserFactory.newInstance();
  57             factory.setNamespaceAware(true);
  58             SAXParser parser = null;
  59                 try {
  60                         parser = factory.newSAXParser();
  61                 } catch (ParserConfigurationException e) {
  62                         logger.error(e);
  63                 } catch (SAXException e) {
  64                         logger.error(e);
  65                 }
  66
  67                 OpenUrlResponseHandler handler = new OpenUrlResponseHandler();
  68
  69             try {
  70                 if(parser != null){
  71                         Reader reader = new InputStreamReader(inputStream, "UTF-8");
  72                         InputSource inputSource = new InputSource(reader);
  73                         parser.parse(inputSource, handler);
  74                         if(handler.status != ResponseStatus.Success){
  75                                 throw new IOException("MOBOT.OpenUrl.Utilities.OpenUrlResponse - Status:" + handler.status.toString() + (handler.message != null ? handler.message : ""));
  76                         }
  77                 } else {
  78                         logger.error("parser is null");
  79                 }
  80                 } catch (SAXException e) {
  81                         logger.error(e);
  82                 }
  83
  84
  85                 return handler.referenceList;
  86         }
  87
  88         class OpenUrlResponseHandler extends DefaultHandler {
  89
  90                 /*
  91                  * Fields of OpenUrlResponse
  92                  *  see http://code.google.com/p/bhl-bits/source/browse/trunk/portal/OpenUrlUtilities/OpenUrlResponse.cs
  93                  */
  94                 private static final String OPENURL_RESPONSE = "OpenUrlResponse";
  95                 private static final String STATUS = "Status";
  96                 private static final String MESSAGE = "Message";
  97                 private static final String CITATIONS = "citations";
  98                 private static final String OPENURL_RESPONSE_CITATION = "OpenUrlResponseCitation";
  99
 100                 /*
 101                  * Fields of OpenUrlResponseCitation
 102                  *  see http://code.google.com/p/bhl-bits/source/browse/trunk/portal/OpenUrlUtilities/OpenUrlResponseCitation.cs
 103                  */
 104
 105                 /**
 106                  * references the specific page in the title
 107                  */
 108                 private static final String URL = "Url";
 109                 /**
 110                  * references the according entry in the bibliography
 111                  */
 112                 private static final String ITEM_URL = "ItemUrl";
 113
 114                 /**
 115                  * references the specific book or journal, that is to the front page
 116                  */
 117                 private static final String TITLE_URL = "TitleUrl";
 118                 private static final String TITLE = "Title";
 119                 private static final String STITLE = "STitle";
 120                 /**
 121                  * seems to contain the type of the reference : book
 122                  */
 123                 private static final String GENRE = "Genre";
 124                 private static final String AUTHORS = "Authors";
 125                 private static final String SUBJECTS = "Subjects";
 126                 private static final String PUBLISHER_NAME = "PublisherName";
 127                 private static final String PUBLISHER_PLACE = "PublisherPlace";
 128                 private static final String DATE = "Date";
 129                 private static final String VOLUME = "Volume";
 130                 private static final String EDITION = "Edition";
 131                 private static final String PUBLICATION_FREQUENCY = "PublicationFrequency";
 132                 private static final String LANGUAGE = "Language";
 133                 private static final String OCLC = "Oclc";
 134                 private static final String LCCN = "Lccn";
 135                 private static final String ISSN = "Issn";
 136                 private static final String ATITLE = "ATitle";
 137                 private static final String SPAGE = "SPage";
 138                 private static final String EPAGE = "EPage";
 139                 private static final String PAGES = "Pages";
 140
 141                 List<Reference> referenceList = new ArrayList<>();
 142
 143                 OpenUrlReference reference = null;
 144
 145                 ResponseStatus status = null;
 146                 Team authorship = null;
 147                 String message = null;
 148
 149                 String elementName = null;
 150                 private String elementNameToStore;
 151                 private StringBuilder textBuffer = new StringBuilder();
 152
 153                 @Override
 154                 public void startElement(String uri, String localName,
 155                                 String qName, Attributes attributes) throws SAXException {
 156
 157                         if (qName.equals(OPENURL_RESPONSE)) {
 158                                 logger.debug("Start " + OPENURL_RESPONSE + "; ");
 159                                 status = ResponseStatus.Undefined; // indicates that the OPENURL_RESPONSE element has ben detected
 160                         } else if (status != null && qName.equals(OPENURL_RESPONSE_CITATION)) {
 161                                 reference = new OpenUrlReference();
 162                         } else if (reference != null && qName.equals(AUTHORS)) {
 163                                 authorship = Team.NewInstance();
 164                         } else if (reference != null && qName.equals(SUBJECTS)) {
 165                                 //TODO implement, but no equivalent in the cdm model
 166                         } else {
 167                                 elementName = qName;
 168                         }
 169                 }
 170
 171                 @Override
 172                 public void endElement(String uri, String localName, String qName) throws SAXException {
 173
 174                         if (qName.equals(OPENURL_RESPONSE)) {
 175
 176                         } else if (qName.equals(OPENURL_RESPONSE_CITATION)) {
 177                                 referenceList.add(reference);
 178                                 reference = null;
 179                         } else if (reference != null && qName.equals(AUTHORS)) {
 180                                 reference.setAuthorship(authorship);
 181                                 authorship = null;
 182                         } else if (reference != null && qName.equals(SUBJECTS)) {
 183                                 //TODO implement, but no equivalent in the cdm model
 184                         }else {
 185                                 elementNameToStore = elementName;
 186                                 elementName = null;
 187                         }
 188
 189                 }
 190
 191                 @Override
 192                 public void characters(char ch[], int start, int length)
 193                                 throws SAXException {
 194
 195                         if(elementNameToStore  == null){
 196                                 textBuffer.append(new String(ch, start, length));
 197                         } else {
 198
 199                                 logger.debug("Characters [" + elementNameToStore + "]: " + textBuffer);
 200                                 String trimmedText = textBuffer.toString().trim();
 201                                 // empty the text buffer
 202                                 textBuffer.delete(0, textBuffer.length());
 203
 204                                 // --- Reference --- //
 205                                 if(reference != null){
 206
 207                                         if(elementNameToStore.equals(URL)){
 208                                                 try {
 209                                                         reference.setUri(new URI(trimmedText));
 210                                                 } catch (URISyntaxException e) {
 211                                                         logger.warn(e.getMessage());
 212                                                 }
 213                                         }
 214                                         if(elementNameToStore.equals(ITEM_URL)){
 215                                                 try {
 216                                                         reference.setItemUri(new URI(trimmedText));
 217                                                 } catch (URISyntaxException e) {
 218                                                         logger.warn(e.getMessage());
 219                                                 }
 220                                         }
 221                                         if(elementNameToStore.equals(TITLE_URL)){
 222                                                 try {
 223                                                         reference.setTitleUri(new URI(trimmedText));
 224                                                 } catch (URISyntaxException e) {
 225                                                         logger.warn(e.getMessage());
 226                                                 }
 227                                         }
 228                                         if(elementNameToStore.equals(TITLE)){
 229                                                 reference.setTitleCache(trimmedText, true);
 230                                         }
 231                                         if(elementNameToStore.equals(STITLE)){
 232                                                 logger.debug(elementNameToStore + " not yet implemented!");//TODO
 233                                         }
 234                                         if(elementNameToStore.equals(ATITLE)){
 235                                                 logger.debug(elementNameToStore + " not yet implemented!");//TODO
 236                                         }
 237                                         if(elementNameToStore.equals(PUBLISHER_NAME)){
 238                                                 reference.setPublisher(trimmedText);
 239                                         }
 240                                         if(elementNameToStore.equals(PUBLISHER_PLACE)){
 241                                                 reference.setPlacePublished(trimmedText);
 242                                         }
 243                                         if(elementNameToStore.equals(DATE)){
 244                                                 /* may be a single year or a range of years 1797-1830 */
 245                                                 Integer startYear = null;
 246                                                 Integer endYear = null;
 247                                                 if(trimmedText.length() == 9 && trimmedText.indexOf("-") == 4){
 248                                                         try {
 249                                                                 startYear = Integer.valueOf(trimmedText.substring(0, 4));
 250                                                                 endYear = Integer.valueOf(trimmedText.substring(5));
 251                                                                 reference.setDatePublished(VerbatimTimePeriod.NewVerbatimInstance(startYear, endYear));
 252                                                         } catch (NumberFormatException e) {
 253                                                                 logger.error("date can not be parsed: "+ trimmedText);
 254                                                         }
 255                                                 } else if(trimmedText.length() == 4) {
 256                                                         try {
 257                                                                 startYear = Integer.valueOf(trimmedText);
 258                                                         } catch (NumberFormatException e) {
 259                                                                 logger.error("date can not be parsed: "+ trimmedText);
 260                                                         }
 261                                                         reference.setDatePublished(VerbatimTimePeriod.NewVerbatimInstance(startYear));
 262                                                 }
 263                                         }
 264                                         if(elementNameToStore.equals(VOLUME)){
 265                                                 reference.setVolume(trimmedText);
 266                                         }
 267                                         if(elementNameToStore.equals(EDITION)){
 268                                                 reference.setEdition(trimmedText);
 269                                         }
 270                                         if(elementNameToStore.equals(SPAGE)){
 271                                                 reference.setPages(trimmedText);
 272                                         }
 273                                         if(elementNameToStore.equals(EPAGE)){
 274                                                 logger.debug(elementNameToStore + " not yet implemented!");//TODO
 275                                         }
 276                                         if(elementNameToStore.equals(PAGES)){
 277                                                 // IGNORE we rather need the start page value SPAGE
 278                                         }
 279                                         if(elementNameToStore.equals(PUBLICATION_FREQUENCY)){
 280                                                 logger.debug(elementNameToStore + " not yet implemented!");//TODO
 281                                         }
 282                                         if(elementNameToStore.equals(LANGUAGE)){
 283                                                 logger.debug(elementNameToStore + " not yet implemented!");//TODO
 284                                         }
 285                                         if(elementNameToStore.equals(OCLC)){
 286                                                 logger.debug(elementNameToStore + " not yet implemented!");//TODO
 287                                         }
 288                                         if(elementNameToStore.equals(LCCN)){
 289                                                 logger.debug(elementNameToStore + " not yet implemented!");//TODO
 290                                         }
 291                                         if(elementNameToStore.equals(ISSN)){
 292                                                 reference.setIssn(trimmedText);
 293                                         }
 294                                 }
 295
 296                                 // --- Reference.authorship --- //
 297                                 if(authorship != null && reference != null){
 298                                         if(elementNameToStore.equals("String")){
 299                                                 authorship.addTeamMember(Person.NewTitledInstance(trimmedText));
 300                                         }
 301                                 }
 302
 303                                 // openUrlResponse //
 304                                 if(reference == null){
 305                                         if(elementNameToStore.equals(STATUS)){
 306                                                 status = ResponseStatus.valueOf(trimmedText);
 307                                         }
 308                                 }
 309
 310                                 elementNameToStore = null;
 311                         }
 312                 }
 313
 314         }
 315
 316          /**
 317          * @see http://code.google.com/p/bhl-bits/source/browse/trunk/portal/OpenUrlUtilities/IOpenUrlResponse.cs
 318          */
 319         public enum ResponseStatus {
 320                 Undefined, // Query not submitted
 321                 Success, Error
 322         }
 323
 324
 325 }