[15222] | 1 | /**
|
---|
| 2 | *#########################################################################
|
---|
| 3 | * FedoraGS3Connection.java - works with the demo-client for Greenstone 3,
|
---|
| 4 | * of the Greenstone digital library suite from the New Zealand Digital
|
---|
| 5 | * Library Project at the * University of Waikato, New Zealand.
|
---|
| 6 | * <BR><BR>
|
---|
| 7 | * Copyright (C) 2008 New Zealand Digital Library Project
|
---|
| 8 | * <BR><BR>
|
---|
| 9 | * This program is free software; you can redistribute it and/or modify
|
---|
| 10 | * it under the terms of the GNU General Public License as published by
|
---|
| 11 | * the Free Software Foundation; either version 2 of the License, or
|
---|
| 12 | * (at your option) any later version.
|
---|
| 13 | * <BR><BR>
|
---|
| 14 | * This program is distributed in the hope that it will be useful,
|
---|
| 15 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 16 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 17 | * GNU General Public License for more details.
|
---|
| 18 | *########################################################################
|
---|
| 19 | */
|
---|
| 20 |
|
---|
| 21 | package org.greenstone.fedora.services;
|
---|
| 22 |
|
---|
| 23 |
|
---|
| 24 | import java.io.StringReader;
|
---|
| 25 |
|
---|
| 26 | import org.apache.log4j.Logger;
|
---|
| 27 | import org.greenstone.fedora.services.FedoraGS3Exception.AuthenticationFailedException;
|
---|
| 28 | import org.greenstone.fedora.services.FedoraGS3Exception.CancelledException;
|
---|
| 29 | import org.greenstone.fedora.services.FedoraGS3Exception.FedoraGS3RunException;
|
---|
| 30 | import org.greenstone.fedora.services.FedoraGS3Exception.NotAFedoraServerException;
|
---|
| 31 | import org.greenstone.gsdl3.util.GSXML;
|
---|
| 32 | import org.w3c.dom.Document;
|
---|
| 33 | import org.w3c.dom.Element;
|
---|
| 34 | import org.w3c.dom.Attr;
|
---|
| 35 | import org.w3c.dom.Text;
|
---|
| 36 | import org.w3c.dom.NodeList;
|
---|
| 37 | import org.w3c.dom.Node;
|
---|
| 38 | import org.xml.sax.InputSource;
|
---|
| 39 |
|
---|
| 40 | import java.io.File;
|
---|
| 41 | import java.util.HashMap;
|
---|
| 42 | import java.util.Properties;
|
---|
| 43 | import java.util.Map;
|
---|
| 44 |
|
---|
| 45 | import javax.swing.JOptionPane;
|
---|
| 46 |
|
---|
| 47 | import org.xml.sax.SAXException;
|
---|
| 48 | import java.io.UnsupportedEncodingException;
|
---|
| 49 | import java.io.IOException;
|
---|
| 50 | import javax.net.ssl.SSLHandshakeException;
|
---|
| 51 | import java.net.ConnectException;
|
---|
| 52 | import java.net.MalformedURLException;
|
---|
| 53 | import java.rmi.RemoteException;
|
---|
| 54 | import javax.xml.parsers.ParserConfigurationException;
|
---|
| 55 | import javax.xml.transform.TransformerException;
|
---|
| 56 |
|
---|
| 57 | /**
|
---|
| 58 | * Class that extends FedoraConnection in order to be able to use
|
---|
| 59 | * Fedora's web services to retrieve the specific datastreams of
|
---|
| 60 | * Greenstone documents stored in Fedora's repository. This class
|
---|
| 61 | * provides methods that convert those datastreams into Greenstone3
|
---|
| 62 | * XML response messages which are returned.
|
---|
| 63 | * @author ak19
|
---|
| 64 | */
|
---|
| 65 | public class FedoraGS3Connection
|
---|
| 66 | extends FedoraConnection implements FedoraToGS3Interface,
|
---|
| 67 | FedoraToGS3Interface.Constants
|
---|
| 68 | {
|
---|
| 69 | /** The logging instance for this class */
|
---|
| 70 | private static final Logger LOG = Logger.getLogger(
|
---|
| 71 | FedoraGS3Connection.class.getName());
|
---|
[21859] | 72 |
|
---|
| 73 | /** Default name of Fedora index */
|
---|
| 74 | private static final String DEFAULT_FEDORA_INDEX = "BasicIndex";
|
---|
| 75 |
|
---|
[15222] | 76 | /** Complete list of services that are supported our FedoraGS3 would
|
---|
| 77 | * support if everything goes well. If a connection to FedoraGSearch
|
---|
| 78 | * cannot be established, the query services will no longer be
|
---|
| 79 | * available. The actual services supported are given by member
|
---|
| 80 | * variable serviceNames. */
|
---|
| 81 | protected static final String[] SERVICES = {
|
---|
| 82 | "DocumentContentRetrieve", "DocumentMetadataRetrieve",
|
---|
| 83 | "DocumentStructureRetrieve",
|
---|
| 84 | "TextQuery", "FieldQuery",
|
---|
| 85 | "ClassifierBrowse", "ClassifierBrowseMetadataRetrieve"
|
---|
| 86 | };
|
---|
| 87 |
|
---|
| 88 | /** List of services actually supported by our FedoraGS3 repository
|
---|
| 89 | * after construction. If FedoraGenericSearch can't be connected to,
|
---|
| 90 | * then query services will not be offered */
|
---|
| 91 | protected String[] serviceNames;
|
---|
| 92 |
|
---|
| 93 | /** constant CHILDREN indicates that a DocumentStructureRetrieve is to
|
---|
| 94 | * return only the child nodes of a section, not any further descendants */
|
---|
| 95 | protected static final int CHILDREN = 0;
|
---|
| 96 |
|
---|
| 97 | /** constant DESCENDANTS indicates that a DocumentStructureRetrieve is to
|
---|
| 98 | * return all descendants of a section */
|
---|
| 99 | protected static final int DESCENDANTS = 1;
|
---|
| 100 |
|
---|
| 101 | /** The object used to connect to FedoraGenericSearch, which is used
|
---|
| 102 | * for doing full-text searching */
|
---|
| 103 | protected GSearchConnection fedoraGSearch;
|
---|
| 104 |
|
---|
| 105 | /** The url for the wsdl file of FedoraGSearch's web services
|
---|
| 106 | * by default this will be the Fedora server's base URL
|
---|
| 107 | * concatenated to "gsearch/services/FgsOperations?wsdl" */
|
---|
| 108 | protected String gSearchWSDLURL;
|
---|
[15733] | 109 |
|
---|
| 110 | /** The last part of the gSearchWSDL URL. The first part is
|
---|
| 111 | * the same as the fedora server's base url. */
|
---|
| 112 | protected String gSearchWSDLSuffix;
|
---|
| 113 |
|
---|
[15437] | 114 | /** The name of the index that FedoraGSearch will index the GS3
|
---|
| 115 | * documents into. If no name is specified in the properties file,
|
---|
| 116 | * this will default to FedoraIndex. */
|
---|
| 117 | protected String gSearchIndexName;
|
---|
| 118 |
|
---|
[15222] | 119 | /** 5 argument constructor is the same as that of superclass FedoraConnection:
|
---|
| 120 | * @param protocol can be either http or https
|
---|
| 121 | * @param host is the host where the fedora server is listening
|
---|
| 122 | * @param port is the port where the fedora server is listening
|
---|
| 123 | * @param fedoraServerUsername is the username for administrative
|
---|
| 124 | * authentication required to access the fedora server.
|
---|
| 125 | * @param fedoraServerPassword is the password for administrative
|
---|
| 126 | * authentication required to access the fedora server. If no password was set
|
---|
| 127 | * when installing Fedora, leave the field "".
|
---|
| 128 | * Instantiates a FedoraGS3Connection object which connects to Fedora's
|
---|
| 129 | * web services through stub classes and tries to connect to FedoraGSearch's
|
---|
| 130 | * web services through the default WSDL location for it
|
---|
| 131 | * ("gsearch/services/FgsOperations?wsdl"). If another url is to be used,
|
---|
| 132 | * call setGSearchWSDLURL(url) after the constructor instead.
|
---|
| 133 | */
|
---|
| 134 | public FedoraGS3Connection(String protocol, String host, int port,
|
---|
| 135 | String fedoraServerUsername, String fedoraServerPassword)
|
---|
| 136 | throws ParserConfigurationException, MalformedURLException,
|
---|
| 137 | SSLHandshakeException, RemoteException, AuthenticationFailedException,
|
---|
| 138 | NotAFedoraServerException, ConnectException, Exception
|
---|
| 139 | {
|
---|
| 140 | super(protocol, host, port, fedoraServerUsername, fedoraServerPassword);
|
---|
| 141 | // super() will call setInitialisationProperties(properties)
|
---|
| 142 | // And that will try to instantiate the GSearchConnection.
|
---|
| 143 | }
|
---|
| 144 |
|
---|
| 145 | /** No-argument constructor which is the same as that of superclass
|
---|
| 146 | * FedoraConnection: it displays a small dialog requesting input for the
|
---|
| 147 | * host, port, administrative password and username of the fedora server.
|
---|
| 148 | * If no password was set on the fedora repository when installing it,
|
---|
| 149 | * the user can leave the password field blank. */
|
---|
| 150 | public FedoraGS3Connection()
|
---|
| 151 | throws ParserConfigurationException, MalformedURLException,
|
---|
| 152 | CancelledException, ConnectException, RemoteException,
|
---|
| 153 | SSLHandshakeException, Exception
|
---|
| 154 | {
|
---|
| 155 | super();
|
---|
| 156 | // super() will call setInitialisationProperties(properties)
|
---|
| 157 | // And that will try to instantiate the GSearchConnection.
|
---|
| 158 | }
|
---|
| 159 |
|
---|
| 160 | /** Single-argument constructor which is the same as that of superclass
|
---|
| 161 | * FedoraConnection: it takes the name of the properties file where
|
---|
| 162 | * connection initialisation values may already be provided and then
|
---|
| 163 | * displays a small dialog requesting input for the host, port,
|
---|
| 164 | * administrative password and username of the fedora server showing
|
---|
| 165 | * the values in the properties file as default. If the necessary
|
---|
[21835] | 166 | * initialisation are not present in the file, the corresponding fields
|
---|
[15222] | 167 | * in the dialog will be blank.
|
---|
| 168 | * If no password was set on the fedora repository when installing it,
|
---|
| 169 | * the user can leave the password field blank. */
|
---|
| 170 | public FedoraGS3Connection(File propertiesFilename)
|
---|
| 171 | throws ParserConfigurationException, MalformedURLException,
|
---|
| 172 | CancelledException, ConnectException, RemoteException,
|
---|
| 173 | SSLHandshakeException, Exception
|
---|
| 174 | {
|
---|
| 175 | super(propertiesFilename);
|
---|
| 176 | // super() will call setInitialisationProperties(properties)
|
---|
| 177 | // And that will try to instantiate the GSearchConnection.
|
---|
| 178 | }
|
---|
| 179 |
|
---|
| 180 | /** The superclass constructor calls this method passing any preset
|
---|
| 181 | * properties loaded from a propertiesFile. This method is overridden
|
---|
| 182 | * here in order to instantiate the gSearchConnection based on the
|
---|
[15733] | 183 | * - gSearchWSDLSuffix that will be appended to the fedora base url.
|
---|
| 184 | * (If one was not provided in the properties file, gSearchWSDLURL defaults
|
---|
| 185 | * to something of the form
|
---|
| 186 | * "http://<fedorahost:port>/fedoragsearch/services/FgsOperations?wsdl"
|
---|
| 187 | * which is the baseURL of fedora concatenated to the default gSearchWSDLSuffix,
|
---|
| 188 | * "gsearch/services/FgsOperations?wsdl".
|
---|
[15437] | 189 | * - name of the index into which the GS3 documents have been indexed
|
---|
| 190 | * and which FedoraGenericSearch should use to perform searches. If none is
|
---|
[21573] | 191 | * given in the properties file, then the index name defaults to "FedoraIndex".
|
---|
[15222] | 192 | * @param properties is the Properties Map loaded from a properties file
|
---|
| 193 | * (if there was any) which specifies such things as host and port of the
|
---|
[15733] | 194 | * FedoraServer, but can also specify the property "gsearch.wsdlURL.suffix".
|
---|
| 195 | * At the end of this method, properties' "gsearch.wsdlURL.suffix" will be set
|
---|
| 196 | * to whatever the final value of this.gSearchWSDLURL' suffix is, and
|
---|
[15437] | 197 | * "gsearch.indexName" will be set to to whatever the final value of
|
---|
| 198 | * this.gSearchIndexName is.
|
---|
[15222] | 199 | */
|
---|
| 200 | protected void setInitialisationProperties(Properties properties)
|
---|
| 201 | throws ParserConfigurationException, MalformedURLException,
|
---|
| 202 | CancelledException, ConnectException, RemoteException,
|
---|
| 203 | SSLHandshakeException, Exception
|
---|
| 204 | {
|
---|
| 205 | super.setInitialisationProperties(properties);
|
---|
[15733] | 206 | // gsearchWSDL URL suffix, if not specified, defaults to
|
---|
| 207 | // "fedoragsearch/services/FgsOperations?wsdl" which is
|
---|
| 208 | // concatenated to the baseURL of fedora to give the gsearchWSDLURL.
|
---|
| 209 | this.gSearchWSDLSuffix = properties.getProperty(
|
---|
| 210 | "gsearch.wsdlURL.suffix", "gsearch/services/FgsOperations?wsdl");
|
---|
| 211 | this.gSearchWSDLURL = this.baseURL + this.gSearchWSDLSuffix;
|
---|
[15222] | 212 | // Set the property to whatever this.gSearchWSDLURL is now,
|
---|
| 213 | // so that it will be written out to the properties file again
|
---|
[15733] | 214 | properties.setProperty("gsearch.wsdlURL.suffix", this.gSearchWSDLSuffix);
|
---|
[15437] | 215 |
|
---|
| 216 | // Similarly for the name of the index FedoraGenericSearch should use
|
---|
| 217 | // when performing searches for GS3 docs stored in Fedora's repository.
|
---|
| 218 | this.gSearchIndexName = properties.getProperty(
|
---|
[21859] | 219 | "gsearch.indexName", DEFAULT_FEDORA_INDEX); // default to FedoraIndex
|
---|
[15437] | 220 | properties.setProperty("gsearch.indexName", this.gSearchIndexName);
|
---|
[15222] | 221 | // Create a connection to FedoraGSearch's web services:
|
---|
| 222 | initSearchFunctionality();
|
---|
| 223 | }
|
---|
[21859] | 224 |
|
---|
| 225 | /** Overridden init method to work with the 5 argument constructor, so that we can
|
---|
| 226 | * bypass using setInitialisationProperties() which works with a Properties map.
|
---|
| 227 | */
|
---|
| 228 | protected void init(String protocol, String host, String port,
|
---|
| 229 | String fedoraServerUsername, String fedoraServerPassword)
|
---|
| 230 | throws ParserConfigurationException, MalformedURLException,
|
---|
| 231 | AuthenticationFailedException, RemoteException, Exception
|
---|
| 232 | {
|
---|
| 233 | super.init(protocol, host, port, fedoraServerUsername, fedoraServerPassword);
|
---|
| 234 | this.gSearchWSDLSuffix = "gsearch/services/FgsOperations?wsdl";
|
---|
| 235 | this.gSearchWSDLURL = this.baseURL + this.gSearchWSDLSuffix;
|
---|
| 236 | this.gSearchIndexName = DEFAULT_FEDORA_INDEX;
|
---|
| 237 | initSearchFunctionality();
|
---|
| 238 | }
|
---|
[15222] | 239 |
|
---|
[21859] | 240 |
|
---|
[15222] | 241 | /** Init method that instantiates a GSearchConnection object used
|
---|
| 242 | * to work with the separate FedoraGSearch web services.
|
---|
| 243 | * The url of the WSDL for FedoraGSearch's web services is worked out
|
---|
| 244 | * from the baseURL of the Fedora server.
|
---|
| 245 | */
|
---|
| 246 | protected void initSearchFunctionality()
|
---|
| 247 | {
|
---|
| 248 | try {
|
---|
[15437] | 249 | this.fedoraGSearch = null;
|
---|
| 250 | this.fedoraGSearch = new GSearchConnection(
|
---|
| 251 | gSearchWSDLURL, gSearchIndexName);
|
---|
[15222] | 252 | this.serviceNames = SERVICES;
|
---|
| 253 | } catch(Exception e){
|
---|
| 254 | LOG.error("Cannot connect to FedoraGSearch's web services at "
|
---|
| 255 | + gSearchWSDLURL + "\nQuery services will not be available.");
|
---|
| 256 | // If an exception occurs, something has gone wrong when
|
---|
| 257 | // trying to connect to FedoraGSearch's web services. This
|
---|
| 258 | // means, we can't offer query services, as that's provided
|
---|
| 259 | // by FedoraGSearch
|
---|
| 260 | serviceNames = null;
|
---|
| 261 | int countOfNonQueryServices = 0;
|
---|
| 262 | for(int i = 0; i < SERVICES.length; i++) {
|
---|
| 263 | // do not count query services
|
---|
| 264 | if(!SERVICES[i].toLowerCase().contains("query")) {
|
---|
| 265 | countOfNonQueryServices++;
|
---|
| 266 | }
|
---|
| 267 | }
|
---|
| 268 | // Services now supported are everything except Query services
|
---|
| 269 | serviceNames = new String[countOfNonQueryServices];
|
---|
[15331] | 270 | int j = 0;
|
---|
[15222] | 271 | for(int i = 0; i < SERVICES.length; i++) {
|
---|
| 272 | if(!SERVICES[i].toLowerCase().contains("query")) {
|
---|
[15331] | 273 | serviceNames[j] = SERVICES[i];
|
---|
| 274 | j++; // valid serviceName, so increment serviceName counter
|
---|
[15222] | 275 | }
|
---|
| 276 |
|
---|
| 277 | }
|
---|
| 278 | }
|
---|
| 279 | }
|
---|
| 280 |
|
---|
| 281 | /** @return the gSearchWSDLURL, the url of the WSDL for the
|
---|
| 282 | * FedoraGSearch web services */
|
---|
| 283 | public String getGSearchWSDLURL() { return gSearchWSDLURL; }
|
---|
| 284 |
|
---|
| 285 | /** Sets the member variable gSearchWSDLURL that specify the location of
|
---|
| 286 | * the WSDL file of FedoraGSearch's web services. Then it attempts
|
---|
| 287 | * to instantiate a connection to those web services.
|
---|
| 288 | * @param url is the new url of the GSearch web services WSDL file */
|
---|
| 289 | public void setGSearchWSDLURL(String url) {
|
---|
[15437] | 290 | this.gSearchWSDLURL = url;
|
---|
[15222] | 291 | initSearchFunctionality();
|
---|
| 292 | }
|
---|
| 293 |
|
---|
[15437] | 294 | /** @return the gSearchIndexName, the name of the index Fedora Generic
|
---|
| 295 | * Search will search in (where GS3 docs have been indexed into). */
|
---|
| 296 | public String getGSearchIndexName() { return gSearchIndexName; }
|
---|
| 297 |
|
---|
| 298 | /** Sets the member variable gSearchIndexName that specifies the name
|
---|
| 299 | * of the index containing indexed GS3 documents. Then it attempts
|
---|
| 300 | * to instantiate a connection to the Fedora GSearch web services using
|
---|
| 301 | * this changed value for indexName.
|
---|
| 302 | * @param indexName is the new name of the index containing indexed GS3
|
---|
| 303 | * docs that GSearch should search in. */
|
---|
| 304 | public void setGSearchIndexName(String indexName) {
|
---|
| 305 | this.gSearchIndexName = indexName;
|
---|
| 306 | initSearchFunctionality();
|
---|
| 307 | }
|
---|
| 308 |
|
---|
[15222] | 309 | /** @return the array of the services actually supported by FedoraGS3 */
|
---|
| 310 | protected String[] getServiceNames() { return this.serviceNames;}
|
---|
| 311 |
|
---|
| 312 | /**
|
---|
| 313 | * For finding out if the sectionNumber is given as part of the docID.
|
---|
| 314 | * @param docID is the String that contains the docPID and may also
|
---|
| 315 | * contain the section number.
|
---|
| 316 | * @return true if the document identifier docID contains a section-
|
---|
| 317 | * number, and false if it consists solely of the docPID.
|
---|
| 318 | * That is, true is returned if
|
---|
| 319 | * <pre>docID = "greenstone:colName-<docPID>-<sectionNum>"</pre>
|
---|
| 320 | * and false is returned if
|
---|
| 321 | * <pre>docID = "greenstone:colName-<docPID>"</pre>
|
---|
| 322 | * */
|
---|
| 323 | protected boolean containsSectionNumber(String docID) {
|
---|
| 324 | // if there are two hyphens in the docID, then there are sections
|
---|
| 325 | // (and the section number is appended at end of docID)
|
---|
| 326 | // docID = "greenstone:colName-<docPID>-<sectionNum>"
|
---|
| 327 | return (docID.lastIndexOf(HYPHEN) != docID.indexOf(HYPHEN));
|
---|
| 328 | }
|
---|
| 329 |
|
---|
| 330 | /** This method will extract the docPID from docID and return it.
|
---|
| 331 | * (If a sectionNumber is suffixed to the docID, the docPID which is
|
---|
| 332 | * the prefix is returned; otherwise the docID is the docPID and is
|
---|
| 333 | * returned)
|
---|
| 334 | * @param docID is the String that contains the docPID and may also
|
---|
| 335 | * contain the section number.
|
---|
| 336 | * @return only the docPID portion of the docID.
|
---|
| 337 | */
|
---|
| 338 | protected String getDocPIDFromDocID(String docID) {
|
---|
| 339 | if(containsSectionNumber(docID))
|
---|
| 340 | return docID.substring(0, docID.lastIndexOf(HYPHEN));
|
---|
| 341 | // else (if there's no sectionNumber), docID is the docPID
|
---|
| 342 | return docID;
|
---|
| 343 | }
|
---|
| 344 |
|
---|
| 345 | /** This method will return the section Number, if there's any
|
---|
| 346 | * suffixed to the docID. Otherwise it will return the empty string
|
---|
| 347 | * @param docID is the String that contains the docPID and may also
|
---|
| 348 | * contain the section number.
|
---|
| 349 | * @return only the sectionID portion of the docID - if any, else "".
|
---|
| 350 | */
|
---|
| 351 | protected String getSectionIDFromDocID(String docID) {
|
---|
| 352 | if(containsSectionNumber(docID))
|
---|
| 353 | return docID.substring(
|
---|
| 354 | docID.lastIndexOf(HYPHEN)+1, docID.length());
|
---|
| 355 | return "";
|
---|
| 356 | }
|
---|
| 357 |
|
---|
| 358 | /** Given a list of collectionIDs, returns a GS3 DocumentMetadataRetrieve
|
---|
| 359 | * response message that gives the metadata for each collection identified
|
---|
| 360 | * @param collIDs is an array of fedora pids identifying collections in the
|
---|
| 361 | * fedora repository
|
---|
| 362 | * @return a GS3 DocumentMetadataRetrieve response message containing the
|
---|
| 363 | * EX metadata for all the requested collections */
|
---|
| 364 | public String getCollectionMetadata(String[] collIDs) {
|
---|
| 365 | return getMetadata(collIDs);
|
---|
| 366 | }
|
---|
| 367 |
|
---|
| 368 | /** Given a list of document identifiers, a GS3 DocumentMetadataRetrieve
|
---|
| 369 | * response message is returned containing the metadata for each document.
|
---|
| 370 | * @param docIDs is an array of document identifiers (docID can either be
|
---|
| 371 | * <pid>s items (documents) in the fedora repository, or
|
---|
| 372 | * "<pid>-sectionNumber".
|
---|
| 373 | * @return a GS3 DocumentMetadataRetrieve response message containing the
|
---|
| 374 | * EX, DC, DLS metadata for all the requested documents */
|
---|
| 375 | public String getDocumentMetadata(String[] docIDs) {
|
---|
| 376 | return getMetadata(docIDs);
|
---|
| 377 | }
|
---|
| 378 |
|
---|
| 379 | /** Given a collectionID, returns a GS3 DocumentMetadataRetrieve
|
---|
| 380 | * response message that gives the metadata for the collection identified
|
---|
| 381 | * @param collID is a fedora pid identifying a collection in its repository
|
---|
| 382 | * @return a GS3 DocumentMetadataRetrieve response message containing the
|
---|
| 383 | * EX metadata for the requested collection */
|
---|
| 384 | public String getCollectionMetadata(String collID) {
|
---|
| 385 | return getMetadata(new String[] {collID});
|
---|
| 386 | }
|
---|
| 387 |
|
---|
| 388 | /** Given a document identifier, returns a GS3 DocumentMetadataRetrieve
|
---|
| 389 | * response message containing the metadata for the document.
|
---|
| 390 | * @param docID is a document identifier (docID can either be a <pid>
|
---|
| 391 | * of an item (document) in the fedora repository, or it can be
|
---|
| 392 | * "<pid>-sectionNumber".
|
---|
| 393 | * @return a GS3 DocumentMetadataRetrieve response message containing the
|
---|
| 394 | * EX, DC, DLS metadata for the requested document */
|
---|
| 395 | public String getDocumentMetadata(String docID) {
|
---|
| 396 | return getMetadata(new String[] {docID});
|
---|
| 397 | }
|
---|
| 398 |
|
---|
| 399 | /** @return a greenstone DocumentMetadataRetrieve response for the
|
---|
| 400 | * documents or collections indicated by the docIDsOrCollIDs.
|
---|
| 401 | * @param docIDsOrCollIDs is an array of identifiers which may be either the
|
---|
| 402 | * fedora pids for collections, or otherwise may be a document identifier.
|
---|
| 403 | * In the last case, the document ID may consist of either
|
---|
| 404 | * "documentPID-sectionNumber" or may just be just fedora documentPID */
|
---|
| 405 | public String getMetadata(String[] docIDsOrCollIDs)
|
---|
| 406 | {
|
---|
| 407 | Document doc = builder.newDocument();
|
---|
| 408 | FedoraGS3RunException ex = null;
|
---|
| 409 |
|
---|
| 410 | Element docNodeList = doc.createElement(
|
---|
| 411 | GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 412 |
|
---|
| 413 | try{
|
---|
| 414 | for(int i = 0; i < docIDsOrCollIDs.length; i++) {
|
---|
| 415 | // create the <documentNode> containing the metadata
|
---|
| 416 | // for each document docID
|
---|
| 417 | Element docNode = getMetadata(doc, docIDsOrCollIDs[i]);
|
---|
| 418 | docNodeList.appendChild(docNode);
|
---|
| 419 | }
|
---|
| 420 | } catch(Exception e) {
|
---|
| 421 | ex = new FedoraGS3RunException(e);
|
---|
| 422 | ex.setSpecifics("EX (and/or DC, DLS) metadata datastream");
|
---|
| 423 | }
|
---|
| 424 |
|
---|
| 425 | Element responseMsg = createResponseMessage(doc, docNodeList, ex,
|
---|
| 426 | GSXML.REQUEST_TYPE_PROCESS, "DocumentMetadataRetrieve");
|
---|
| 427 | try{
|
---|
| 428 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 429 | } catch(TransformerException e) {
|
---|
| 430 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 431 | + " " + e;
|
---|
| 432 | }
|
---|
| 433 | }
|
---|
| 434 |
|
---|
| 435 | /** Method that takes a new DOM document, as well as an identifier of either
|
---|
| 436 | * a collection or document (which may be a fedora pid for the collection
|
---|
| 437 | * or document, or may be the documentPid-sectionNumber for a document) and
|
---|
| 438 | * returns a documentNode element for it:
|
---|
| 439 | * <documentNode><metadataList>
|
---|
| 440 | * <metadata name="">value</metadata>
|
---|
| 441 | * ...
|
---|
| 442 | * </metadataList></documentNode>
|
---|
| 443 | * @return documentNode containing the metadata for the collection or
|
---|
| 444 | * document given by parameter ID
|
---|
| 445 | * @param id denotes a collection pid, a document pid or a docID of the
|
---|
| 446 | * form "documentpid-sectionNumber" */
|
---|
| 447 | protected Element getMetadata(Document doc, String id)
|
---|
| 448 | throws RemoteException, UnsupportedEncodingException,
|
---|
| 449 | SAXException, IOException
|
---|
| 450 | {
|
---|
| 451 | // We're going to create the documentNode nested inside the following
|
---|
| 452 | // documentNodeList:
|
---|
| 453 | // <documentNodeList>
|
---|
| 454 | // <documentNode nodeID=""><metadataList>
|
---|
| 455 | // <metadata name="">value</metadata>
|
---|
| 456 | // </metadataList></documentNode>
|
---|
| 457 | // <documentNode>...</documentNode>
|
---|
| 458 | // </documentNodeList>
|
---|
| 459 | // <documentNodeList>
|
---|
| 460 |
|
---|
| 461 | // <documentNode nodeID="docID"> - the docNode on which a structure
|
---|
| 462 | // retrieve is being performed
|
---|
| 463 | Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
| 464 | Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
| 465 | attribute.setValue(id);
|
---|
| 466 | docNode.setAttributeNode(attribute);
|
---|
| 467 |
|
---|
| 468 | // <metadataList>
|
---|
| 469 | Element metadataList = doc.createElement(
|
---|
| 470 | GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 471 |
|
---|
| 472 | String ex = "";
|
---|
| 473 | String dc = "";
|
---|
| 474 | String dls = "";
|
---|
| 475 | if(id.endsWith(_COLLECTION)) { // docID refers to a collection
|
---|
| 476 | // Obtain the "EX" datastream (extracted metadata) for the collection
|
---|
| 477 | ex = this.getEX(id);
|
---|
| 478 | }
|
---|
| 479 | else { // docID refers to a document
|
---|
| 480 | // work out the document's fedora PID and section ID, and then
|
---|
| 481 | // obtain the EX (extracted metadata) and DC datastreams for the doc
|
---|
| 482 |
|
---|
| 483 | // Note that EX/DC for pid="greenstone:<colname>-docPID-1"
|
---|
| 484 | // is the same as for pid="greenstone:<colname>-docPID"
|
---|
| 485 | // That is, <Section id="1"> refers to the toplevel document docPID
|
---|
| 486 | // If requested for top-level document, there may also be DLS meta
|
---|
| 487 | String sectionID = getSectionIDFromDocID(id);
|
---|
| 488 | String docPID = getDocPIDFromDocID(id);
|
---|
| 489 | if(sectionID.equals("") || sectionID.equals("1")) {
|
---|
[21573] | 490 | // metadata of toplevel document is requested
|
---|
[15222] | 491 | ex = this.getEX(docPID); // slightly faster than doing
|
---|
| 492 | //getSectionEXMetadata(docID, "1")
|
---|
| 493 | dc = this.getDC(docPID);
|
---|
| 494 | dls = this.getDLS(docPID);
|
---|
| 495 | }
|
---|
| 496 | else {
|
---|
| 497 | ex = getSectionEXMetadata(docPID, sectionID);
|
---|
| 498 | dc = getSectionDCMetadata(docPID, sectionID);
|
---|
| 499 | }
|
---|
| 500 | }
|
---|
| 501 |
|
---|
| 502 | // Adding in metadata sets in alphabetical order
|
---|
| 503 | // DC metadata for a top-level document is different from EX, DLS:
|
---|
| 504 | // only the element's namespace prefix is "dc", the rest of a tagname
|
---|
| 505 | // is unknown.
|
---|
| 506 | if(!dc.equals("")) {
|
---|
| 507 | addMetadataWithNamespacedTagNames(doc, metadataList,
|
---|
| 508 | dc, DC);
|
---|
| 509 | }
|
---|
| 510 |
|
---|
| 511 | // Check if we were supposed to process dls and dc metadata
|
---|
| 512 | // as well. We only ever do this for top-level documents,
|
---|
| 513 | // in which case, dls and dc will be non-empty strings
|
---|
| 514 | if(!dls.equals("")) {
|
---|
| 515 | addMetadataWithFixedTagName(doc, metadataList, dls, DLS);
|
---|
| 516 | }
|
---|
| 517 |
|
---|
| 518 | // we definitely have an EX metadatastream for each
|
---|
| 519 | // collection object, top-level document object,
|
---|
| 520 | // and document section item
|
---|
| 521 | addMetadataWithFixedTagName(doc, metadataList, ex, EX);
|
---|
| 522 |
|
---|
| 523 | // now the metadataList has been built up
|
---|
| 524 | docNode.appendChild(metadataList);
|
---|
| 525 |
|
---|
| 526 | return docNode; // return <documentNode> containing the metadata
|
---|
| 527 | }
|
---|
| 528 |
|
---|
| 529 | /** This method retrieves all the metadata elements in the metaDataStream
|
---|
| 530 | * parameter of the form <"metadataSetNS:metadata">"value"</metadata> where
|
---|
| 531 | * metadataSetNS is the namespace of each tag, and creates a new element of
|
---|
| 532 | * the form <metadata name="metadataSetNS:metadata">"value"</metadata> for
|
---|
| 533 | * each. Each of these are then appended to the metadataList parameter.
|
---|
| 534 | * @param doc is the Document object using which the new metadata Elements
|
---|
| 535 | * are to be constructed
|
---|
| 536 | * @param metadataList is the <metadataList> Element to which the new
|
---|
| 537 | * metadata Elements are to be appended as children.
|
---|
| 538 | * @param metaDatastream the metadata datastream in string form (e.g. the
|
---|
| 539 | * Dublin Core metadata stored in the Fedora repository).
|
---|
| 540 | * @param metadataSet is the constant datastream identifier, e.g. "DC".
|
---|
| 541 | * At present this method only applies to the DC metadata as that's the only
|
---|
| 542 | * one where each tagname is different except for the constant dc: namespace.
|
---|
| 543 | */
|
---|
| 544 | protected void addMetadataWithNamespacedTagNames(Document doc,
|
---|
| 545 | Element metadataList, String metaDatastream, String metadataSet)
|
---|
| 546 | throws SAXException, IOException
|
---|
| 547 | {
|
---|
| 548 | Document src = builder.parse(
|
---|
| 549 | new InputSource(new StringReader(metaDatastream)));
|
---|
| 550 |
|
---|
| 551 | // The following doesn't work for some reason: to retrieve all elements
|
---|
| 552 | // whose namespace prefix starts with "dc", we pass "*" for localName
|
---|
| 553 | //NodeList dcMetaTags = src.getElementsByTagNameNS(DC.toLowerCase(), "*");
|
---|
| 554 |
|
---|
| 555 | // Longer way: get the children of the root document
|
---|
| 556 | NodeList children = src.getDocumentElement().getChildNodes();
|
---|
| 557 |
|
---|
| 558 | for(int i = 0; i < children.getLength(); i++) {
|
---|
| 559 | String nodeName = children.item(i).getNodeName();
|
---|
| 560 | // check that the nodename starts with the "dc" namespace,
|
---|
| 561 | // which simultaneously ensures that the node's an element:
|
---|
| 562 | if(nodeName.startsWith(DC.toLowerCase())) {
|
---|
| 563 | // need to have a period for Greenstone instead of Fedora's colon
|
---|
| 564 | nodeName = nodeName.replace(COLON, PERIOD);
|
---|
| 565 | Element metatag = (Element)children.item(i);
|
---|
| 566 | String value = FedoraCommons.getValue(metatag);
|
---|
| 567 | // <dc:tagname>value</dc:tagname>
|
---|
| 568 | // we're going to put use this in our metadata element as
|
---|
| 569 | // <metadata name="dc:tagname">value</metadata>
|
---|
| 570 |
|
---|
| 571 | // create metadata of (name, value) pairs in target DOM (doc)
|
---|
| 572 | Element metadata = doc.createElement(GSXML.METADATA_ELEM);
|
---|
| 573 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 574 | attribute.setValue(nodeName);
|
---|
| 575 | metadata.setAttributeNode(attribute);
|
---|
| 576 | Text content = doc.createTextNode(value);
|
---|
| 577 | metadata.appendChild(content);
|
---|
| 578 | metadataList.appendChild(metadata);
|
---|
| 579 | }
|
---|
| 580 | }
|
---|
| 581 | }
|
---|
| 582 |
|
---|
| 583 | /** This method retrieves all the metadata elements in the metaDataStream
|
---|
| 584 | * of the form <"namespace:"metadata name="metadataName">value</metadata>
|
---|
| 585 | * where "namespace" is the namespace prefix of each tag, and metadataName
|
---|
| 586 | * is the name of the metadata (like author, title). For each element
|
---|
| 587 | * it creates a corresponding new element of the form
|
---|
| 588 | * <metadata name="namespace:metadataName">value</metadata>. Each of these
|
---|
| 589 | * are then appended to the metadataList parameter.
|
---|
| 590 | * @param doc is the Document object using which the new metadata Elements
|
---|
| 591 | * are to be constructed
|
---|
| 592 | * @param metadataList is the <metadataList> Element to which the new
|
---|
| 593 | * metadata Elements are to be appended as children.
|
---|
| 594 | * @param metaDatastream the metadata datastream in string form (e.g. the
|
---|
| 595 | * EX/Greenstone extracted metadata or DLS metadata stored in the Fedora
|
---|
| 596 | * repository).
|
---|
| 597 | * @param metadataSet is the constant datastream identifier,
|
---|
| 598 | * e.g. "DLS" or "EX".
|
---|
| 599 | * At present this method applies to the DLS and EX metadata as they have
|
---|
| 600 | * constant tagnames throughout.
|
---|
| 601 | */
|
---|
| 602 | protected void addMetadataWithFixedTagName(Document doc,
|
---|
| 603 | Element metadataList, String metaDatastream, String metadataSet)
|
---|
| 604 | throws SAXException, IOException
|
---|
| 605 | {
|
---|
| 606 | // Namespace prefix can be "ex:" or "dls:"
|
---|
| 607 | String namespacePrefix = "";
|
---|
| 608 | if(!metadataSet.equals(EX)) {
|
---|
| 609 | // need to have a period for Greenstone instead of Fedora's colon
|
---|
| 610 | namespacePrefix = metadataSet.toLowerCase() + PERIOD;
|
---|
| 611 | }
|
---|
| 612 |
|
---|
| 613 | Document src = builder.parse(
|
---|
| 614 | new InputSource(new StringReader(metaDatastream)));
|
---|
| 615 | NodeList metaTags = src.getElementsByTagName(
|
---|
| 616 | metadataSet.toLowerCase()+COLON+METADATA);
|
---|
| 617 | // Looking for tagnames: <ex:metadata> or <dls:metadata>
|
---|
| 618 |
|
---|
| 619 | for(int i = 0; i < metaTags.getLength(); i++) {
|
---|
| 620 | Element metatag = (Element)metaTags.item(i);
|
---|
| 621 |
|
---|
| 622 | // extract the metadata of (name, value) pairs from src DOM
|
---|
| 623 | // look for <metadata name="name">value</metadata>
|
---|
| 624 | String name = metatag.hasAttribute(NAME) ?
|
---|
| 625 | metatag.getAttribute(NAME) : "";
|
---|
| 626 | // sometimes, there are several metadata for the same name, in this
|
---|
| 627 | // case, look for a qualifier and append its value to the name to
|
---|
| 628 | // distinguish it uniquely:
|
---|
| 629 | if(metatag.hasAttribute(QUALIFIER)) {
|
---|
| 630 | name = name + HYPHEN + metatag.getAttribute(QUALIFIER);
|
---|
| 631 | }
|
---|
| 632 | String value = FedoraCommons.getValue(metatag);
|
---|
| 633 |
|
---|
| 634 | // create metadata of (name, value) pairs in target DOM (doc)
|
---|
| 635 | Element metadata = doc.createElement(GSXML.METADATA_ELEM);
|
---|
| 636 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 637 | attribute.setValue(namespacePrefix + name);
|
---|
| 638 | // prefix with namespace, if any
|
---|
| 639 | metadata.setAttributeNode(attribute);
|
---|
| 640 | Text content = doc.createTextNode(value);
|
---|
| 641 | metadata.appendChild(content);
|
---|
| 642 |
|
---|
| 643 | metadataList.appendChild(metadata);
|
---|
| 644 | }
|
---|
| 645 | }
|
---|
| 646 |
|
---|
| 647 | /** Given a document identifier, returns a GS3 DocumentMetadataRetrieve
|
---|
| 648 | * response message containing ONLY the Title metadata for the document.
|
---|
| 649 | * @param docID is a document identifier (docID can either be a <pid>
|
---|
| 650 | * of an item (document) in the fedora repository, or it can be
|
---|
| 651 | * "<pid>-sectionNumber".
|
---|
| 652 | * @return a GS3 DocumentMetadataRetrieve response message containing the
|
---|
| 653 | * Title metadata for the requested document */
|
---|
| 654 | public String getTitleMetadata(String docID) {
|
---|
| 655 | return getTitleMetadata(new String[] { docID });
|
---|
| 656 | }
|
---|
| 657 |
|
---|
| 658 | /** Given a document identifier, returns a GS3 DocumentMetadataRetrieve
|
---|
| 659 | * response message containing ONLY the Title metadata for the documents.
|
---|
| 660 | * @param docIDs is a list of document identifiers (where docID can either be
|
---|
| 661 | * a <pid> of an item (document) in the fedora repository, or it can be
|
---|
| 662 | * "<pid>-sectionNumber".
|
---|
| 663 | * @return a GS3 DocumentMetadataRetrieve response message containing the
|
---|
| 664 | * Title metadata for all the requested documents */
|
---|
| 665 | public String getTitleMetadata(String[] docIDs) {
|
---|
| 666 | // Must create message of the following form:
|
---|
| 667 | // <documentNodeList><documentNode nodeID="docID">
|
---|
| 668 | // <metadataList><metadata name="Title">sometitle</metadata>
|
---|
| 669 | // </metadataList></documentNode>
|
---|
| 670 |
|
---|
| 671 | Document doc = builder.newDocument();
|
---|
| 672 | FedoraGS3RunException ex = null;
|
---|
| 673 |
|
---|
| 674 | Element docNodeList = doc.createElement(
|
---|
| 675 | GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 676 | try{
|
---|
| 677 | for(int i = 0; i < docIDs.length; i++) {
|
---|
| 678 | Element docNode = getTitleMetadata(doc, docIDs[i]);
|
---|
| 679 | docNodeList.appendChild(docNode);
|
---|
| 680 | }
|
---|
| 681 | }catch(Exception e) {
|
---|
| 682 | ex = new FedoraGS3RunException(e);
|
---|
[21573] | 683 | //ex.setSpecifics("EX metadata datastream PID: |" + docIDs[i] + "|"); // for debugging PID
|
---|
[15222] | 684 | ex.setSpecifics("EX metadata datastream");
|
---|
| 685 | }
|
---|
| 686 |
|
---|
| 687 | Element responseMsg = createResponseMessage(doc, docNodeList, ex,
|
---|
| 688 | GSXML.REQUEST_TYPE_PROCESS, "DocumentMetadataRetrieve");
|
---|
| 689 | try{
|
---|
| 690 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 691 | } catch(TransformerException e) {
|
---|
| 692 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 693 | + " " + e;
|
---|
| 694 | }
|
---|
| 695 | }
|
---|
| 696 |
|
---|
| 697 | /** Method that takes a new DOM document, as well as an identifier of either
|
---|
| 698 | * a document or document section and returns a documentNode element containing
|
---|
| 699 | * the title metadata for it:
|
---|
| 700 | * <documentNode nodeID="docID"><metadataList>
|
---|
| 701 | * <metadata name="Title">sometitle</metadata>
|
---|
| 702 | * </metadataList></documentNode>
|
---|
| 703 | * @return documentNode containing the metadata for the collection or
|
---|
| 704 | * document given by parameter ID
|
---|
| 705 | * @param docID denotes the id of a document or a document section, so id
|
---|
| 706 | * is either a document-pid or it's of the form documentpid-sectionNumber */
|
---|
| 707 | protected Element getTitleMetadata(Document doc, String docID)
|
---|
| 708 | throws RemoteException, UnsupportedEncodingException,
|
---|
| 709 | SAXException, IOException
|
---|
| 710 | {
|
---|
| 711 | // Returns a docNode element of the following form:
|
---|
| 712 | // <documentNode nodeID="docID">
|
---|
| 713 | // <metadataList><metadata name="Title">sometitle</metadata></metadataList>
|
---|
| 714 | // </documentNode>
|
---|
| 715 |
|
---|
| 716 | // <documentNode nodeID="docID">
|
---|
| 717 | Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
| 718 | Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
| 719 | attribute.setValue(docID);
|
---|
| 720 | docNode.setAttributeNode(attribute);
|
---|
| 721 |
|
---|
| 722 | // <metadataList>
|
---|
| 723 | Element metaList = doc.createElement(
|
---|
| 724 | GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 725 | // <metadata name="Title">
|
---|
| 726 | Element metadata = doc.createElement(GSXML.METADATA_ELEM);
|
---|
| 727 | // if we connect it all up (append children), we can immediately add
|
---|
| 728 | // the name attribute into the metadata element:
|
---|
| 729 | metaList.appendChild(metadata);
|
---|
| 730 | docNode.appendChild(metaList);
|
---|
| 731 | metadata.setAttribute(GSXML.NAME_ATT, TITLE); // immediately add attribute
|
---|
| 732 |
|
---|
| 733 | String title = "";
|
---|
| 734 | String sectionID = getSectionIDFromDocID(docID);
|
---|
| 735 | String docPID = getDocPIDFromDocID(docID);
|
---|
| 736 |
|
---|
| 737 | // check if title of toplevel document is requested
|
---|
| 738 | if(sectionID.equals(""))
|
---|
| 739 | title = this.getDocTitle(docPID);
|
---|
| 740 | else { // title of document section
|
---|
| 741 | title = this.getSectionTitle(docPID, sectionID);
|
---|
| 742 | }
|
---|
| 743 |
|
---|
| 744 | metadata.appendChild(doc.createTextNode(title));
|
---|
| 745 |
|
---|
| 746 | return docNode;
|
---|
| 747 | }
|
---|
| 748 |
|
---|
| 749 |
|
---|
| 750 | /** @return the documentStructure of the document or section given by docID.
|
---|
| 751 | * The structure is returned in the XML format of a Greenstone3
|
---|
| 752 | * DocumentStructureRetrieve response message. This method returns the entire
|
---|
| 753 | * subSection of the docID (that is, all descendants included).
|
---|
| 754 | * @param docID the identifier for the document whose structure is required.
|
---|
| 755 | * This is of the format "greenstone:<collectionName>-<docPID>"
|
---|
| 756 | * OR "greenstone:<collectionName>-<docPID>-<sectioNumber>"
|
---|
| 757 | * where "greenstone:<collectionName>-<docPID>-1" is the same as
|
---|
| 758 | * "greenstone:<collectionName>-<docPID>" and will return the
|
---|
| 759 | * same response */
|
---|
| 760 | public String getDocumentStructure(String docID) {
|
---|
| 761 | return getStructure(new String[]{docID}, DESCENDANTS);
|
---|
| 762 | }
|
---|
| 763 |
|
---|
| 764 | /** @return a view of the structure of the document or section given by docID
|
---|
| 765 | * which contains only the section and its direct children. This structure is
|
---|
| 766 | * returned in the XML format of a Greenstone3 DocumentStructureRetrieve
|
---|
| 767 | * response message.
|
---|
| 768 | * @param docID the identifier for the document whose structure is required.
|
---|
| 769 | * This is of the format "greenstone:<collectionName>-<docPID>"
|
---|
| 770 | * OR "greenstone:<collectionName>-<docPID>-<sectioNumber>"
|
---|
| 771 | * where "greenstone:<collectionName>-<docPID>-1" is the same as
|
---|
| 772 | * "greenstone:<collectionName>-<docPID>" and will return the
|
---|
| 773 | * same response */
|
---|
| 774 | public String getChildren(String docID) {
|
---|
| 775 | return getStructure(new String[]{docID}, CHILDREN);
|
---|
| 776 | }
|
---|
| 777 |
|
---|
| 778 | /** @return the documentStructure of the documents or sections given by docIDs.
|
---|
| 779 | * The structure is returned in the XML format of a Greenstone3
|
---|
| 780 | * DocumentStructureRetrieve response message. This method returns the entire
|
---|
| 781 | * subSection of each docID (that is, all descendants included).
|
---|
| 782 | * @param docIDs is an array of identifiers for the documents whose structures
|
---|
| 783 | * are required.
|
---|
| 784 | * This is of the format "greenstone:<collectionName>-<docPID>"
|
---|
| 785 | * OR "greenstone:<collectionName>-<docPID>-<sectioNumber>"
|
---|
| 786 | * where "greenstone:<collectionName>-<docPID>-1" is the same as
|
---|
| 787 | * "greenstone:<collectionName>-<docPID>" and will return the
|
---|
| 788 | * same response */
|
---|
| 789 | public String getDocumentStructure(String[] docIDs) {
|
---|
| 790 | return getStructure(docIDs, DESCENDANTS);
|
---|
| 791 | }
|
---|
| 792 |
|
---|
| 793 | /** @return the documentStructure of the documents or sections given by docIDs
|
---|
| 794 | * but only the sections and their children (not any further descendants).
|
---|
| 795 | * The structure is returned in the XML format of a Greenstone3
|
---|
| 796 | * DocumentStructureRetrieve response message.
|
---|
| 797 | * @param docIDs the identifiers for the documents whose structures are
|
---|
| 798 | * required. The docids are of the format "greenstone:<collectionName>-<docPID>"
|
---|
| 799 | * OR "greenstone:<collectionName>-<docPID>-<sectioNumber>"
|
---|
| 800 | * where "greenstone:<collectionName>-<docPID>-1" is the same as
|
---|
| 801 | * "greenstone:<collectionName>-<docPID>" and will return the
|
---|
| 802 | * same response */
|
---|
| 803 | public String getChildren(String[] docIDs) {
|
---|
| 804 | return getStructure(docIDs, CHILDREN);
|
---|
| 805 | }
|
---|
| 806 |
|
---|
| 807 | /**
|
---|
| 808 | * Returns a greenstone3 DocumentStructureRetrieve XML response message
|
---|
| 809 | * containing the document structures for the given docIDs.
|
---|
| 810 | * Similar to FedoraConnection.getTOC(), but instead of fedora formatted XML,
|
---|
| 811 | * greenstone formatted XML is returned. The requested section of the table
|
---|
| 812 | * of contents (TOC) for a document is converted into the greenstone3 xml
|
---|
| 813 | * format that is returned upon DocumentStructureRetrieve requests.
|
---|
| 814 | * @param docIDs the documentIDs for which the section's structure is returned;
|
---|
| 815 | * where a docID is either a fedora pid <docPID> or <docPID>-<sectionNumber>.
|
---|
| 816 | * @param levels - either CHILDREN or DESCENDANTS.
|
---|
| 817 | * CHILDREN returns only the first-level descendants (children) of the
|
---|
| 818 | * requested document sections indicated by docIDs.
|
---|
| 819 | * DESCENDANTS returns all descendants of all the document-sections denoted by
|
---|
| 820 | * docIDs.
|
---|
| 821 | * @return a greenstone3 DocumentStructureRetrieve XML response message in
|
---|
| 822 | * String format with the structure of the docIDs requested.
|
---|
| 823 | */
|
---|
| 824 | protected String getStructure(String[] docIDs, int levels)
|
---|
| 825 | {
|
---|
| 826 | Document doc = builder.newDocument();
|
---|
| 827 | FedoraGS3RunException ex = null;
|
---|
| 828 | // <documentNodeList>
|
---|
| 829 | Element docNodeList = doc.createElement(
|
---|
| 830 | GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 831 |
|
---|
| 832 | try{
|
---|
| 833 | // append the <documentNodes> for the docIDs
|
---|
| 834 | // to the docNodeList
|
---|
| 835 | getStructureElement(docNodeList, docIDs, levels);
|
---|
| 836 | } catch(Exception e) {
|
---|
| 837 | ex = new FedoraGS3RunException(e);
|
---|
| 838 | ex.setSpecifics("(requested portion of) TOC datastream");
|
---|
| 839 | }
|
---|
| 840 | // insert our <documentNodeList> into a GS3 response message
|
---|
| 841 | Element responseMsg = createResponseMessage(doc, docNodeList, ex,
|
---|
| 842 | GSXML.REQUEST_TYPE_PROCESS, "DocumentStructureRetrieve");
|
---|
| 843 | try{
|
---|
| 844 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 845 | } catch(TransformerException e) {
|
---|
| 846 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 847 | + " " + e;
|
---|
| 848 | }
|
---|
| 849 | }
|
---|
| 850 |
|
---|
| 851 | /** Given a <documentNodeList> portion of a greenstone3
|
---|
| 852 | * DocumentStructureRetrieve XML response message, this method will populate
|
---|
| 853 | * it with the <documentNodes> that represent the structure of the given docIDs.
|
---|
| 854 | * @param docNodeList is a <documentNodeList> to which <documentNodes> of
|
---|
| 855 | * the doc structures are appended.
|
---|
| 856 | * @param docIDs the documentIDs for which the section's structure is returned;
|
---|
| 857 | * where a docID is either a fedora pid <docPID> or <docPID>-<sectionNumber>.
|
---|
| 858 | * @param levels - either CHILDREN or DESCENDANTS.
|
---|
| 859 | * CHILDREN returns only the first-level descendants (children) of the
|
---|
| 860 | * requested document sections indicated by docIDs.
|
---|
| 861 | * DESCENDANTS returns all descendants of all the document-sections denoted by
|
---|
| 862 | * docIDs.
|
---|
| 863 | */
|
---|
| 864 | protected void getStructureElement(Element docNodeList,
|
---|
| 865 | String[] docIDs, int levels)
|
---|
| 866 | throws RemoteException, UnsupportedEncodingException, SAXException,
|
---|
| 867 | IOException
|
---|
| 868 | {
|
---|
| 869 | // process each docID
|
---|
| 870 | for(int i = 0; i < docIDs.length; i++) {
|
---|
| 871 | // work out the document's fedora PID and section ID
|
---|
| 872 | String sectionID = getSectionIDFromDocID(docIDs[i]);
|
---|
| 873 | String docPID = getDocPIDFromDocID(docIDs[i]);
|
---|
| 874 |
|
---|
| 875 | // get the required section, along with children or descendants
|
---|
| 876 | Element srcDocElement = null;
|
---|
| 877 | if(levels == CHILDREN) // get the requested section with its children
|
---|
| 878 | srcDocElement = this.getChildrenOfSectionXML(docPID, sectionID);
|
---|
| 879 | else // levels == DESCENDANTS, get the section with all its descendants
|
---|
| 880 | srcDocElement = this.getSubsectionXML(docPID, sectionID);
|
---|
| 881 |
|
---|
| 882 | // copy-and-convert that structure into a structure format for GS3
|
---|
| 883 | Element docNode = getStructure(docNodeList.getOwnerDocument(),
|
---|
| 884 | docIDs[i], docPID, srcDocElement);
|
---|
| 885 |
|
---|
| 886 | // add it to our list of documentNodes
|
---|
| 887 | docNodeList.appendChild(docNode);
|
---|
| 888 | }
|
---|
| 889 | }
|
---|
| 890 |
|
---|
| 891 | /**
|
---|
| 892 | * Takes the portion of the XML document outlining the structure of the
|
---|
| 893 | * document (section)--in the format this is stored in Fedora--and returns
|
---|
| 894 | * Greenstone 3 DOM XML format for outlining document structure.
|
---|
| 895 | * @return a <documentNode> element that contains a greenstone3
|
---|
| 896 | * DocumentStructureRetrieve XML corresponding to the parameter Element section
|
---|
| 897 | * (which is in fedora XML), for the document indicated by docID.
|
---|
| 898 | * @param requestingDocID is the identifier of the document for which the
|
---|
| 899 | * structure was requested. It's this document's children or descendants that
|
---|
| 900 | * will be returned. Note that this is not always the same as (clear from)
|
---|
| 901 | * parameter docID.
|
---|
| 902 | * @param docID is the documentID for which the section's structure is
|
---|
| 903 | * returned where docID = "docPID-sectionNumber".
|
---|
| 904 | * @param section - the fedora section XML that is being mirrored in
|
---|
| 905 | * greenstone3 format.
|
---|
| 906 | */
|
---|
| 907 | protected Element getStructure(Document doc, String requestingDocID,
|
---|
| 908 | String docID, Element section)
|
---|
| 909 | {
|
---|
| 910 | // we want to mirror the section's DOM (given in fedora XML) in
|
---|
| 911 | // greenstone3's XML for a DocumentStructureRetrieve response.
|
---|
| 912 |
|
---|
| 913 | // <documentNode nodeID="docID"> - the docNode on which a structure retrieve
|
---|
| 914 | // is being performed
|
---|
| 915 | Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
| 916 | Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
| 917 | attribute.setValue(requestingDocID); //requestingDocID.replace(HYPHEN+SECTION, "")
|
---|
| 918 | docNode.setAttributeNode(attribute);
|
---|
| 919 |
|
---|
| 920 | // <nodeStructure>
|
---|
| 921 | Element nodeStructure = doc.createElement(GSXML.NODE_STRUCTURE_ELEM);
|
---|
| 922 |
|
---|
| 923 | // <documentNode nodeID="docID" docType="hierarchy" nodeType="root">
|
---|
| 924 | Element rootNode = createDocNodeFromSubsection(doc, section, docID);
|
---|
| 925 |
|
---|
| 926 | // fills in the subtree of the rootNode in our nodeStructure element
|
---|
| 927 | createDocStructure(doc, section, rootNode, docID);
|
---|
| 928 | //where section represents the root section
|
---|
| 929 |
|
---|
| 930 | nodeStructure.appendChild(rootNode);
|
---|
| 931 | docNode.appendChild(nodeStructure);
|
---|
| 932 | return docNode;
|
---|
| 933 | }
|
---|
| 934 |
|
---|
| 935 | /** Recursive method that creates a documentStructure mirroring parameter
|
---|
| 936 | * section, starting from parameter parent down to all descendants
|
---|
| 937 | * @param section is the XML <Section> in the fedora repository's TOC
|
---|
| 938 | * for the docPID whose substructure is to be mirrored
|
---|
| 939 | * @param parent is the XML documentNode in the greenstone repository whose
|
---|
| 940 | * descendants created by this method will correspond to the descendants of
|
---|
| 941 | * parameter section.
|
---|
| 942 | * @param doc is the document containing the parent;
|
---|
| 943 | * @param docPID is the prefix of all nodeIDs in the parent's structure
|
---|
| 944 | */
|
---|
| 945 | protected void createDocStructure(
|
---|
| 946 | Document doc, Element section, Element parent, String docPID)
|
---|
| 947 | {
|
---|
| 948 | // get the section's children (if any)
|
---|
| 949 | NodeList children = section.getChildNodes();
|
---|
| 950 | for(int i = 0; i < children.getLength(); i++) {
|
---|
| 951 | Node n = children.item(i);
|
---|
| 952 |
|
---|
| 953 | if(n.getNodeName().equals(SECTION_ELEMENT)) {
|
---|
| 954 | //then we know it's an element AND that its tagname is "Section"
|
---|
| 955 | Element subsection = (Element)n;
|
---|
| 956 | Element child = createDocNodeFromSubsection(doc, subsection, docPID);
|
---|
| 957 | parent.appendChild(child);
|
---|
| 958 |
|
---|
| 959 | // recursion call on newly found child-element and subsection
|
---|
| 960 | createDocStructure(doc, subsection, child, docPID);
|
---|
| 961 | }
|
---|
| 962 | }
|
---|
| 963 | }
|
---|
| 964 |
|
---|
| 965 | /** Given a particular subsection element, this method creates a
|
---|
| 966 | * Greenstone3 DocumentNode element that mirrors it.
|
---|
| 967 | * @param doc is the document that will contain the created DocumentNode
|
---|
| 968 | * @param docID is the prefix of all nodeIDs in the parent's structure
|
---|
| 969 | * @param subSection is the XML <Section> in the fedora repository's
|
---|
| 970 | * TOC for the docPID which will be mirrored in the greenstone XML
|
---|
| 971 | * documentNode that will be returned.
|
---|
| 972 | * @return a greenstone <documentNode> that represents the fedora TOC's
|
---|
| 973 | * <Section> element passed as parameter subSection. */
|
---|
| 974 | protected Element createDocNodeFromSubsection(
|
---|
| 975 | Document doc, Element subSection, String docID)
|
---|
| 976 | {
|
---|
| 977 | Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
| 978 | Attr docType = doc.createAttribute(GSXML.DOC_TYPE_ATT);
|
---|
| 979 | docType.setValue(GSXML.DOC_TYPE_HIERARCHY);
|
---|
| 980 | docNode.setAttributeNode(docType);
|
---|
| 981 |
|
---|
| 982 | Attr nodeID = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
| 983 | String sectionID = subSection.hasAttribute(ID) ?
|
---|
| 984 | subSection.getAttribute(ID) : "";
|
---|
| 985 | nodeID.setValue(docID + HYPHEN + sectionID);
|
---|
| 986 | docNode.setAttributeNode(nodeID);
|
---|
| 987 |
|
---|
| 988 | Attr nodeType = doc.createAttribute(GSXML.NODE_TYPE_ATT);
|
---|
| 989 | if(sectionID.equals("1")) { // root case
|
---|
| 990 | nodeType.setValue(GSXML.NODE_TYPE_ROOT);
|
---|
| 991 | // reset the attribute without the section number
|
---|
| 992 | docNode.setAttribute(GSXML.NODE_ID_ATT, docID);
|
---|
| 993 | }
|
---|
| 994 | else if(subSection.getElementsByTagName(SECTION_ELEMENT).getLength() > 0)
|
---|
| 995 | // this section has further <Section> children, so it's an internal node
|
---|
| 996 | nodeType.setValue(GSXML.NODE_TYPE_INTERNAL);
|
---|
| 997 | else if(subSection.hasAttribute(TYPE))
|
---|
| 998 | nodeType.setValue(GSXML.NODE_TYPE_INTERNAL);
|
---|
| 999 | else // leaf
|
---|
| 1000 | nodeType.setValue(GSXML.NODE_TYPE_LEAF);
|
---|
| 1001 | docNode.setAttributeNode(nodeType);
|
---|
| 1002 | return docNode;
|
---|
| 1003 | }
|
---|
| 1004 |
|
---|
| 1005 |
|
---|
| 1006 | /** Given an identifier that is either a docPID or a concatenation of
|
---|
| 1007 | * docPID+sectionID, this method works out the fedora assigned docPID and
|
---|
| 1008 | * sectionID and then calls getContentBody(docPID, sectionID) with those.
|
---|
| 1009 | * @param docID is expected to be of the form
|
---|
| 1010 | * "greenstone:<collectionName>-<docPID>-<sectionNumber>" or
|
---|
| 1011 | * "greenstone:<collectionName>-<docPID>"
|
---|
| 1012 | * If it is "greenstone:<collectionName>-<docPID>", then the content for
|
---|
| 1013 | * "greenstone:<collectionName>-1" ("greenstone:<collectionName>-Section1")
|
---|
| 1014 | * is returned! */
|
---|
| 1015 | public String getContent(String docID) {
|
---|
| 1016 | return this.getContent(new String[]{docID});
|
---|
| 1017 | }
|
---|
| 1018 |
|
---|
| 1019 | /** Given an identifier that is a concatenation of docID+sectionID, this
|
---|
| 1020 | * method works out the fedora assigned docPID and sectionID and then calls
|
---|
| 1021 | * getContentBody(docPID, sectionID) with those.
|
---|
| 1022 | * @param docIDs is an array of document identifiers of the form
|
---|
| 1023 | * "greenstone:<collectionName>-<docPID>-<sectionNumber>"
|
---|
| 1024 | * If it is "greenstone:<collectionName>-<docPID>", then the content for
|
---|
| 1025 | * "greenstone:<collectionName>-Section1" is returned! */
|
---|
| 1026 | public String getContent(String[] docIDs) {
|
---|
| 1027 | Document doc = builder.newDocument();
|
---|
| 1028 | FedoraGS3RunException ex = null;
|
---|
| 1029 |
|
---|
| 1030 | //<documentNodeList>
|
---|
| 1031 | Element docNodeList = doc.createElement(
|
---|
| 1032 | GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1033 |
|
---|
| 1034 | try{
|
---|
| 1035 | for(int i = 0; i < docIDs.length; i++) {
|
---|
| 1036 | // get the sectionID and docPID from the docID
|
---|
| 1037 | String sectionID = this.removePrefix(
|
---|
| 1038 | getSectionIDFromDocID(docIDs[i]), SECTION);
|
---|
| 1039 | String docPID = getDocPIDFromDocID(docIDs[i]);
|
---|
| 1040 | if(sectionID.equals("")) // if no section is specified, get
|
---|
| 1041 | sectionID = "1"; // get the content for Section id="1"
|
---|
| 1042 |
|
---|
| 1043 | // Get the contents for the requested section of document docPID
|
---|
| 1044 | String sectionContent = this.getContentBody(docPID, sectionID);
|
---|
| 1045 |
|
---|
| 1046 | // set the nodeID attribute
|
---|
| 1047 | Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
| 1048 | Attr nodeId = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
| 1049 |
|
---|
| 1050 | nodeId.setValue(docIDs[i]); // just set the docID which will contain
|
---|
| 1051 | // the docPID (and sectionID if already present)
|
---|
| 1052 |
|
---|
| 1053 | docNode.setAttributeNode(nodeId);
|
---|
| 1054 | // set the text content to what was retrieved
|
---|
| 1055 | Element nodeContent = doc.createElement(GSXML.NODE_CONTENT_ELEM);
|
---|
[21775] | 1056 | Text textNode = doc.createTextNode(sectionContent.trim());
|
---|
[15222] | 1057 |
|
---|
| 1058 | nodeContent.appendChild(textNode);
|
---|
| 1059 | docNode.appendChild(nodeContent);
|
---|
| 1060 | //add the documentNode to the docNodeList
|
---|
| 1061 | docNodeList.appendChild(docNode);
|
---|
| 1062 | }
|
---|
| 1063 | } catch(Exception e) {
|
---|
| 1064 | ex = new FedoraGS3RunException(e);
|
---|
| 1065 | ex.setSpecifics("requested doc Section datastream");
|
---|
| 1066 | }
|
---|
| 1067 | Element responseMsg = createResponseMessage(doc, docNodeList, ex,
|
---|
| 1068 | GSXML.REQUEST_TYPE_PROCESS, "DocumentContentRetrieve");
|
---|
| 1069 | try{
|
---|
| 1070 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 1071 | } catch(TransformerException e) {
|
---|
| 1072 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 1073 | + " " + e;
|
---|
| 1074 | }
|
---|
| 1075 | }
|
---|
| 1076 |
|
---|
| 1077 | /** Gets the contents of a textNode from a section.
|
---|
| 1078 | * @return the text content of a section.
|
---|
| 1079 | * @param docPID the pid of the document from which a section's text is to
|
---|
| 1080 | * be retrieved.
|
---|
| 1081 | * @param sectionID is the section identifier of the document denoted by
|
---|
| 1082 | * docPID whose text is to be returned.
|
---|
| 1083 | */
|
---|
| 1084 | protected String getContentBody(String docPID, String sectionID)
|
---|
| 1085 | throws RemoteException, UnsupportedEncodingException,
|
---|
| 1086 | SAXException, IOException
|
---|
| 1087 | {
|
---|
| 1088 | String section = this.getSection(docPID, sectionID);
|
---|
| 1089 |
|
---|
| 1090 | // the content is nested inside a <Section> element,
|
---|
| 1091 | // we extract it from there:
|
---|
| 1092 | InputSource source = new InputSource(new StringReader(section));
|
---|
| 1093 | Document doc = builder.parse(source);
|
---|
| 1094 |
|
---|
| 1095 | // The document Element is the <Section> we want.
|
---|
| 1096 | // Get its text contents:
|
---|
| 1097 | section = FedoraCommons.getValue(doc.getDocumentElement());
|
---|
| 1098 |
|
---|
| 1099 | // we are going to remove all occurrences of "_httpdocimg_/"
|
---|
| 1100 | // that precede associated filenames, because that's a GS3
|
---|
| 1101 | // defined macro for resolving relative urls. It won't help
|
---|
| 1102 | // with documents stored in fedora.
|
---|
| 1103 | section = section.replaceAll(GS3FilePathMacro+"/", "");
|
---|
| 1104 | return section;
|
---|
| 1105 | }
|
---|
| 1106 |
|
---|
| 1107 | /** Here we create the greenstone's response message element:
|
---|
| 1108 | * <message≶<response><content></response></message>
|
---|
| 1109 | * @return a greenstone response-message element.
|
---|
| 1110 | * @param doc - the Document object which should me used to create the
|
---|
| 1111 | * <message> and <response> elements
|
---|
| 1112 | * @param content - the element that is to be nested inside <response>
|
---|
| 1113 | * @param ex - any exception that occurred when trying to create
|
---|
| 1114 | * the content parameter
|
---|
| 1115 | * @param responseType - the value for the type attribute of <response>,
|
---|
| 1116 | * such as "describe", "retrieve", "browse", "query"...
|
---|
| 1117 | * @param originator - indiates the collectionName or service (like
|
---|
| 1118 | * DocumentContentRetrieve) from where this response message originates
|
---|
| 1119 | */
|
---|
| 1120 | protected Element createResponseMessage(Document doc, Element content,
|
---|
| 1121 | Exception ex, String responseType, String originator)
|
---|
| 1122 | {
|
---|
| 1123 | Element response = doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
| 1124 | // from = "FedoraGS3"
|
---|
| 1125 | Attr attribute = doc.createAttribute(GSXML.FROM_ATT);
|
---|
| 1126 | String from = originator.equals("") ? FEDORA_GS3
|
---|
| 1127 | : FEDORA_GS3+"/"+originator;
|
---|
| 1128 |
|
---|
| 1129 | attribute.setValue(from);
|
---|
| 1130 | response.setAttributeNode(attribute);
|
---|
| 1131 |
|
---|
| 1132 | // type = "describe" or "process" - whatever's given in requestType:
|
---|
| 1133 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1134 | attribute.setValue(responseType);
|
---|
| 1135 | response.setAttributeNode(attribute);
|
---|
| 1136 |
|
---|
| 1137 | if(content != null)
|
---|
| 1138 | response.appendChild(content);
|
---|
| 1139 |
|
---|
| 1140 | // we'll create an error element for RemoteExceptions (web service problems)
|
---|
| 1141 | // and UnsupportedEncodingExceptions and
|
---|
| 1142 | if(ex != null) {
|
---|
| 1143 | Element error = doc.createElement(GSXML.ERROR_ELEM);
|
---|
| 1144 | error.appendChild(doc.createTextNode(ex.getMessage()));
|
---|
| 1145 | // now append the error to the <response> element (after
|
---|
| 1146 | // the content element whatever that was)
|
---|
| 1147 | response.appendChild(error);
|
---|
| 1148 | }
|
---|
| 1149 |
|
---|
| 1150 | Element message = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
| 1151 | message.appendChild(response);
|
---|
| 1152 | doc.appendChild(message);
|
---|
| 1153 | return message;
|
---|
| 1154 | }
|
---|
| 1155 |
|
---|
| 1156 | /** @return a <serviceList> Element as defined by GS3: containing all the
|
---|
| 1157 | * services (denoted by <service> elements) that are supported by FedoraGS3.
|
---|
| 1158 | * At present these are: DocumentContentRetrieve, DocumentMetadataRetrieve,
|
---|
| 1159 | * DocumentStructureRetrieve, TextQuery, FieldQuery, ClassifierBrowse,
|
---|
| 1160 | * ClassifierBrowseMetadataRetrieve (as indicated by member var serviceNames).
|
---|
| 1161 | * @param doc - the Document object which should me used to create the
|
---|
| 1162 | * <serviceList> element */
|
---|
| 1163 | protected Element createServiceList(Document doc)
|
---|
| 1164 | {
|
---|
| 1165 | Element serviceList = doc.createElement(
|
---|
| 1166 | GSXML.SERVICE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1167 |
|
---|
| 1168 | for(int i = 0; i < serviceNames.length; i++) {
|
---|
| 1169 | // create the <service name="serviceName[i]" type="servicetype" />
|
---|
| 1170 | Element service = doc.createElement(GSXML.SERVICE_ELEM);
|
---|
| 1171 |
|
---|
| 1172 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1173 | attribute.setValue(serviceNames[i]);
|
---|
| 1174 | service.setAttributeNode(attribute);
|
---|
| 1175 |
|
---|
| 1176 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1177 | if(serviceNames[i].equals("ClassifierBrowse")) //browseTitlesByLetter
|
---|
| 1178 | attribute.setValue(GSXML.SERVICE_TYPE_BROWSE);
|
---|
| 1179 | else if(serviceNames[i].contains("Query")) // search services
|
---|
| 1180 | attribute.setValue(GSXML.SERVICE_TYPE_QUERY);
|
---|
| 1181 | else
|
---|
| 1182 | attribute.setValue(GSXML.SERVICE_TYPE_RETRIEVE);
|
---|
| 1183 | service.setAttributeNode(attribute);
|
---|
| 1184 |
|
---|
| 1185 | // add the service element to the serviceList element
|
---|
| 1186 | // <serviceList><service /></serviceList>
|
---|
| 1187 | serviceList.appendChild(service);
|
---|
| 1188 | }
|
---|
| 1189 | return serviceList;
|
---|
| 1190 | }
|
---|
| 1191 |
|
---|
| 1192 | /** @return a GS3 response message for a describe services request:
|
---|
| 1193 | * indicating the list of services supported by the Fedora-Greenstone
|
---|
| 1194 | * interface. These are DocumentContentRetrieve, DocumentMetadataRetrieve,
|
---|
| 1195 | * DocumentStructureRetrieve, ClassifierBrowse, TextQuery, FieldQuery,
|
---|
| 1196 | * ClassifierBrowseMetadataRetrieve - as indicated by member variable
|
---|
| 1197 | * serviceNames. */
|
---|
| 1198 | public String getServiceList()
|
---|
| 1199 | {
|
---|
| 1200 | Document doc = builder.newDocument();
|
---|
| 1201 | Element serviceList = createServiceList(doc);
|
---|
| 1202 | // make <serviceList> the body of the responseMessage:
|
---|
| 1203 | // <message><response><serviceList></response></message>
|
---|
| 1204 | Element responseMsg = createResponseMessage(doc, serviceList, null,
|
---|
| 1205 | GSXML.REQUEST_TYPE_DESCRIBE, "");
|
---|
| 1206 | try {
|
---|
| 1207 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 1208 | }catch(TransformerException e) {
|
---|
| 1209 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 1210 | + " " + e;
|
---|
| 1211 | }
|
---|
| 1212 | }
|
---|
| 1213 |
|
---|
| 1214 | /** @return a GS3 describe response message listing the collections and
|
---|
| 1215 | * collection-specific metadata stored in the Fedora-Greenstone repository. */
|
---|
| 1216 | public String getCollectionList()
|
---|
| 1217 | {
|
---|
| 1218 | Document doc = builder.newDocument();
|
---|
| 1219 | FedoraGS3RunException ex = null; // any RemoteException
|
---|
| 1220 |
|
---|
| 1221 | // create the <collectionList /> element
|
---|
| 1222 | Element collectionList = doc.createElement(
|
---|
| 1223 | GSXML.COLLECTION_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1224 | try{
|
---|
| 1225 | String[] collectionNames = this.getCollectionNames(
|
---|
| 1226 | this.getCollections()); // this line could throw RemoteException
|
---|
| 1227 | for(int i = 0; i < collectionNames.length; i++) {
|
---|
| 1228 | // create the <collection name="somename" /> element
|
---|
| 1229 | Element collection = doc.createElement(GSXML.COLLECTION_ELEM);
|
---|
| 1230 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1231 | attribute.setValue(collectionNames[i]);
|
---|
| 1232 | collection.setAttributeNode(attribute);
|
---|
| 1233 |
|
---|
| 1234 | // append the <collection> element as child of <collectionList>
|
---|
| 1235 | collectionList.appendChild(collection);
|
---|
| 1236 |
|
---|
| 1237 | //if(collection.hasAttribute(GSXML.NAME_ATT))
|
---|
| 1238 | //LOG.debug(collection.getAttribute(GSXML.NAME_ATT));
|
---|
| 1239 | }
|
---|
| 1240 | } catch(RemoteException e) { // if this happens, perhaps it's because it
|
---|
| 1241 | // can't find Greenstone collections in fedora repository?
|
---|
| 1242 | ex = new FedoraGS3RunException(e);
|
---|
| 1243 | ex.setSpecifics(
|
---|
| 1244 | "greenstone collections in fedora repository");
|
---|
| 1245 | }
|
---|
| 1246 |
|
---|
| 1247 | // make <collectionList> the body of the responseMessage:
|
---|
| 1248 | // <message><response><collectionList></response></message>
|
---|
| 1249 | Element responseMsg = createResponseMessage(doc, collectionList, ex,
|
---|
| 1250 | GSXML.REQUEST_TYPE_DESCRIBE, "");
|
---|
| 1251 | try{
|
---|
| 1252 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 1253 | }catch(TransformerException e) {
|
---|
| 1254 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 1255 | + " " + e;
|
---|
| 1256 | }
|
---|
| 1257 | }
|
---|
| 1258 |
|
---|
| 1259 | /** @return a GS3 describe response message for a collection in the
|
---|
| 1260 | * Fedora-Greenstone repository.
|
---|
| 1261 | * @param collectionName - the name of the collection that is to be described.
|
---|
| 1262 | * It will be converted to a fedora collection pid, which is of the form
|
---|
| 1263 | * "greenstone:<collectionName>-collection". */
|
---|
| 1264 | public String describeCollection(String collectionName)
|
---|
| 1265 | {
|
---|
| 1266 | Document doc = builder.newDocument();
|
---|
| 1267 | FedoraGS3RunException ex = null;
|
---|
| 1268 |
|
---|
| 1269 | Element collection = doc.createElement(GSXML.COLLECTION_ELEM);
|
---|
| 1270 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1271 | attribute.setValue(collectionName);
|
---|
| 1272 | collection.setAttributeNode(attribute);
|
---|
| 1273 |
|
---|
| 1274 | //<displayItem assigned="true" lang="en" name="name">
|
---|
| 1275 | //"some display name"</displayItem>
|
---|
| 1276 | Element displayItem = doc.createElement(GSXML.DISPLAY_TEXT_ELEM);
|
---|
| 1277 |
|
---|
| 1278 | attribute = doc.createAttribute(GSXML.LANG_ATT);
|
---|
| 1279 | attribute.setValue(this.lang);
|
---|
| 1280 | displayItem.setAttributeNode(attribute);
|
---|
| 1281 |
|
---|
| 1282 | attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1283 | attribute.setValue(GSXML.DISPLAY_TEXT_NAME);
|
---|
| 1284 | displayItem.setAttributeNode(attribute);
|
---|
| 1285 |
|
---|
| 1286 | try{
|
---|
| 1287 | Text textNode = doc.createTextNode(
|
---|
| 1288 | this.getCollectionTitle(getCollectionPID(collectionName)));
|
---|
| 1289 | displayItem.appendChild(textNode);
|
---|
| 1290 | } catch(Exception e) {
|
---|
| 1291 | // can't find Greenstone collections in fedora repository or problem
|
---|
| 1292 | // getting their titles from their metadata datastream?
|
---|
| 1293 | ex = new FedoraGS3RunException(e);
|
---|
| 1294 | ex.setSpecifics("greenstone collections or their metadata"
|
---|
| 1295 | + "in the fedora repository");
|
---|
| 1296 | }
|
---|
| 1297 | // now append the displayItem element as child of the collection element
|
---|
| 1298 | collection.appendChild(displayItem);
|
---|
| 1299 | // get the <serviceList> and add it into the collection description.
|
---|
| 1300 | // Services for all collections in the FedoraGS3 repository are the
|
---|
| 1301 | // same, offering a ClassifierBrowse to browse titles by starting letter
|
---|
| 1302 | // and DocRetrieve services: Content, Metadata and Structure.
|
---|
| 1303 |
|
---|
| 1304 | Element serviceList = createServiceList(doc);
|
---|
| 1305 | collection.appendChild(serviceList);
|
---|
| 1306 |
|
---|
| 1307 | Element responseMsg = createResponseMessage(doc, collection, ex,
|
---|
| 1308 | GSXML.REQUEST_TYPE_DESCRIBE, collectionName);
|
---|
| 1309 | try{
|
---|
| 1310 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 1311 | }catch(TransformerException e) {
|
---|
| 1312 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 1313 | + " " + e;
|
---|
| 1314 | }
|
---|
| 1315 | }
|
---|
| 1316 |
|
---|
| 1317 | /** @return a GS3 describe response message for the services of a collection
|
---|
| 1318 | * in the Fedora-Greenstone repository. So far, these services are the same for
|
---|
| 1319 | * all fedora collections: they are the services given in member variable
|
---|
| 1320 | * serviceNames: DocumentContent/Metadata/StructureRetrieve, ClassifierBrowse,
|
---|
| 1321 | * ClassifierBrowseMetadataRetrieve.
|
---|
[21835] | 1322 | * All collections in this Digital Library (Fedora Repository) share the
|
---|
| 1323 | * same services, so this method returns the same services as getServiceList();
|
---|
[15222] | 1324 | * @param collectionName - the name of the collection whose services are to
|
---|
| 1325 | * be described. It will be converted to a fedora collection pid, which is of
|
---|
| 1326 | * the form "greenstone:<collectionName>-collection". */
|
---|
| 1327 | public String describeCollectionServices(String collectionName)
|
---|
| 1328 | {
|
---|
| 1329 | Document doc = builder.newDocument();
|
---|
| 1330 |
|
---|
| 1331 | Element collection = doc.createElement(GSXML.COLLECTION_ELEM);
|
---|
| 1332 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1333 | attribute.setValue(collectionName);
|
---|
| 1334 | collection.setAttributeNode(attribute);
|
---|
| 1335 |
|
---|
| 1336 | Element serviceList = createServiceList(doc);
|
---|
| 1337 | collection.appendChild(serviceList);
|
---|
| 1338 |
|
---|
| 1339 | Element responseMsg = createResponseMessage(doc, collection, null,
|
---|
| 1340 | GSXML.REQUEST_TYPE_DESCRIBE, collectionName);
|
---|
| 1341 | try{
|
---|
| 1342 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 1343 | }catch(TransformerException e) {
|
---|
| 1344 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 1345 | + " " + e;
|
---|
| 1346 | }
|
---|
| 1347 | }
|
---|
| 1348 |
|
---|
| 1349 | /** All collections in this Digital Library (Fedora Repository) share
|
---|
| 1350 | * the same services, so this method returns the same as
|
---|
| 1351 | * describeCollectionService(collName, serviceName).
|
---|
| 1352 | * @return a GS3 describe response message for the requested service
|
---|
| 1353 | * of the given collection. DocumentContent/Metadata/StructureRetrieve
|
---|
| 1354 | * return nothing special except their names; browse (and any query)
|
---|
| 1355 | * return more complex XML responses.
|
---|
| 1356 | * @param serviceName - the name of the service in the collection which is to
|
---|
| 1357 | * be described.*/
|
---|
| 1358 | public String describeService(String serviceName)
|
---|
| 1359 | {
|
---|
| 1360 | // For all the *retrieve* services (incl ClassifierBrowseMetadataRetrieve)
|
---|
| 1361 | // we return:
|
---|
| 1362 | // <message><response from="<name>Retrieve" type="describe">
|
---|
| 1363 | // <service name="<name>Retrieve" type="retrieve" /></response></message>
|
---|
| 1364 | // But for browse (and any query) service, we return the data necessary
|
---|
| 1365 | // for displaying it
|
---|
| 1366 |
|
---|
| 1367 | Document doc = this.builder.newDocument();
|
---|
| 1368 | Element service = doc.createElement(GSXML.SERVICE_ELEM);
|
---|
| 1369 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1370 | attribute.setValue(serviceName);
|
---|
| 1371 | service.setAttributeNode(attribute);
|
---|
| 1372 |
|
---|
| 1373 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1374 |
|
---|
[21775] | 1375 | if(serviceName.toLowerCase().endsWith("retrieve")) {
|
---|
[15222] | 1376 | attribute.setValue(GSXML.SERVICE_TYPE_RETRIEVE);
|
---|
[21775] | 1377 | }
|
---|
[15222] | 1378 | else if(serviceName.toLowerCase().contains("browse")) {
|
---|
| 1379 | attribute.setValue(GSXML.SERVICE_TYPE_BROWSE);
|
---|
| 1380 |
|
---|
| 1381 | // we need name and description <displayItem> elements
|
---|
| 1382 | Element displayItem
|
---|
| 1383 | = createNameValuePairElement(doc,
|
---|
| 1384 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, "Browse");
|
---|
| 1385 | service.appendChild(displayItem);
|
---|
| 1386 |
|
---|
| 1387 | displayItem = createNameValuePairElement(doc,
|
---|
| 1388 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_DESCRIPTION,
|
---|
| 1389 | "Browse pre-defined classification hierarchies");
|
---|
| 1390 | service.appendChild(displayItem);
|
---|
| 1391 |
|
---|
| 1392 | // now need a classifierList
|
---|
| 1393 | Element classifierList = doc.createElement(
|
---|
| 1394 | GSXML.CLASSIFIER_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1395 |
|
---|
| 1396 | int classifierNum = 1;
|
---|
| 1397 | // append a <classifier content="some letter" name="CL#">
|
---|
| 1398 | // for each letter of the alphabet:
|
---|
| 1399 | Element classifier = createClassifierElement(doc, "TitleByLetter",
|
---|
[15670] | 1400 | classifierNum++, "titles by letter", "Browse titles by letter");
|
---|
[15222] | 1401 | // now add this <classifier> to the <classifierList>
|
---|
| 1402 | classifierList.appendChild(classifier);
|
---|
| 1403 |
|
---|
| 1404 | // ANY MORE CLASSIFIERS? ADD THEM HERE
|
---|
| 1405 |
|
---|
| 1406 | service.appendChild(classifierList);
|
---|
| 1407 | } // ELSE check for whether it is a query service
|
---|
| 1408 | else if(serviceName.toLowerCase().contains("query")) {
|
---|
| 1409 | attribute.setValue(GSXML.SERVICE_TYPE_QUERY);
|
---|
[21775] | 1410 | if(serviceName.equals("TextQuery")) {
|
---|
[15222] | 1411 | describeTextQueryService(service);
|
---|
[21775] | 1412 | } else if(serviceName.equals("FieldQuery")) {
|
---|
[15222] | 1413 | describeFieldQueryService(service);
|
---|
[21775] | 1414 | }
|
---|
[15222] | 1415 | }
|
---|
| 1416 |
|
---|
| 1417 | // don't forget to add the type attribute to the service!
|
---|
| 1418 | service.setAttributeNode(attribute);
|
---|
| 1419 |
|
---|
| 1420 | String from = serviceName;
|
---|
| 1421 |
|
---|
| 1422 | Element responseMsg = createResponseMessage(doc, service, null,
|
---|
| 1423 | GSXML.REQUEST_TYPE_DESCRIBE, from);
|
---|
| 1424 | try{
|
---|
| 1425 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 1426 | }catch(TransformerException e) {
|
---|
| 1427 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 1428 | + " " + e;
|
---|
| 1429 | }
|
---|
| 1430 | }
|
---|
| 1431 |
|
---|
| 1432 | /** Appends children to the parameter service Element that make the
|
---|
| 1433 | * final service Element into a describe response XML for FedoraGS3's
|
---|
| 1434 | * TextQuery service.
|
---|
| 1435 | * @param service is the service Element that is being filled out. */
|
---|
| 1436 | protected void describeTextQueryService(Element service) {
|
---|
| 1437 | Document doc = service.getOwnerDocument();
|
---|
| 1438 | // we need name, submit (button) and description <displayItem> elements
|
---|
| 1439 | Element displayItem = createNameValuePairElement(doc,
|
---|
| 1440 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME,
|
---|
| 1441 | "Text Search");
|
---|
| 1442 | service.appendChild(displayItem);
|
---|
| 1443 |
|
---|
| 1444 | displayItem = createNameValuePairElement(doc,
|
---|
| 1445 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_SUBMIT, "Search");
|
---|
| 1446 | service.appendChild(displayItem);
|
---|
| 1447 |
|
---|
| 1448 | displayItem = createNameValuePairElement(doc,
|
---|
| 1449 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_DESCRIPTION,
|
---|
| 1450 | "Title and full-text search service");
|
---|
| 1451 | service.appendChild(displayItem);
|
---|
| 1452 |
|
---|
| 1453 | //create the <paramList>
|
---|
| 1454 | Element paramList = doc.createElement(
|
---|
| 1455 | GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1456 |
|
---|
| 1457 | // we ignore granularity to search at: it will always be
|
---|
| 1458 | // document and section level
|
---|
| 1459 | // we ignore casefolding: always on (that is, case is irrelevant)
|
---|
| 1460 | // we ignore document display order: always ranked
|
---|
| 1461 |
|
---|
| 1462 | // Constructing the following:
|
---|
| 1463 | // <param default="100" name="maxDocs" type="integer">
|
---|
| 1464 | // <displayItem name="name">Maximum hits to return</displayItem>
|
---|
| 1465 | // </param>
|
---|
| 1466 | Element param = doc.createElement(GSXML.PARAM_ELEM);
|
---|
| 1467 |
|
---|
| 1468 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1469 | attribute.setValue(MAXDOCS);
|
---|
| 1470 | param.setAttributeNode(attribute);
|
---|
| 1471 |
|
---|
| 1472 | attribute = doc.createAttribute(GSXML.DEFAULT_ATT);
|
---|
| 1473 | attribute.setValue("100");
|
---|
| 1474 | param.setAttributeNode(attribute);
|
---|
| 1475 |
|
---|
| 1476 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1477 | attribute.setValue(GSXML.PARAM_TYPE_INTEGER);
|
---|
| 1478 | param.setAttributeNode(attribute);
|
---|
| 1479 |
|
---|
| 1480 | displayItem = createNameValuePairElement(doc,
|
---|
| 1481 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME,
|
---|
| 1482 | "Maximum hits to return");
|
---|
| 1483 | param.appendChild(displayItem);
|
---|
| 1484 |
|
---|
| 1485 | paramList.appendChild(param);
|
---|
| 1486 |
|
---|
| 1487 | // Constructing the following:
|
---|
| 1488 | // <param name="query" type="string">
|
---|
| 1489 | // <displayItem name="name">Query string</displayItem>
|
---|
| 1490 | // </param>
|
---|
| 1491 | param = doc.createElement(GSXML.PARAM_ELEM);
|
---|
| 1492 |
|
---|
| 1493 | attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1494 | attribute.setValue(QUERY);
|
---|
| 1495 | param.setAttributeNode(attribute);
|
---|
| 1496 |
|
---|
| 1497 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1498 | attribute.setValue(GSXML.PARAM_TYPE_STRING);
|
---|
| 1499 | param.setAttributeNode(attribute);
|
---|
| 1500 |
|
---|
| 1501 | displayItem = createNameValuePairElement(doc,
|
---|
| 1502 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME,
|
---|
| 1503 | "Query string");
|
---|
| 1504 | param.appendChild(displayItem);
|
---|
| 1505 |
|
---|
| 1506 | paramList.appendChild(param);
|
---|
| 1507 |
|
---|
| 1508 | service.appendChild(paramList);
|
---|
| 1509 | }
|
---|
| 1510 |
|
---|
| 1511 | /** Appends children to the parameter service Element that make the
|
---|
| 1512 | * final service Element into a describe response XML for FedoraGS3's
|
---|
| 1513 | * FieldQuery service.
|
---|
| 1514 | * @param service is the service Element that is being filled out. */
|
---|
| 1515 | protected void describeFieldQueryService(Element service) {
|
---|
| 1516 | Document doc = service.getOwnerDocument();
|
---|
| 1517 | // we need name, submit (button) and description <displayItem> elements
|
---|
| 1518 | Element displayItem = createNameValuePairElement(doc,
|
---|
| 1519 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME,
|
---|
| 1520 | "Form Search");
|
---|
| 1521 | service.appendChild(displayItem);
|
---|
| 1522 |
|
---|
| 1523 | displayItem = createNameValuePairElement(doc,
|
---|
| 1524 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_SUBMIT, "Search");
|
---|
| 1525 | service.appendChild(displayItem);
|
---|
| 1526 |
|
---|
| 1527 | displayItem = createNameValuePairElement(doc,
|
---|
| 1528 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_DESCRIPTION,
|
---|
| 1529 | "Simple fielded search");
|
---|
| 1530 | service.appendChild(displayItem);
|
---|
| 1531 |
|
---|
| 1532 | //create the <paramList>
|
---|
| 1533 | Element paramList = doc.createElement(
|
---|
| 1534 | GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1535 |
|
---|
| 1536 | // we ignore granularity to search at: it will always be
|
---|
| 1537 | // document and section level
|
---|
| 1538 | // we ignore casefolding: always on (that is, case is irrelevant)
|
---|
| 1539 | // we ignore document display order: always ranked
|
---|
| 1540 |
|
---|
| 1541 | // Constructing the following:
|
---|
| 1542 | // <param default="100" name="maxDocs" type="integer">
|
---|
| 1543 | // <displayItem name="name">Maximum hits to return</displayItem>
|
---|
| 1544 | // </param>
|
---|
| 1545 | Element param = doc.createElement(GSXML.PARAM_ELEM);
|
---|
| 1546 |
|
---|
| 1547 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1548 | attribute.setValue(MAXDOCS);
|
---|
| 1549 | param.setAttributeNode(attribute);
|
---|
| 1550 |
|
---|
| 1551 | attribute = doc.createAttribute(GSXML.DEFAULT_ATT);
|
---|
| 1552 | attribute.setValue("100");
|
---|
| 1553 | param.setAttributeNode(attribute);
|
---|
| 1554 |
|
---|
| 1555 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1556 | attribute.setValue(GSXML.PARAM_TYPE_INTEGER);
|
---|
| 1557 | param.setAttributeNode(attribute);
|
---|
| 1558 |
|
---|
| 1559 | displayItem = createNameValuePairElement(doc,
|
---|
| 1560 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME,
|
---|
| 1561 | "Maximum hits to return");
|
---|
| 1562 | param.appendChild(displayItem);
|
---|
| 1563 |
|
---|
| 1564 | paramList.appendChild(param);
|
---|
| 1565 |
|
---|
| 1566 | // Constructing the following:
|
---|
| 1567 | // <param name="simpleField" occurs="4" type="multi">
|
---|
| 1568 | // <displayItem name="name"></displayItem>
|
---|
| 1569 | //
|
---|
| 1570 | // <param name="query" type="string">
|
---|
| 1571 | // <displayItem name="name">Word or phrase </displayItem>
|
---|
| 1572 | // </param>
|
---|
| 1573 | //
|
---|
| 1574 | // <param default="allFields" name="fieldname" type="enum_single">
|
---|
| 1575 | // <displayItem name="name">in field</displayItem>
|
---|
| 1576 | //
|
---|
| 1577 | // <option name="docTitles">
|
---|
| 1578 | // <displayItem name="name">document titles</displayItem>
|
---|
| 1579 | // </option>
|
---|
| 1580 | // <option name="allTitles">
|
---|
| 1581 | // <displayItem name="name">document and section titles</displayItem>
|
---|
| 1582 | // </option>
|
---|
| 1583 | // <option name="fullText">
|
---|
| 1584 | // <displayItem name="name">full text</displayItem>
|
---|
| 1585 | // </option>
|
---|
| 1586 | // <option name="all">
|
---|
| 1587 | // <displayItem name="name">titles and full text</displayItem>
|
---|
| 1588 | // </option>
|
---|
| 1589 | // <option name="">
|
---|
| 1590 | // <displayItem name="name"></displayItem>
|
---|
| 1591 | // </option>
|
---|
| 1592 | // </param>
|
---|
| 1593 | // </param>
|
---|
| 1594 | Element rowOfParams = doc.createElement(GSXML.PARAM_ELEM);
|
---|
| 1595 | attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1596 | attribute.setValue(SIMPLEFIELD_ATT);
|
---|
| 1597 | rowOfParams.setAttributeNode(attribute);
|
---|
| 1598 |
|
---|
| 1599 | // we want the row of controls to occur multiple times
|
---|
| 1600 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1601 | attribute.setValue(GSXML.PARAM_TYPE_MULTI);
|
---|
| 1602 | rowOfParams.setAttributeNode(attribute);
|
---|
| 1603 |
|
---|
| 1604 | attribute = doc.createAttribute(OCCURS_ATT);
|
---|
| 1605 | attribute.setValue("4"); // we want this row to occur 4 times
|
---|
| 1606 | rowOfParams.setAttributeNode(attribute);
|
---|
| 1607 |
|
---|
| 1608 | // <param name="query" type="string">
|
---|
| 1609 | // <displayItem name="name">Word or phrase </displayItem>
|
---|
| 1610 | // </param>
|
---|
| 1611 | param = doc.createElement(GSXML.PARAM_ELEM);
|
---|
| 1612 |
|
---|
| 1613 | attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1614 | attribute.setValue(QUERY);
|
---|
| 1615 | param.setAttributeNode(attribute);
|
---|
| 1616 |
|
---|
| 1617 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1618 | attribute.setValue(GSXML.PARAM_TYPE_STRING);
|
---|
| 1619 | param.setAttributeNode(attribute);
|
---|
| 1620 |
|
---|
| 1621 | displayItem = createNameValuePairElement(doc,
|
---|
| 1622 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME,
|
---|
| 1623 | "Word or phrase");
|
---|
| 1624 | param.appendChild(displayItem);
|
---|
| 1625 | rowOfParams.appendChild(param);
|
---|
| 1626 |
|
---|
| 1627 | // <param default="allFields" name="fieldName" type="enum_single">
|
---|
| 1628 | // <displayItem name="name">in field</displayItem>
|
---|
| 1629 | param = doc.createElement(GSXML.PARAM_ELEM);
|
---|
| 1630 | attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1631 | attribute.setValue(FIELDNAME_ATT);
|
---|
| 1632 | param.setAttributeNode(attribute);
|
---|
| 1633 |
|
---|
| 1634 | attribute = doc.createAttribute(GSXML.TYPE_ATT);
|
---|
| 1635 | attribute.setValue(GSXML.PARAM_TYPE_ENUM_SINGLE);
|
---|
| 1636 | param.setAttributeNode(attribute);
|
---|
| 1637 |
|
---|
| 1638 | attribute = doc.createAttribute(GSXML.DEFAULT_ATT);
|
---|
| 1639 | attribute.setValue(ALL_FIELDS);
|
---|
| 1640 | param.setAttributeNode(attribute);
|
---|
| 1641 |
|
---|
| 1642 | displayItem = createNameValuePairElement(doc,
|
---|
| 1643 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME,
|
---|
| 1644 | "in field");
|
---|
| 1645 | param.appendChild(displayItem);
|
---|
| 1646 |
|
---|
| 1647 | String[] searchFieldNames
|
---|
| 1648 | = {ALL_FIELDS, DOC_TITLES, ALL_TITLES, FULLTEXT};
|
---|
| 1649 | String[] searchFieldDisplay = {"all titles and full-text",
|
---|
| 1650 | "document titles only", "document and section titles",
|
---|
| 1651 | "full-text only"};
|
---|
| 1652 |
|
---|
| 1653 | // for each fieldName create an option element and insert
|
---|
| 1654 | // the option into the enum_multi drop-down param:
|
---|
| 1655 | // <option name="fieldName">
|
---|
| 1656 | // <displayItem name="name">fieldName</displayItem>
|
---|
| 1657 | // </option>
|
---|
| 1658 | for(int i = 0; i < searchFieldNames.length; i++) {
|
---|
| 1659 | Element option = doc.createElement(GSXML.PARAM_OPTION_ELEM);
|
---|
| 1660 | attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1661 | attribute.setValue(searchFieldNames[i]);
|
---|
| 1662 | option.setAttributeNode(attribute);
|
---|
| 1663 |
|
---|
| 1664 | displayItem = createNameValuePairElement(doc,
|
---|
| 1665 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME,
|
---|
| 1666 | searchFieldDisplay[i]);
|
---|
| 1667 | option.appendChild(displayItem);
|
---|
| 1668 | param.appendChild(option); // add option to the drop-down box
|
---|
| 1669 | }
|
---|
| 1670 |
|
---|
| 1671 | rowOfParams.appendChild(param);
|
---|
| 1672 | paramList.appendChild(rowOfParams);
|
---|
| 1673 | service.appendChild(paramList);
|
---|
| 1674 | }
|
---|
| 1675 |
|
---|
| 1676 | /**
|
---|
| 1677 | * @return a GS3 describe response message for the requested service
|
---|
| 1678 | * of the given collection. DocumentContent/Metadata/StructureRetrieve
|
---|
| 1679 | * return nothing special except their names; browse (and any query)
|
---|
| 1680 | * return more complex XML responses.
|
---|
| 1681 | * All collections in this Digital Library (Fedora Repository) share
|
---|
| 1682 | * the same services, so this method returns the same as
|
---|
| 1683 | * describeService(serviceName).
|
---|
| 1684 | * @param collectionName - the name of the collection whose service is to
|
---|
| 1685 | * be described. It will be converted to a fedora collection pid, which is of
|
---|
| 1686 | * the form "greenstone:<collectionName>-collection".
|
---|
| 1687 | * @param serviceName - the name of the service in the collection which is to
|
---|
| 1688 | * be described. */
|
---|
| 1689 | public String describeCollectionService(String collectionName,
|
---|
| 1690 | String serviceName) {
|
---|
| 1691 | // collectionName can be ignored, because all services are FedoraGS3
|
---|
| 1692 | // services and are not unique to any particular (greenstone) collection.
|
---|
| 1693 | return describeService(serviceName);
|
---|
| 1694 | }
|
---|
| 1695 |
|
---|
| 1696 | /** This method performs the implemented browse operation: allowing the
|
---|
| 1697 | * user to browse the titles of documents in the given collection by letter
|
---|
| 1698 | * and returning the results.
|
---|
[21835] | 1699 | * @param classifierIDs are the ids of the classifiers on which to browse. In
|
---|
[15222] | 1700 | * this case, the classifier indicates whether we browse titles by letter, or
|
---|
| 1701 | * browse (documents) by collection; and it is of the form <CL(letter)>.
|
---|
| 1702 | * @param collectionName is the name of the collection whose documents
|
---|
| 1703 | * starting with the given letter will be returned.
|
---|
| 1704 | * @return a GS3 DocumentStructureRetrieve response message which lists all
|
---|
| 1705 | * the documents that start with the letter indicated by parameter classifier.
|
---|
| 1706 | */
|
---|
[21835] | 1707 | public String browse(String collectionName, String[] classifierIDs)
|
---|
[15222] | 1708 | {
|
---|
| 1709 | Document doc = builder.newDocument();
|
---|
| 1710 | FedoraGS3RunException ex = null; //any RemoteException or UnsupportedEncodingException
|
---|
| 1711 |
|
---|
| 1712 | // <classifierNodeList>
|
---|
| 1713 | Element classifierNodeList = doc.createElement(
|
---|
| 1714 | GSXML.CLASS_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1715 |
|
---|
[21835] | 1716 | for(int i = 0; i < classifierIDs.length; i++) {
|
---|
[15222] | 1717 | // <classifierNode nodeID="classifierNum">
|
---|
| 1718 | Element requestedClassifierNode = doc.createElement(
|
---|
| 1719 | GSXML.CLASS_NODE_ELEM);
|
---|
| 1720 | Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
[21835] | 1721 | attribute.setValue(classifierIDs[i]);
|
---|
[15222] | 1722 | requestedClassifierNode.setAttributeNode(attribute);
|
---|
| 1723 | classifierNodeList.appendChild(requestedClassifierNode);
|
---|
| 1724 |
|
---|
| 1725 | // <nodeStructure>
|
---|
| 1726 | Element nodeStructure = doc.createElement(GSXML.NODE_STRUCTURE_ELEM);
|
---|
| 1727 | requestedClassifierNode.appendChild(nodeStructure);
|
---|
| 1728 |
|
---|
| 1729 | // And one more time, the top level classifierNode:
|
---|
| 1730 | Element classifierNode = doc.createElement(GSXML.CLASS_NODE_ELEM);
|
---|
| 1731 | attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
[21835] | 1732 | attribute.setValue(classifierIDs[i]);
|
---|
[15222] | 1733 | classifierNode.setAttributeNode(attribute);
|
---|
| 1734 | nodeStructure.appendChild(classifierNode);
|
---|
| 1735 |
|
---|
| 1736 | // Work out what we're browsing base on the classifierID's number
|
---|
| 1737 | // classifier CL1 = browse titles by letter;
|
---|
| 1738 | // classifier CL2 = browse by collection;
|
---|
| 1739 | //remove the CL prefix and obtain the number from the id:
|
---|
[21835] | 1740 | int classifierNum = Integer.parseInt(classifierIDs[i].replace("CL", ""));
|
---|
[15222] | 1741 | switch(classifierNum) {
|
---|
| 1742 | case 1:
|
---|
| 1743 | // we're going to loop to the end of the alphabet
|
---|
| 1744 | int num = 1;
|
---|
| 1745 | for(char ch = 'A'; ch <= 'Z'; ch++, num++) {
|
---|
[15670] | 1746 | // Retrieve the document structure for each subClassifierID:
|
---|
| 1747 | // all the documents that begin with its letter.
|
---|
| 1748 | String letter = String.valueOf(ch);
|
---|
[21835] | 1749 | try {
|
---|
[15670] | 1750 | String[] docPIDs = this.browseTitlesByLetter(
|
---|
| 1751 | collectionName, letter);
|
---|
| 1752 | if(docPIDs.length == 0) {
|
---|
| 1753 | continue; // skip letters that don't have any kids
|
---|
| 1754 | }
|
---|
| 1755 |
|
---|
[15222] | 1756 | // <classifierNode nodeID="CL3.1">
|
---|
| 1757 | Element subClassifier = doc.createElement(
|
---|
[15670] | 1758 | GSXML.CLASS_NODE_ELEM);
|
---|
[15222] | 1759 | attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
[21835] | 1760 | attribute.setValue(classifierIDs[i]+"."+num);
|
---|
[15222] | 1761 | subClassifier.setAttributeNode(attribute);
|
---|
| 1762 | classifierNode.appendChild(subClassifier);
|
---|
[15670] | 1763 |
|
---|
| 1764 | // append the <docNodes> for the docPIDs found as children
|
---|
| 1765 | // of subclassifier
|
---|
| 1766 | getStructureElement(subClassifier, docPIDs, DESCENDANTS);
|
---|
| 1767 | //CHILDREN); // for testing
|
---|
| 1768 | } catch(Exception e) {
|
---|
| 1769 | ex = new FedoraGS3RunException(e);
|
---|
| 1770 | ex.setSpecifics("requested portion of TOC file or "
|
---|
| 1771 | + "trouble with fielded search ");
|
---|
| 1772 | }
|
---|
[15222] | 1773 | }
|
---|
[15672] | 1774 | // No titles in this collection that start with a letter at all
|
---|
| 1775 | if(!classifierNode.hasChildNodes()) {
|
---|
| 1776 | // <classifierNode nodeID="CL1.0"> which we will equate with A-Z
|
---|
| 1777 | Element subClassifier = doc.createElement(
|
---|
| 1778 | GSXML.CLASS_NODE_ELEM);
|
---|
| 1779 | attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
[21835] | 1780 | attribute.setValue(classifierIDs[i]+"."+0); //
|
---|
[15672] | 1781 | subClassifier.setAttributeNode(attribute);
|
---|
| 1782 | classifierNode.appendChild(subClassifier);
|
---|
| 1783 | }
|
---|
[15222] | 1784 | break;
|
---|
| 1785 | case 2:
|
---|
| 1786 | break;
|
---|
| 1787 | default:
|
---|
| 1788 | ex = new FedoraGS3RunException( // cause is regular exception
|
---|
[21835] | 1789 | new Exception("Unknown classifier ID: " + classifierIDs[i]));
|
---|
[15222] | 1790 | }
|
---|
[21835] | 1791 | }
|
---|
[15222] | 1792 | Element responseMsg = createResponseMessage(doc, classifierNodeList, ex,
|
---|
| 1793 | GSXML.REQUEST_TYPE_DESCRIBE, collectionName+"/ClassifierBrowse");
|
---|
| 1794 | try{
|
---|
| 1795 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 1796 | }catch(TransformerException e) {
|
---|
| 1797 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 1798 | + " " + e;
|
---|
| 1799 | }
|
---|
| 1800 | }
|
---|
| 1801 |
|
---|
| 1802 | /** This method performs something equivalent to a greenstone3
|
---|
| 1803 | * ClassifierBrowseMetadataRetrieve on the classifierNodeIDs
|
---|
| 1804 | * @param classNodeIDs array of classifierNode IDs of for which the metadata
|
---|
| 1805 | * needs to be returned.
|
---|
| 1806 | * @return a GS3 ClassifierBrowseMetadataRetrieve response message which
|
---|
| 1807 | * lists the metadata for all the classifierNodes passed as parameter.*/
|
---|
| 1808 | public String browseMetadataRetrieve(String[] classNodeIDs)
|
---|
| 1809 | {
|
---|
| 1810 | Document doc = this.builder.newDocument();
|
---|
| 1811 | // <classifierNodeList>
|
---|
| 1812 | Element classifierNodeList = doc.createElement(
|
---|
| 1813 | GSXML.CLASS_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1814 |
|
---|
| 1815 | // create <classifierNode><metadataList><metadata>s
|
---|
| 1816 | // </metadataList></classifierNode> for all letters of alphabet
|
---|
| 1817 | for(int i = 0; i < classNodeIDs.length; i++) {
|
---|
| 1818 | // strip ID of everything before the first '.' (i.e. remove "CL#.")
|
---|
| 1819 | int index = classNodeIDs[i].indexOf('.');
|
---|
| 1820 | String subClassifierNumber = classNodeIDs[i].substring(index+1);
|
---|
| 1821 | int subClassifierNum = Integer.parseInt(subClassifierNumber);
|
---|
[15672] | 1822 | String classifierName = "";
|
---|
| 1823 | if(subClassifierNum == 0) { // no document titles started with a letter
|
---|
| 1824 | classifierName = "A-Z";
|
---|
| 1825 | } else {
|
---|
| 1826 | char letter = (char)('A' + subClassifierNum - 1); // A = 1
|
---|
| 1827 | classifierName = String.valueOf(letter);
|
---|
| 1828 | }
|
---|
[15222] | 1829 |
|
---|
| 1830 | // <classifierNode nodeID="CL#.subNum">
|
---|
| 1831 | Element classifierNode = doc.createElement(GSXML.CLASS_NODE_ELEM);
|
---|
| 1832 | Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
| 1833 | attribute.setValue(classNodeIDs[i]);
|
---|
| 1834 | classifierNode.setAttributeNode(attribute);
|
---|
| 1835 |
|
---|
| 1836 | // <metadataList>
|
---|
| 1837 | Element metadataList = doc.createElement(
|
---|
| 1838 | GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 1839 |
|
---|
| 1840 | // at least one metadata element: that of the title of this
|
---|
| 1841 | // classifierNode:
|
---|
| 1842 | // <metadata name="Title">letter</metadata>
|
---|
| 1843 | Element metadata = this.createNameValuePairElement(doc,
|
---|
[15672] | 1844 | GSXML.METADATA_ELEM, "Title", classifierName);
|
---|
[15222] | 1845 |
|
---|
| 1846 | // now connect up everything
|
---|
| 1847 | metadataList.appendChild(metadata);
|
---|
| 1848 | classifierNode.appendChild(metadataList);
|
---|
| 1849 | classifierNodeList.appendChild(classifierNode);
|
---|
| 1850 | }
|
---|
| 1851 |
|
---|
| 1852 | Element responseMsg = createResponseMessage(doc, classifierNodeList, null,
|
---|
| 1853 | GSXML.REQUEST_TYPE_PROCESS, //collName +
|
---|
| 1854 | "ClassifierBrowseMetadataRetrieve");
|
---|
| 1855 | try{
|
---|
| 1856 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 1857 | }catch(TransformerException e) {
|
---|
| 1858 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 1859 | + " " + e;
|
---|
| 1860 | }
|
---|
| 1861 | }
|
---|
| 1862 |
|
---|
| 1863 | /** @return a newly created element of the following format:
|
---|
| 1864 | * <classifier content="somecontent" name="CL+num">
|
---|
| 1865 | * <displayItem name="name">someClassifierName</displayItem>
|
---|
| 1866 | * <displayItem name="description">Browse by classifier name</displayItem>
|
---|
| 1867 | * </classifier>
|
---|
| 1868 | * @param doc - the document used to create the element
|
---|
| 1869 | * @param content - value of the content attribute
|
---|
| 1870 | * @param classifierNum - the number suffixed to the CL, together forming
|
---|
| 1871 | * the classifier Node's ID
|
---|
| 1872 | * @param displayNameVal is the bodytext of a named displayItem element
|
---|
| 1873 | * @param displayDescrVal is the bodytext of a displayItem element with
|
---|
| 1874 | * description */
|
---|
| 1875 | protected Element createClassifierElement(Document doc, String content,
|
---|
| 1876 | int classifierNum, String displayNameVal, String displayDescrVal)
|
---|
| 1877 | {
|
---|
| 1878 | final String CL = "CL";
|
---|
| 1879 | Element classifier = doc.createElement(GSXML.CLASSIFIER_ELEM);
|
---|
| 1880 | // content attribute
|
---|
| 1881 | Attr att = doc.createAttribute(GSXML.CLASSIFIER_CONTENT_ATT);
|
---|
| 1882 | att.setValue(content);
|
---|
| 1883 | classifier.setAttributeNode(att);
|
---|
| 1884 | // name attribute
|
---|
| 1885 | att = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1886 | att.setValue(CL + classifierNum);
|
---|
| 1887 | classifier.setAttributeNode(att);
|
---|
| 1888 |
|
---|
| 1889 | // now create the displayItem children for classifier:
|
---|
| 1890 | // <displayItem name="name">#letter</displayItem>
|
---|
| 1891 | // <displayItem name="description">
|
---|
| 1892 | //Browse titles starting with #letter</displayItem>
|
---|
| 1893 | Element displayItem = createNameValuePairElement(doc,
|
---|
| 1894 | GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, displayNameVal);
|
---|
| 1895 | classifier.appendChild(displayItem);
|
---|
| 1896 | displayItem = createNameValuePairElement(doc, GSXML.DISPLAY_TEXT_ELEM,
|
---|
| 1897 | GSXML.DISPLAY_TEXT_DESCRIPTION, displayDescrVal);
|
---|
| 1898 | classifier.appendChild(displayItem);
|
---|
| 1899 |
|
---|
| 1900 | return classifier;
|
---|
| 1901 | }
|
---|
| 1902 |
|
---|
| 1903 |
|
---|
| 1904 | /** @return a newly created element of the following format:
|
---|
| 1905 | * <elementName name="somename">"some display value"</elementName>
|
---|
| 1906 | * @param doc - the document used to create the element
|
---|
| 1907 | * @param elementName - the tag name
|
---|
| 1908 | * @param name - value of attribute name
|
---|
| 1909 | * @param value - the body text of the element */
|
---|
| 1910 | protected Element createNameValuePairElement(Document doc, String elementName,
|
---|
| 1911 | String name, String value) {
|
---|
| 1912 | // <elementName name="somename">"some display value"</elementName>
|
---|
| 1913 | Element element = doc.createElement(elementName);
|
---|
| 1914 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 1915 | attribute.setValue(name);
|
---|
| 1916 | element.setAttributeNode(attribute);
|
---|
| 1917 |
|
---|
| 1918 | element.appendChild(doc.createTextNode(value));
|
---|
| 1919 | return element;
|
---|
| 1920 | }
|
---|
| 1921 |
|
---|
| 1922 | /**
|
---|
| 1923 | * @param collection is the collection to search in
|
---|
| 1924 | * @param query is the query term to search for. It won't specify the
|
---|
| 1925 | * indexed field to search in, which will mean that GSearch will
|
---|
| 1926 | * search all default indexed fields.
|
---|
| 1927 | * @param maxDocs is the maximum number of results to return (which
|
---|
| 1928 | * at present we consider equivalent to FedoraGSearch's hitpageSize).
|
---|
| 1929 | */
|
---|
| 1930 | public String[] textQuery(String collection, String query,
|
---|
| 1931 | int maxDocs)
|
---|
| 1932 | throws Exception
|
---|
| 1933 | {
|
---|
| 1934 | // no need to search there is no query or query is empty spaces
|
---|
| 1935 | if(query.trim().equals(""))
|
---|
| 1936 | return new String[]{};
|
---|
| 1937 |
|
---|
| 1938 | // QUERY value won't specify indexed field to search, Fedora
|
---|
| 1939 | // Gsearch will take that as meaning all default indexed fields.
|
---|
| 1940 | // Params to search() method below: string of fielded query terms;
|
---|
| 1941 | // hitpageStart, hitpageEnd, snippetsMax (leave that 0)
|
---|
| 1942 | query = query + " " + "PID" + COLON + GREENSTONE;
|
---|
| 1943 |
|
---|
| 1944 | String searchResult = this.fedoraGSearch.search(query, 1, maxDocs, 0);
|
---|
| 1945 | // now we have the XML returned by FedoraGSearch, get the pids
|
---|
| 1946 | // of the documents returned (if any)
|
---|
| 1947 | String[] pids = this.fedoraGSearch.getPIDsFromSearchResult(
|
---|
| 1948 | collection, searchResult);
|
---|
| 1949 | return pids;
|
---|
| 1950 | }
|
---|
| 1951 |
|
---|
| 1952 | /**
|
---|
| 1953 | * This method performs a fieldquery, searching for x number of phrases
|
---|
| 1954 | * in each of the 4 indexed fields.
|
---|
| 1955 | * @param collection is the collection to search in
|
---|
| 1956 | * @param nameValParamsMap is a Map of several(key, value) entries,
|
---|
| 1957 | * 4 of which we're concerned with here:
|
---|
| 1958 | * - the keys are ALL_FIELDS, DOC_TITLES, ALL_TITLES, FULLTEXT
|
---|
| 1959 | * - the values are a comma separated list of terms (phrases or single
|
---|
| 1960 | * words) to search that field in. There may be more than 1 or
|
---|
| 1961 | * there may be none (in which case there may be N empty values or
|
---|
| 1962 | * spaces separated by commas).
|
---|
| 1963 | * @param maxDocs is the maximum number of results to return (which
|
---|
| 1964 | * at present we consider equivalent to FedoraGSearch's hitpageSize).
|
---|
| 1965 | * */
|
---|
| 1966 | public String[] fieldQuery(String collection, Map nameValParamsMap,
|
---|
| 1967 | int maxDocs)
|
---|
| 1968 | throws Exception
|
---|
| 1969 | {
|
---|
| 1970 | // we're going to maintain a list of UNIQUE pids that were returned
|
---|
| 1971 | // in search results. Hence we use Set:
|
---|
| 1972 | java.util.Set set = new java.util.HashSet();
|
---|
| 1973 |
|
---|
| 1974 | // (1) Use Fedora's search to search document titles, if they were
|
---|
| 1975 | // specified:
|
---|
| 1976 | String[] docTitlepids = {};
|
---|
| 1977 |
|
---|
| 1978 | String docTitleTerms = (String)nameValParamsMap.get(DOC_TITLES);
|
---|
| 1979 | if(docTitleTerms != null) { // no doc titles may have been specified
|
---|
| 1980 | String[] phrases = docTitleTerms.split(COMMA);
|
---|
| 1981 |
|
---|
| 1982 | // search the individual phrases first:
|
---|
| 1983 | for(int i = 0; i < phrases.length; i++) {
|
---|
| 1984 | if(phrases.equals("") || phrases.equals(" "))
|
---|
| 1985 | continue; //skip when there are no terms
|
---|
| 1986 | docTitlepids = this.searchDocumentTitles(
|
---|
| 1987 | collection, phrases[i], false);
|
---|
| 1988 | for(int j = 0; j < docTitlepids.length; j++)
|
---|
| 1989 | set.add(docTitlepids[j]);
|
---|
| 1990 | }
|
---|
| 1991 | }
|
---|
| 1992 | // (2) use FedoraGSearch to search doc AND section titles, and
|
---|
| 1993 | // fulltext (in case these were specified in nameValParamsMap):
|
---|
| 1994 | String searchResult = this.fedoraGSearch.search(
|
---|
| 1995 | nameValParamsMap, 1, maxDocs);
|
---|
| 1996 |
|
---|
| 1997 | String[] pids = this.fedoraGSearch.getPIDsFromSearchResult(
|
---|
| 1998 | collection, searchResult);
|
---|
| 1999 |
|
---|
| 2000 | for(int i = 0; i < pids.length; i++)
|
---|
| 2001 | set.add(pids[i]);
|
---|
| 2002 |
|
---|
| 2003 | pids = null;
|
---|
| 2004 | pids = new String[set.size()];
|
---|
| 2005 | set.toArray(pids); // unique pids
|
---|
| 2006 | return pids;
|
---|
| 2007 | }
|
---|
| 2008 |
|
---|
| 2009 | /** @return a String representing Greenstone3 XML for a query process
|
---|
| 2010 | * response returning the results for the query denoted by parameter
|
---|
| 2011 | * nameValParamsMap.
|
---|
| 2012 | * @param nameValParamsMap is a Hashmap of name and value pairs for all the
|
---|
| 2013 | * query field data values. The names match the field names that
|
---|
| 2014 | * describeCollectionService() would have returned for the query service.
|
---|
| 2015 | * @param collection is the name of the collection
|
---|
| 2016 | * @param service is the name of the query service
|
---|
| 2017 | * This method is only ever called when any of the services in the digital
|
---|
| 2018 | * library described themselves as type=query. Therefore any digital
|
---|
| 2019 | * libraries that have no query services, can just return emtpy message
|
---|
| 2020 | * strings (or even "") since this method will never be called on them
|
---|
| 2021 | * anyway. */
|
---|
| 2022 | public String query(String collection, String service,
|
---|
| 2023 | Map nameValParamsMap)
|
---|
| 2024 | {
|
---|
| 2025 | FedoraGS3RunException ex = null;
|
---|
| 2026 | // (1) obtain the requested number of maximum result documents
|
---|
| 2027 | int maxDocs = 100;
|
---|
| 2028 | try{
|
---|
| 2029 | maxDocs = Integer.parseInt((String)nameValParamsMap.get(MAXDOCS));
|
---|
| 2030 | } catch(NumberFormatException e) {
|
---|
| 2031 | maxDocs = 100;
|
---|
| 2032 | }
|
---|
| 2033 |
|
---|
| 2034 | String pids[] = {};
|
---|
| 2035 | // (2) for Textquery, we simply search ALL_FIELDS using FedoraGSearch
|
---|
| 2036 | if(service.equals("TextQuery")) {
|
---|
| 2037 | try {
|
---|
| 2038 | // get the Query field:
|
---|
| 2039 | String query = (String)nameValParamsMap.get(QUERY);
|
---|
| 2040 | pids = textQuery(collection, query, maxDocs);
|
---|
| 2041 | }
|
---|
| 2042 | catch(Exception e) {
|
---|
| 2043 | LOG.error("Error in TextQuery processing: " + e);
|
---|
| 2044 | ex = new FedoraGS3RunException(
|
---|
| 2045 | "When trying to use FedoraGenericSearch for a TextQuery", e);
|
---|
| 2046 |
|
---|
| 2047 | }
|
---|
| 2048 | } else { // (3) FieldQuery
|
---|
| 2049 | // first get the comma-separated lists
|
---|
| 2050 | String listOfFieldNames = (String)nameValParamsMap.get(FIELDNAME_ATT);
|
---|
| 2051 | String listOfSearchTerms = (String)nameValParamsMap.get(QUERY);
|
---|
| 2052 | // both are comma separated lists, so split both on 'comma'
|
---|
| 2053 | String[] fieldNames = listOfFieldNames.split(COMMA);
|
---|
| 2054 | String[] searchTerms = listOfSearchTerms.split(COMMA);
|
---|
| 2055 |
|
---|
| 2056 | // In the fieldNames and searchTerms lists of nameValParamsMap,
|
---|
| 2057 | // each searchTerm element was matched with its correspondingly
|
---|
| 2058 | // indexed fieldName.
|
---|
| 2059 | // A new map is going to reorganise this, by putting all terms
|
---|
| 2060 | // for a particular fieldName together in a comma separated list
|
---|
| 2061 | // and associating that with the fieldName. I.e. (key, value) ->
|
---|
| 2062 | // (fieldName, comma-separated list of all terms in that field)
|
---|
| 2063 | Map map = new HashMap();
|
---|
| 2064 | for(int i = 0; i < searchTerms.length; i++) {
|
---|
| 2065 | // there may be fewer searchTerms than fieldNames (since some
|
---|
| 2066 | // fieldNames may have been left empty), so loop on searchTerms
|
---|
[21775] | 2067 | if(map.containsKey(fieldNames[i])) { // fieldName is already
|
---|
[15222] | 2068 | // in the list, so append comma with new value
|
---|
| 2069 | String termsList = (String)map.get(fieldNames[i]);
|
---|
| 2070 | termsList = termsList + COMMA + searchTerms[i];
|
---|
| 2071 | map.put(fieldNames[i], termsList);
|
---|
| 2072 | } else { // this is the first time this fieldName occurred
|
---|
| 2073 | // just put the fieldName with searchTerm as-is
|
---|
| 2074 | map.put(fieldNames[i], searchTerms[i]);
|
---|
| 2075 | }
|
---|
| 2076 | }
|
---|
| 2077 |
|
---|
| 2078 | try {
|
---|
| 2079 | // For fieldquery, we search on all the fieldNames specified
|
---|
| 2080 | // - if DOC_TITLES is specified then we use Fedora's search
|
---|
| 2081 | // - for all other fieldNames specified, we use FedoraGSearch
|
---|
| 2082 | pids = fieldQuery(collection, map, maxDocs);
|
---|
| 2083 | }
|
---|
| 2084 | catch(Exception e) {
|
---|
| 2085 | LOG.error("Error in FieldQuery processing: " + e);
|
---|
| 2086 | ex = new FedoraGS3RunException(
|
---|
| 2087 | "When trying to use FedoraGenericSearch for a FieldQuery", e);
|
---|
| 2088 | }
|
---|
| 2089 | }
|
---|
| 2090 |
|
---|
| 2091 | // Build Greenstone XML Query response message for from
|
---|
| 2092 | // the pids (which should be document identifiers)
|
---|
| 2093 | Document doc = builder.newDocument();
|
---|
| 2094 | // <metadataList><metadata name="numDocsMatched" value="n" />
|
---|
| 2095 | // </metadataList>
|
---|
| 2096 | Element metadataList = doc.createElement(
|
---|
| 2097 | GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 2098 | Element metadata = doc.createElement(GSXML.METADATA_ELEM);
|
---|
| 2099 |
|
---|
| 2100 | Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
|
---|
| 2101 | attribute.setValue(NUM_DOCS_MATCHED);
|
---|
| 2102 | metadata.setAttributeNode(attribute);
|
---|
| 2103 |
|
---|
| 2104 | attribute = doc.createAttribute(GSXML.VALUE_ATT);
|
---|
| 2105 | attribute.setValue(Integer.toString(pids.length));
|
---|
| 2106 | metadata.setAttributeNode(attribute);
|
---|
| 2107 |
|
---|
| 2108 | metadataList.appendChild(metadata);
|
---|
| 2109 |
|
---|
| 2110 | // <documentNodeList>
|
---|
| 2111 | // <documentNode nodeID="HASHac0a04dd14571c60d7fbfd.4.2"
|
---|
| 2112 | // docType='hierarchy' nodeType="leaf" />
|
---|
| 2113 | // ...
|
---|
| 2114 | // ...
|
---|
| 2115 | // </documentNodeList>
|
---|
| 2116 | Element docNodeList = doc.createElement(
|
---|
| 2117 | GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 2118 | // for each
|
---|
| 2119 | for(int i = 0; i < pids.length; i++) {
|
---|
| 2120 | Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
| 2121 | attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
|
---|
| 2122 | attribute.setValue(pids[i]);
|
---|
| 2123 | docNode.setAttributeNode(attribute);
|
---|
| 2124 |
|
---|
| 2125 | attribute = doc.createAttribute(GSXML.DOC_TYPE_ATT);
|
---|
| 2126 | attribute.setValue("hierarchy");
|
---|
| 2127 | docNode.setAttributeNode(attribute);
|
---|
| 2128 |
|
---|
| 2129 | attribute = doc.createAttribute(GSXML.NODE_TYPE_ATT);
|
---|
| 2130 | attribute.setValue("root");
|
---|
| 2131 | docNode.setAttributeNode(attribute);
|
---|
| 2132 | docNodeList.appendChild(docNode);
|
---|
| 2133 | }
|
---|
| 2134 |
|
---|
| 2135 | Element responseMsg = createResponseMessage(doc, docNodeList, ex,
|
---|
| 2136 | GSXML.REQUEST_TYPE_PROCESS, service);
|
---|
| 2137 | try{
|
---|
| 2138 | return FedoraCommons.elementToFormattedString(responseMsg);
|
---|
| 2139 | }catch(TransformerException e) {
|
---|
| 2140 | return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
|
---|
| 2141 | + " " + e;
|
---|
| 2142 | }
|
---|
| 2143 | }
|
---|
| 2144 |
|
---|
| 2145 | public static void main(String args[]) {
|
---|
| 2146 | try{
|
---|
| 2147 | // testing default constructor
|
---|
| 2148 | //FedoraGS3Connection con = new FedoraGS3Connection();
|
---|
| 2149 |
|
---|
| 2150 | // testing constructor that takes properties file to show initial
|
---|
| 2151 | // fedora server values
|
---|
| 2152 | java.io.File propertyFilename
|
---|
| 2153 | = new java.io.File("fedoraGS3.properties");
|
---|
| 2154 | FedoraGS3Connection con = new FedoraGS3Connection(propertyFilename);
|
---|
| 2155 |
|
---|
| 2156 | // DESCRIBE: serviceList, collectionList
|
---|
| 2157 | System.out.println("serviceList:\n" + con.getServiceList());
|
---|
| 2158 |
|
---|
| 2159 | System.out.println("collectionList:\n" + con.getCollectionList());
|
---|
| 2160 |
|
---|
| 2161 | String[] colPIDs = con.getCollections();
|
---|
| 2162 | String[] collectionNames = con.getCollectionNames(con.getCollections());
|
---|
| 2163 |
|
---|
| 2164 |
|
---|
| 2165 | for(int i = 0; i < collectionNames.length; i++) {
|
---|
| 2166 | System.out.println("Describing collections:\n");
|
---|
| 2167 | System.out.println(con.describeCollection(collectionNames[i]));
|
---|
| 2168 | System.out.println("Describing collection services:\n"
|
---|
| 2169 | + con.describeCollectionServices(collectionNames[i]));
|
---|
| 2170 | }
|
---|
| 2171 |
|
---|
| 2172 | String[] serviceNames = con.getServiceNames();
|
---|
| 2173 | for(int i = 0; i < serviceNames.length; i++) {
|
---|
| 2174 | System.out.println("Describing " + serviceNames[i] + ":\n"
|
---|
| 2175 | + con.describeCollectionService("demo", serviceNames[i]));
|
---|
| 2176 | }
|
---|
| 2177 |
|
---|
| 2178 |
|
---|
| 2179 | // TRYING OUT SPECIAL top-level document metadata retrieval (DLS, DC)
|
---|
| 2180 | // along with EX of the top-level document:
|
---|
| 2181 | System.out.println("\nGET META for greenstone:gs2mgdemo-HASH01d667303fe98545f03c14ae:");
|
---|
| 2182 | System.out.println(con.getDocumentMetadata(new String[]{"greenstone:gs2mgdemo-HASH01d667303fe98545f03c14ae"}));
|
---|
| 2183 |
|
---|
| 2184 |
|
---|
| 2185 | String[] docIDs = con.getCollectionDocs(colPIDs[0]);
|
---|
| 2186 | System.out.println("\nGET CONTENT:");
|
---|
| 2187 | for(int i = 0; i < docIDs.length; i++) {
|
---|
| 2188 | System.out.println(con.getContent(docIDs[i]));
|
---|
| 2189 | }
|
---|
| 2190 |
|
---|
| 2191 | System.out.println("\nGET META:");
|
---|
| 2192 | for(int i = 0; i < docIDs.length; i++) {
|
---|
| 2193 | System.out.println(con.getDocumentMetadata(docIDs[i]));
|
---|
| 2194 | }
|
---|
| 2195 |
|
---|
| 2196 | String[] getTitlesFor = {
|
---|
| 2197 | "greenstone:gs2mgdemo-HASH01d667303fe98545f03c14ae",
|
---|
| 2198 | "greenstone:gs2mgdemo-HASHa568bac1d8d7bd12a0938b",
|
---|
| 2199 | "greenstone:gs2mgdemo-HASHa568bac1d8d7bd12a0938b-1",
|
---|
| 2200 | "greenstone:gs2mgdemo-HASH01d667303fe98545f03c14ae-1.7",
|
---|
| 2201 | "greenstone:gs2mgdemo-HASHa568bac1d8d7bd12a0938b-1.5.1"
|
---|
| 2202 | };
|
---|
| 2203 |
|
---|
| 2204 | // first let's display the regular meta for top-level docs and
|
---|
| 2205 | // their sections
|
---|
| 2206 | for(int i = 0; i < getTitlesFor.length; i++) {
|
---|
| 2207 | System.out.println(con.getDocumentMetadata(getTitlesFor[i]));
|
---|
| 2208 | }
|
---|
| 2209 |
|
---|
| 2210 | System.out.println("\nTitles are:");
|
---|
| 2211 | System.out.println(con.getTitleMetadata(getTitlesFor));
|
---|
| 2212 |
|
---|
| 2213 | System.out.println("\nGET STRUCTURE:");
|
---|
| 2214 | for(int i = 0; i < docIDs.length; i++) {
|
---|
| 2215 | System.out.println(con.getChildren(docIDs[i]));
|
---|
| 2216 | System.out.println(con.getDocumentStructure(docIDs[i]));
|
---|
| 2217 | }
|
---|
| 2218 |
|
---|
| 2219 | // TEST ERROR CASES:
|
---|
| 2220 | System.out.println("\nTESTING ERROR CASES");
|
---|
| 2221 | System.out.println(con.getContent("greenstone:demo-pinky"));
|
---|
| 2222 | String[] errorCases = { "greenstone:demo-HASH23d1019b589e2ef6a680e3-1.5.1.5",
|
---|
| 2223 | "greenstone:demo-pinky" };
|
---|
| 2224 | System.out.println(con.getContent(errorCases));
|
---|
| 2225 | System.out.println(con.getDocumentMetadata(errorCases));
|
---|
| 2226 | System.out.println(con.getDocumentStructure(errorCases));
|
---|
| 2227 |
|
---|
| 2228 | System.out.println("\nCLASSIFIER BROWSE");
|
---|
| 2229 | System.out.println(con.browse("gs2mgdemo", //"ClassifierBrowse",
|
---|
[21835] | 2230 | new String[]{"CL1"}));
|
---|
[15222] | 2231 |
|
---|
| 2232 | System.out.println("\nCLASSIFIER BROWSE METADATA RETRIEVE");
|
---|
| 2233 | String[] classNodeIDs = new String[26];
|
---|
| 2234 | for(int i = 0; i < classNodeIDs.length; i++) {
|
---|
| 2235 | int subClassifierNum = i + 1;
|
---|
| 2236 | classNodeIDs[i] = "CL1." + subClassifierNum;
|
---|
| 2237 | }
|
---|
| 2238 | System.out.println(con.browseMetadataRetrieve(//"gs2mgdemo",
|
---|
| 2239 | classNodeIDs));
|
---|
| 2240 |
|
---|
| 2241 | System.out.println("Testing query services");
|
---|
| 2242 | System.out.println("TEXT QUERY:");
|
---|
| 2243 | Map formControlValsMap = new HashMap();
|
---|
| 2244 | formControlValsMap.put(MAXDOCS, "100");
|
---|
| 2245 | formControlValsMap.put(QUERY, "snails");
|
---|
| 2246 | String searchResponse
|
---|
| 2247 | = con.query("gs2mgdemo", "TextQuery", formControlValsMap);
|
---|
| 2248 | System.out.println(searchResponse);
|
---|
| 2249 |
|
---|
| 2250 | System.out.println("FIELD QUERY:");
|
---|
| 2251 | formControlValsMap.clear();
|
---|
| 2252 | formControlValsMap.put(MAXDOCS, "100");
|
---|
| 2253 | formControlValsMap.put(QUERY, "interview,Gender equality,cyclone");
|
---|
| 2254 | formControlValsMap.put(FIELDNAME_ATT,
|
---|
| 2255 | "allFields,docTitles,allFields,allFields");
|
---|
| 2256 | searchResponse
|
---|
| 2257 | = con.query("gs2mgdemo", "FieldQuery", formControlValsMap);
|
---|
| 2258 | System.out.println(searchResponse);
|
---|
| 2259 |
|
---|
| 2260 | System.exit(0);
|
---|
| 2261 | }catch(Exception e) {
|
---|
| 2262 | JOptionPane.showMessageDialog(
|
---|
| 2263 | null, e, "Error", JOptionPane.ERROR_MESSAGE);
|
---|
| 2264 | //System.err.println("ERROR: " + e);
|
---|
| 2265 | e.printStackTrace();
|
---|
| 2266 | }
|
---|
| 2267 | }
|
---|
[15733] | 2268 | }
|
---|