source: other-projects/gs3-webservices-java-client/trunk/src/GS3Fedora/org/greenstone/fedora/services/FedoraGS3Connection.java@ 21835

/**
 *#########################################################################
 * FedoraGS3Connection.java - works with the demo-client for Greenstone 3, 
 * of the Greenstone digital library suite from the New Zealand Digital 
 * Library Project at the  * University of Waikato, New Zealand.
 * <BR><BR>
 * Copyright (C) 2008 New Zealand Digital Library Project
 * <BR><BR>
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 * <BR><BR>
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *########################################################################
 */

package org.greenstone.fedora.services;


import java.io.StringReader;

import org.apache.log4j.Logger;
import org.greenstone.fedora.services.FedoraGS3Exception.AuthenticationFailedException;
import org.greenstone.fedora.services.FedoraGS3Exception.CancelledException;
import org.greenstone.fedora.services.FedoraGS3Exception.FedoraGS3RunException;
import org.greenstone.fedora.services.FedoraGS3Exception.NotAFedoraServerException;
import org.greenstone.gsdl3.util.GSXML;
import org.w3c.dom.Document;
import org.w3c.dom.Element;
import org.w3c.dom.Attr;
import org.w3c.dom.Text;
import org.w3c.dom.NodeList;
import org.w3c.dom.Node;
import org.xml.sax.InputSource;

import java.io.File; 
import java.util.HashMap;
import java.util.Properties;
import java.util.Map;

import javax.swing.JOptionPane;

import org.xml.sax.SAXException;
import java.io.UnsupportedEncodingException;
import java.io.IOException;
import javax.net.ssl.SSLHandshakeException;
import java.net.ConnectException;
import java.net.MalformedURLException;
import java.rmi.RemoteException;
import javax.xml.parsers.ParserConfigurationException;
import javax.xml.transform.TransformerException;

/**
 * Class that extends FedoraConnection in order to be able to use
 * Fedora's web services to retrieve the specific datastreams of 
 * Greenstone documents stored in Fedora's repository. This class
 * provides methods that convert those datastreams into Greenstone3 
 * XML response messages which are returned.
 * @author ak19
*/
public class FedoraGS3Connection 
    extends FedoraConnection implements FedoraToGS3Interface, 
        FedoraToGS3Interface.Constants  
{
    /** The logging instance for this class */
    private static final Logger LOG = Logger.getLogger(
            FedoraGS3Connection.class.getName());
        
    /** Complete list of services that are supported our FedoraGS3 would  
     * support if everything goes well. If a connection to FedoraGSearch
     * cannot be established, the query services will no longer be
     * available. The actual services supported are given by member
     * variable serviceNames. */
    protected static final String[] SERVICES = {
        "DocumentContentRetrieve", "DocumentMetadataRetrieve",
        "DocumentStructureRetrieve", 
        "TextQuery", "FieldQuery",
        "ClassifierBrowse", "ClassifierBrowseMetadataRetrieve" 
    };
    
    /** List of services actually supported by our FedoraGS3 repository 
     * after construction. If FedoraGenericSearch can't be connected to,
     * then query services will not be offered */
    protected String[] serviceNames;
        
    /** constant CHILDREN indicates that a DocumentStructureRetrieve is to
     * return only the child nodes of a section, not any further descendants */
    protected static final int CHILDREN = 0;
    
    /** constant DESCENDANTS indicates that a DocumentStructureRetrieve is to
     * return all descendants of a section */
    protected static final int DESCENDANTS = 1;
    
    /** The object used to connect to FedoraGenericSearch, which is used 
     * for doing full-text searching */
    protected GSearchConnection fedoraGSearch; 

    /** The url for the wsdl file of FedoraGSearch's web services
     * by default this will be the Fedora server's base URL 
     * concatenated to "gsearch/services/FgsOperations?wsdl" */
    protected String gSearchWSDLURL;
        
    /** The last part of the gSearchWSDL URL. The first part is
     * the same as the fedora server's base url. */
        protected String gSearchWSDLSuffix;
            
    /** The name of the index that FedoraGSearch will index the GS3
     * documents into. If no name is specified in the properties file,
     * this will default to FedoraIndex. */
    protected String gSearchIndexName;
    
    /** 5 argument constructor is the same as that of superclass FedoraConnection:
     * @param protocol can be either http or https
     * @param host is the host where the fedora server is listening
     * @param port is the port where the fedora server is listening
     * @param fedoraServerUsername is the username for administrative 
     * authentication required to access the fedora server. 
     * @param fedoraServerPassword is the password for administrative 
     * authentication required to access the fedora server. If no password was set 
     * when installing Fedora, leave the field "".
     * Instantiates a FedoraGS3Connection object which connects to Fedora's
     * web services through stub classes and tries to connect to FedoraGSearch's
     * web services through the default WSDL location for it 
     * ("gsearch/services/FgsOperations?wsdl"). If another url is to be used,
     * call setGSearchWSDLURL(url) after the constructor instead.
    */
    public FedoraGS3Connection(String protocol, String host, int port, 
            String fedoraServerUsername, String fedoraServerPassword)
        throws ParserConfigurationException, MalformedURLException, 
            SSLHandshakeException, RemoteException, AuthenticationFailedException,   
                NotAFedoraServerException, ConnectException, Exception  
    { 
        super(protocol, host, port, fedoraServerUsername, fedoraServerPassword);
        // super() will call setInitialisationProperties(properties)
        // And that will try to instantiate the GSearchConnection.
    }
    
    /** No-argument constructor which is the same as that of superclass 
     * FedoraConnection: it displays a small dialog requesting input for the 
     * host, port, administrative password and username of the fedora server. 
     * If no password was set on the fedora repository when installing it, 
     * the user can leave the password field blank. */
    public FedoraGS3Connection() 
        throws ParserConfigurationException, MalformedURLException, 
            CancelledException, ConnectException, RemoteException, 
            SSLHandshakeException, Exception
    { 
        super();
        // super() will call setInitialisationProperties(properties)
        // And that will try to instantiate the GSearchConnection.
    }
    
    /** Single-argument constructor which is the same as that of superclass 
     * FedoraConnection: it takes the name of the properties file where
     * connection initialisation values may already be provided and then
     * displays a small dialog requesting input for the host, port, 
     * administrative password and username of the fedora server showing 
     * the values in the properties file as default. If the necessary 
     * initialisation are not present in the file, the corresponding fields
     * in the dialog will be blank. 
     * If no password was set on the fedora repository when installing it, 
     * the user can leave the password field blank. */
    public FedoraGS3Connection(File propertiesFilename) 
        throws ParserConfigurationException, MalformedURLException, 
            CancelledException, ConnectException, RemoteException, 
            SSLHandshakeException, Exception
    { 
        super(propertiesFilename);
        // super() will call setInitialisationProperties(properties)
        // And that will try to instantiate the GSearchConnection. 
    }
    
    /** The superclass constructor calls this method passing any preset  
     * properties loaded from a propertiesFile. This method is overridden  
     * here in order to instantiate the gSearchConnection based on the  
     * - gSearchWSDLSuffix that will be appended to the fedora base url.
     * (If one was not provided in the properties file, gSearchWSDLURL defaults 
     * to something of the form
     * "http://&lt;fedorahost:port&gt;/fedoragsearch/services/FgsOperations?wsdl"
     * which is the baseURL of fedora concatenated to the default gSearchWSDLSuffix,
     * "gsearch/services/FgsOperations?wsdl".
     * - name of the index into which the GS3 documents have been indexed 
     * and which FedoraGenericSearch should use to perform searches. If none is
     * given in the properties file, then the index name defaults to "FedoraIndex".
     * @param properties is the Properties Map loaded from a properties file
     * (if there was any) which specifies such things as host and port of the
     * FedoraServer, but can also specify the property "gsearch.wsdlURL.suffix".
     * At the end of this method, properties' "gsearch.wsdlURL.suffix" will be set
     * to whatever the final value of this.gSearchWSDLURL' suffix is, and 
     * "gsearch.indexName" will be set to to whatever the final value of
     * this.gSearchIndexName is. 
    */
    protected void setInitialisationProperties(Properties properties)
        throws ParserConfigurationException, MalformedURLException, 
        CancelledException, ConnectException, RemoteException, 
        SSLHandshakeException, Exception
    {
        super.setInitialisationProperties(properties);
        // gsearchWSDL URL suffix, if not specified, defaults to 
        // "fedoragsearch/services/FgsOperations?wsdl" which is 
        // concatenated to the baseURL of fedora to give the gsearchWSDLURL.
        this.gSearchWSDLSuffix = properties.getProperty(
            "gsearch.wsdlURL.suffix", "gsearch/services/FgsOperations?wsdl");
        this.gSearchWSDLURL = this.baseURL + this.gSearchWSDLSuffix;
        // Set the property to whatever this.gSearchWSDLURL is now,
        // so that it will be written out to the properties file again
        properties.setProperty("gsearch.wsdlURL.suffix", this.gSearchWSDLSuffix);
        
        // Similarly for the name of the index FedoraGenericSearch should use
        // when performing searches for GS3 docs stored in Fedora's repository.
        this.gSearchIndexName = properties.getProperty(
                "gsearch.indexName", "FedoraIndex"); // default to FedoraIndex
        properties.setProperty("gsearch.indexName", this.gSearchIndexName);
        // Create a connection to FedoraGSearch's web services:
        initSearchFunctionality();
    }

    /** Init method that instantiates a GSearchConnection object used
     * to work with the separate FedoraGSearch web services. 
     * The url of the WSDL for FedoraGSearch's web services is worked out 
     * from the baseURL of the Fedora server.
    */
    protected void initSearchFunctionality() 
    {
        try {
            this.fedoraGSearch = null;
            this.fedoraGSearch = new GSearchConnection(
                    gSearchWSDLURL, gSearchIndexName);
            this.serviceNames = SERVICES;
        } catch(Exception e){
            LOG.error("Cannot connect to FedoraGSearch's web services at "
                + gSearchWSDLURL + "\nQuery services will not be available.");
            // If an exception occurs, something has gone wrong when
            // trying to connect to FedoraGSearch's web services. This
            // means, we can't offer query services, as that's provided
            // by FedoraGSearch
            serviceNames = null;
            int countOfNonQueryServices = 0;
            for(int i = 0; i < SERVICES.length; i++) {
                // do not count query services
                if(!SERVICES[i].toLowerCase().contains("query")) {
                    countOfNonQueryServices++; 
                }
            }
            // Services now supported are everything except Query services
            serviceNames = new String[countOfNonQueryServices];
            int j = 0;
            for(int i = 0; i < SERVICES.length; i++) {
                if(!SERVICES[i].toLowerCase().contains("query")) {
                    serviceNames[j] = SERVICES[i];
                    j++; // valid serviceName, so increment serviceName counter
                }
                 
            }
        }
    }
    
    /** @return the gSearchWSDLURL, the url of the WSDL for the
     * FedoraGSearch web services */
    public String getGSearchWSDLURL() { return gSearchWSDLURL; }
    
    /** Sets the member variable gSearchWSDLURL that specify the location of
     * the WSDL file of FedoraGSearch's web services. Then it attempts 
     * to instantiate a connection to those web services. 
     * @param url is the new url of the GSearch web services WSDL file */
    public void setGSearchWSDLURL(String url) { 
        this.gSearchWSDLURL = url;
        initSearchFunctionality();
    }
    
    /** @return the gSearchIndexName, the name of the index Fedora Generic
     * Search will search in (where GS3 docs have been indexed into). */
    public String getGSearchIndexName() { return gSearchIndexName; }
    
    /** Sets the member variable gSearchIndexName that specifies the name
     * of the index containing indexed GS3 documents. Then it attempts 
     * to instantiate a connection to the Fedora GSearch web services using
     * this changed value for indexName. 
     * @param indexName is the new name of the index containing indexed GS3 
     * docs that GSearch should search in. */
    public void setGSearchIndexName(String indexName) { 
        this.gSearchIndexName = indexName;
        initSearchFunctionality();
    }
    
    /** @return the array of the services actually supported by FedoraGS3 */
    protected String[] getServiceNames() { return this.serviceNames;}
    
    /** 
     * For finding out if the sectionNumber is given as part of the docID.
     * @param docID is the String that contains the docPID and may also
     * contain the section number. 
     * @return true if the document identifier docID contains a section-
     * number, and false if it consists solely of the docPID.
     * That is, true is returned if 
     * <pre>docID = "greenstone:colName-&lt;docPID&gt;-&lt;sectionNum&gt;"</pre>
     * and false is returned if
     * <pre>docID = "greenstone:colName-&lt;docPID&gt;"</pre>
     * */
    protected boolean containsSectionNumber(String docID) {
        // if there are two hyphens in the docID, then there are sections 
        // (and the section number is appended at end of docID)
        // docID = "greenstone:colName-<docPID>-<sectionNum>"
        return (docID.lastIndexOf(HYPHEN) != docID.indexOf(HYPHEN));
    }
    
    /** This method will extract the docPID from docID and return it.
     * (If a sectionNumber is suffixed to the docID, the docPID which is
     * the prefix is returned; otherwise the docID is the docPID and is
     * returned)
     * @param docID is the String that contains the docPID and may also
     * contain the section number.
     * @return only the docPID portion of the docID.
    */
    protected String getDocPIDFromDocID(String docID) {
        if(containsSectionNumber(docID))
            return docID.substring(0, docID.lastIndexOf(HYPHEN));
        // else (if there's no sectionNumber), docID is the docPID
        return docID; 
    }
    
    /** This method will return the section Number, if there's any
     * suffixed to the docID. Otherwise it will return the empty string
     * @param docID is the String that contains the docPID and may also
     * contain the section number.
     * @return only the sectionID portion of the docID - if any, else "".
    */
    protected String getSectionIDFromDocID(String docID) {
        if(containsSectionNumber(docID))
            return docID.substring(
                    docID.lastIndexOf(HYPHEN)+1, docID.length());
        return "";
    }
    
    /** Given a list of collectionIDs, returns a GS3 DocumentMetadataRetrieve 
     * response message that gives the metadata for each collection identified
     * @param collIDs is an array of fedora pids identifying collections in the
     * fedora repository
     * @return a GS3 DocumentMetadataRetrieve response message containing the
     * EX metadata for all the requested collections */
    public String getCollectionMetadata(String[] collIDs) {
        return getMetadata(collIDs);
    }
    
    /** Given a list of document identifiers, a GS3 DocumentMetadataRetrieve 
     * response message is returned containing the metadata for each document. 
     * @param docIDs is an array of document identifiers (docID can either be 
     * &lt;pid&gt;s items (documents) in the fedora repository, or 
     * "&lt;pid&gt;-sectionNumber". 
     * @return a GS3 DocumentMetadataRetrieve response message containing the
     * EX, DC, DLS metadata for all the requested documents */
    public String getDocumentMetadata(String[] docIDs) {
        return getMetadata(docIDs);
    }
    
    /** Given a collectionID, returns a GS3 DocumentMetadataRetrieve 
     * response message that gives the metadata for the collection identified
     * @param collID is a fedora pid identifying a collection in its repository
     * @return a GS3 DocumentMetadataRetrieve response message containing the
     * EX metadata for the requested collection */
    public String getCollectionMetadata(String collID) {
        return getMetadata(new String[] {collID});
    }
    
    /** Given a document identifier, returns a GS3 DocumentMetadataRetrieve 
     * response message containing the metadata for the document. 
     * @param docID is a document identifier (docID can either be a &lt;pid&gt; 
     * of an item (document) in the fedora repository, or it can be 
     * "&lt;pid&gt;-sectionNumber". 
     * @return a GS3 DocumentMetadataRetrieve response message containing the
     * EX, DC, DLS metadata for the requested document */
    public String getDocumentMetadata(String docID) {
        return getMetadata(new String[] {docID});
    }

    /** @return a greenstone DocumentMetadataRetrieve response for the 
     * documents or collections indicated by the docIDsOrCollIDs.
     * @param docIDsOrCollIDs is an array of identifiers which may be either the
     * fedora pids for collections, or otherwise may be a document identifier.
     * In the last case, the document ID may consist of either
     * "documentPID-sectionNumber" or may just be just fedora documentPID */
    public String getMetadata(String[] docIDsOrCollIDs)
    {
        Document doc = builder.newDocument();
        FedoraGS3RunException ex = null; 
        
        Element docNodeList = doc.createElement(
                GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
        
        try{
            for(int i = 0; i < docIDsOrCollIDs.length; i++) {
                // create the <documentNode> containing the metadata 
                // for each document docID
                Element docNode = getMetadata(doc, docIDsOrCollIDs[i]); 
                docNodeList.appendChild(docNode);
            }
        } catch(Exception e) { 
            ex = new FedoraGS3RunException(e);
            ex.setSpecifics("EX (and/or DC, DLS) metadata datastream");
        }
        
        Element responseMsg = createResponseMessage(doc, docNodeList, ex,  
                GSXML.REQUEST_TYPE_PROCESS, "DocumentMetadataRetrieve");
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        } catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        } 
    }
    
    /** Method that takes a new DOM document, as well as an identifier of either
     * a collection or document (which may be a fedora pid for the collection 
     * or document, or may be the documentPid-sectionNumber for a document) and
     * returns a documentNode element for it: 
     * &lt;documentNode&gt;&lt;metadataList&gt;
     * &lt;metadata name=""&gt;value&lt;/metadata&gt; 
     * ...
     * &lt;/metadataList&gt;&lt;/documentNode&gt;
     * @return documentNode containing the metadata for the collection or 
     * document given by parameter ID
     * @param id denotes a collection pid, a document pid or a docID of the  
     * form "documentpid-sectionNumber" */
    protected Element getMetadata(Document doc, String id) 
        throws RemoteException, UnsupportedEncodingException, 
            SAXException, IOException 
    {
        // We're going to create the documentNode nested inside the following
        // documentNodeList: 
            // <documentNodeList>
            // <documentNode nodeID=""><metadataList>
            // <metadata name="">value</metadata>
            // </metadataList></documentNode>
            // <documentNode>...</documentNode>
            // </documentNodeList>
            // <documentNodeList>
        
        // <documentNode nodeID="docID"> - the docNode on which a structure  
        // retrieve is being performed
        Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
        Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
        attribute.setValue(id);
        docNode.setAttributeNode(attribute);
        
        // <metadataList>
        Element metadataList = doc.createElement(
                GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
        
        String ex = "";
        String dc = "";
        String dls = "";
        if(id.endsWith(_COLLECTION)) { // docID refers to a collection
            // Obtain the "EX" datastream (extracted metadata) for the collection
            ex = this.getEX(id);
        }
        else { // docID refers to a document 
            // work out the document's fedora PID and section ID, and then 
            // obtain the EX (extracted metadata) and DC datastreams for the doc
            
            // Note that EX/DC for pid="greenstone:<colname>-docPID-1"
            // is the same as for pid="greenstone:<colname>-docPID"
            // That is, <Section id="1"> refers to the toplevel document docPID
            // If requested for top-level document, there may also be DLS meta 
            String sectionID = getSectionIDFromDocID(id);
            String docPID = getDocPIDFromDocID(id);
            if(sectionID.equals("") || sectionID.equals("1")) { 
                // metadata of toplevel document is requested
                ex = this.getEX(docPID); // slightly faster than doing
                        //getSectionEXMetadata(docID, "1")
                dc = this.getDC(docPID);
                dls = this.getDLS(docPID);
            }
            else {
                ex = getSectionEXMetadata(docPID, sectionID);
                dc = getSectionDCMetadata(docPID, sectionID);
            }
        }
        
        // Adding in metadata sets in alphabetical order
        // DC metadata for a top-level document is different from EX, DLS: 
        // only the element's namespace prefix is "dc", the rest of a tagname
        // is unknown.
        if(!dc.equals("")) {
            addMetadataWithNamespacedTagNames(doc, metadataList,
                    dc, DC);
        }
        
        // Check if we were supposed to process dls and dc metadata
        // as well. We only ever do this for top-level documents,
        // in which case, dls and dc will be non-empty strings
        if(!dls.equals("")) {
            addMetadataWithFixedTagName(doc, metadataList, dls, DLS);
        }
        
        // we definitely have an EX metadatastream for each 
        // collection object, top-level document object, 
        // and document section item 
        addMetadataWithFixedTagName(doc, metadataList, ex, EX);
        
        // now the metadataList has been built up
        docNode.appendChild(metadataList);
        
        return docNode; // return <documentNode> containing the metadata 
    }
    
    /** This method retrieves all the metadata elements in the metaDataStream
     * parameter of the form &lt;"metadataSetNS:metadata"&gt;"value"&lt;/metadata&gt; where 
     * metadataSetNS is the namespace of each tag, and creates a new element of
     * the form &lt;metadata name="metadataSetNS:metadata"&gt;"value"&lt;/metadata&gt; for 
     * each. Each of these are then appended to the metadataList parameter.
     * @param doc is the Document object using which the new metadata Elements
     * are to be constructed
     * @param metadataList is the &lt;metadataList&gt; Element to which the new
     * metadata Elements are to be appended as children.
     * @param metaDatastream the metadata datastream in string form (e.g. the
     * Dublin Core metadata stored in the Fedora repository).
     * @param metadataSet is the constant datastream identifier, e.g. "DC".
     * At present this method only applies to the DC metadata as that's the only 
     * one where each tagname is different except for the constant dc: namespace.
    */
    protected void addMetadataWithNamespacedTagNames(Document doc, 
            Element metadataList, String metaDatastream, String metadataSet)
        throws SAXException, IOException
    {
        Document src = builder.parse(
                new InputSource(new StringReader(metaDatastream)));
        
        // The following doesn't work for some reason: to retrieve all elements 
        // whose namespace prefix starts with "dc", we pass "*" for localName
        //NodeList dcMetaTags = src.getElementsByTagNameNS(DC.toLowerCase(), "*");
        
        // Longer way: get the children of the root document
        NodeList children = src.getDocumentElement().getChildNodes();
        
        for(int i = 0; i < children.getLength(); i++) {
            String nodeName = children.item(i).getNodeName();
            // check that the nodename starts with the "dc" namespace,
            // which simultaneously ensures that the node's an element:
            if(nodeName.startsWith(DC.toLowerCase())) {
                // need to have a period for Greenstone instead of Fedora's colon
                nodeName = nodeName.replace(COLON, PERIOD);
                Element metatag = (Element)children.item(i);
                String value = FedoraCommons.getValue(metatag);
                // <dc:tagname>value</dc:tagname>
                // we're going to put use this in our metadata element as 
                // <metadata name="dc:tagname">value</metadata>
                
                // create metadata of (name, value) pairs in target DOM (doc)
                Element metadata = doc.createElement(GSXML.METADATA_ELEM);
                Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
                attribute.setValue(nodeName);
                metadata.setAttributeNode(attribute);
                Text content = doc.createTextNode(value);
                metadata.appendChild(content);
                metadataList.appendChild(metadata);
            }
        }       
    }
    
    /** This method retrieves all the metadata elements in the metaDataStream
     * of the form &lt;"namespace:"metadata name="metadataName"&gt;value&lt;/metadata&gt;  
     * where "namespace" is the namespace prefix of each tag, and metadataName  
     * is the name of the metadata (like author, title). For each element
     * it creates a corresponding new element of the form 
     * &lt;metadata name="namespace:metadataName"&gt;value&lt;/metadata&gt;. Each of these 
     * are then appended to the metadataList parameter.
     * @param doc is the Document object using which the new metadata Elements
     * are to be constructed
     * @param metadataList is the &lt;metadataList&gt; Element to which the new
     * metadata Elements are to be appended as children.
     * @param metaDatastream the metadata datastream in string form (e.g. the
     * EX/Greenstone extracted metadata or DLS metadata stored in the Fedora 
     * repository).
     * @param metadataSet is the constant datastream identifier, 
     * e.g. "DLS" or "EX".
     * At present this method applies to the DLS and EX metadata as they have
     * constant tagnames throughout.
    */
    protected void addMetadataWithFixedTagName(Document doc, 
            Element metadataList, String metaDatastream, String metadataSet) 
        throws SAXException, IOException
    {
        // Namespace prefix can be "ex:" or "dls:"
        String namespacePrefix = "";
        if(!metadataSet.equals(EX)) { 
            // need to have a period for Greenstone instead of Fedora's colon
            namespacePrefix = metadataSet.toLowerCase() + PERIOD;
        }
                
        Document src = builder.parse(
                new InputSource(new StringReader(metaDatastream)));
        NodeList metaTags = src.getElementsByTagName(
                metadataSet.toLowerCase()+COLON+METADATA); 
                // Looking for tagnames: <ex:metadata> or <dls:metadata>
            
        for(int i = 0; i < metaTags.getLength(); i++) {
            Element metatag = (Element)metaTags.item(i);
            
            // extract the metadata of (name, value) pairs from src DOM
            // look for <metadata name="name">value</metadata>
            String name = metatag.hasAttribute(NAME) ? 
                    metatag.getAttribute(NAME) : "";
            // sometimes, there are several metadata for the same name, in this
            // case, look for a qualifier and append its value to the name to
            // distinguish it uniquely: 
            if(metatag.hasAttribute(QUALIFIER)) {
                name = name + HYPHEN + metatag.getAttribute(QUALIFIER);
            }
            String value = FedoraCommons.getValue(metatag);
            
            // create metadata of (name, value) pairs in target DOM (doc)
            Element metadata = doc.createElement(GSXML.METADATA_ELEM);
            Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
            attribute.setValue(namespacePrefix + name);  
                                    // prefix with namespace, if any
            metadata.setAttributeNode(attribute);
            Text content = doc.createTextNode(value);
            metadata.appendChild(content);
            
            metadataList.appendChild(metadata);
        }
    }
    
    /** Given a document identifier, returns a GS3 DocumentMetadataRetrieve 
     * response message containing ONLY the Title metadata for the document. 
     * @param docID is a document identifier (docID can either be a &lt;pid&gt; 
     * of an item (document) in the fedora repository, or it can be 
     * "&lt;pid&gt;-sectionNumber". 
     * @return a GS3 DocumentMetadataRetrieve response message containing the
     * Title metadata for the requested document */
    public String getTitleMetadata(String docID) {
        return getTitleMetadata(new String[] { docID });
    }
    
    /** Given a document identifier, returns a GS3 DocumentMetadataRetrieve 
     * response message containing ONLY the Title metadata for the documents. 
     * @param docIDs is a list of document identifiers (where docID can either be 
     * a &lt;pid&gt; of an item (document) in the fedora repository, or it can be 
     * "&lt;pid&gt;-sectionNumber". 
     * @return a GS3 DocumentMetadataRetrieve response message containing the
     * Title metadata for all the requested documents */
    public String getTitleMetadata(String[] docIDs) {
        // Must create message of the following form: 
        // <documentNodeList><documentNode nodeID="docID">
        // <metadataList><metadata name="Title">sometitle</metadata>
        // </metadataList></documentNode>
        
        Document doc = builder.newDocument();
        FedoraGS3RunException ex = null; 
        
        Element docNodeList = doc.createElement(
                GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
        try{
            for(int i = 0; i < docIDs.length; i++) {
                Element docNode = getTitleMetadata(doc, docIDs[i]);
                docNodeList.appendChild(docNode);
            }
        }catch(Exception e) { 
            ex = new FedoraGS3RunException(e);
            //ex.setSpecifics("EX metadata datastream PID: |" + docIDs[i] + "|"); // for debugging PID
            ex.setSpecifics("EX metadata datastream");
        }
        
        Element responseMsg = createResponseMessage(doc, docNodeList, ex,  
                GSXML.REQUEST_TYPE_PROCESS, "DocumentMetadataRetrieve");
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        } catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        } 
    }
    
    /** Method that takes a new DOM document, as well as an identifier of either
     * a document or document section and returns a documentNode element containing
     * the title metadata for it: 
     * &lt;documentNode nodeID="docID"&gt;&lt;metadataList&gt;
     * &lt;metadata name="Title"&gt;sometitle&lt;/metadata&gt;
     * &lt;/metadataList&gt;&lt;/documentNode&gt;
     * @return documentNode containing the metadata for the collection or 
     * document given by parameter ID
     * @param docID denotes the id of a document or a document section, so id 
     * is either a document-pid or it's of the form documentpid-sectionNumber */
    protected Element getTitleMetadata(Document doc, String docID) 
        throws RemoteException, UnsupportedEncodingException, 
            SAXException, IOException 
    {
        // Returns a docNode element of the following form:
        // <documentNode nodeID="docID">
        // <metadataList><metadata name="Title">sometitle</metadata></metadataList>
        // </documentNode>
        
        // <documentNode nodeID="docID">
        Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
        Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
        attribute.setValue(docID);
        docNode.setAttributeNode(attribute);
        
        // <metadataList>
        Element metaList = doc.createElement(
                GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
        // <metadata name="Title">
        Element metadata = doc.createElement(GSXML.METADATA_ELEM);
        // if we connect it all up (append children), we can immediately add
        // the name attribute into the metadata element:
        metaList.appendChild(metadata);
        docNode.appendChild(metaList);
        metadata.setAttribute(GSXML.NAME_ATT, TITLE); // immediately add attribute 
        
        String title = "";
        String sectionID = getSectionIDFromDocID(docID);
        String docPID = getDocPIDFromDocID(docID);
    
        // check if title of toplevel document is requested
        if(sectionID.equals("")) 
            title = this.getDocTitle(docPID);
        else { // title of document section
            title = this.getSectionTitle(docPID, sectionID);
        }
        
        metadata.appendChild(doc.createTextNode(title));
        
        return docNode;
    } 
    
        
    /** @return the documentStructure of the document or section given by docID. 
     * The structure is returned in the XML format of a Greenstone3 
     * DocumentStructureRetrieve response message. This method returns the entire
     * subSection of the docID (that is, all descendants included).
     * @param docID the identifier for the document whose structure is required.
     * This is of the format "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;"
     * OR "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-&lt;sectioNumber&gt;"  
     * where "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-1" is the same as 
     * "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;" and will return the 
     * same response */
    public String getDocumentStructure(String docID) {
        return getStructure(new String[]{docID}, DESCENDANTS);
    }
    
    /** @return a view of the structure of the document or section given by docID
     * which contains only the section and its direct children. This structure is 
     * returned in the XML format of a Greenstone3 DocumentStructureRetrieve 
     * response message. 
     * @param docID the identifier for the document whose structure is required.
     * This is of the format "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;"
     * OR "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-&lt;sectioNumber&gt;"  
     * where "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-1" is the same as 
     * "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;" and will return the 
     * same response */
    public String getChildren(String docID) {
        return getStructure(new String[]{docID}, CHILDREN);
    }
    
    /** @return the documentStructure of the documents or sections given by docIDs. 
     * The structure is returned in the XML format of a Greenstone3 
     * DocumentStructureRetrieve response message. This method returns the entire
     * subSection of each docID (that is, all descendants included).
     * @param docIDs is an array of identifiers for the documents whose structures
     * are required.
     * This is of the format "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;"
     * OR "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-&lt;sectioNumber&gt;"  
     * where "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-1" is the same as 
     * "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;" and will return the 
     * same response */
    public String getDocumentStructure(String[] docIDs) {
        return getStructure(docIDs, DESCENDANTS);
    }
    
    /** @return the documentStructure of the documents or sections given by docIDs
     * but only the sections and their children (not any further descendants). 
     * The structure is returned in the XML format of a Greenstone3 
     * DocumentStructureRetrieve response message.
     * @param docIDs the identifiers for the documents whose structures are 
     * required. The docids are of the format "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;"
     * OR "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-&lt;sectioNumber&gt;"  
     * where "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-1" is the same as 
     * "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;" and will return the 
     * same response */
    public String getChildren(String[] docIDs) {
        return getStructure(docIDs, CHILDREN);
    }
    
    /** 
     * Returns a greenstone3 DocumentStructureRetrieve XML response message 
     * containing the document structures for the given docIDs.
     * Similar to FedoraConnection.getTOC(), but instead of fedora formatted XML,  
     * greenstone formatted XML is returned. The requested section of the table
     * of contents (TOC) for a document is converted into the greenstone3 xml  
     * format that is returned upon DocumentStructureRetrieve requests.
     * @param docIDs the documentIDs for which the section's structure is returned; 
     * where a docID is either a fedora pid &lt;docPID&gt; or &lt;docPID&gt;-&lt;sectionNumber&gt;.
     * @param levels - either CHILDREN or DESCENDANTS. 
     * CHILDREN returns only the first-level descendants (children) of the 
     * requested document sections indicated by docIDs.
     * DESCENDANTS returns all descendants of all the document-sections denoted by 
     * docIDs.
     * @return a greenstone3 DocumentStructureRetrieve XML response message in
     * String format with the structure of the docIDs requested.
    */
    protected String getStructure(String[] docIDs, int levels) 
    {
        Document doc = builder.newDocument();
        FedoraGS3RunException ex = null;
        // <documentNodeList>
        Element docNodeList = doc.createElement(
                GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
        
        try{
            // append the <documentNodes> for the docIDs
            // to the docNodeList
            getStructureElement(docNodeList, docIDs, levels); 
        } catch(Exception e) { 
            ex = new FedoraGS3RunException(e);
            ex.setSpecifics("(requested portion of) TOC datastream");
        }
        // insert our <documentNodeList> into a GS3 response message 
        Element responseMsg = createResponseMessage(doc, docNodeList, ex,
                GSXML.REQUEST_TYPE_PROCESS, "DocumentStructureRetrieve");
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        } catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    /** Given a &lt;documentNodeList&gt; portion of a greenstone3 
     * DocumentStructureRetrieve XML response message, this method will populate 
     * it with the &lt;documentNodes&gt; that represent the structure of the given docIDs.
     * @param docNodeList is a &lt;documentNodeList&gt; to which &lt;documentNodes&gt; of
     * the doc structures are appended.
     * @param docIDs the documentIDs for which the section's structure is returned; 
     * where a docID is either a fedora pid &lt;docPID&gt; or &lt;docPID&gt;-&lt;sectionNumber&gt;.
     * @param levels - either CHILDREN or DESCENDANTS. 
     * CHILDREN returns only the first-level descendants (children) of the 
     * requested document sections indicated by docIDs.
     * DESCENDANTS returns all descendants of all the document-sections denoted by 
     * docIDs.
    */
    protected void getStructureElement(Element docNodeList, 
            String[] docIDs, int levels) 
        throws RemoteException, UnsupportedEncodingException, SAXException, 
            IOException 
    {
        // process each docID
        for(int i = 0; i < docIDs.length; i++) {
            // work out the document's fedora PID and section ID
            String sectionID = getSectionIDFromDocID(docIDs[i]);
            String docPID = getDocPIDFromDocID(docIDs[i]);
            
            // get the required section, along with children or descendants
            Element srcDocElement = null;
            if(levels == CHILDREN) // get the requested section with its children 
                srcDocElement = this.getChildrenOfSectionXML(docPID, sectionID);
            else // levels == DESCENDANTS, get the section with all its descendants
                srcDocElement = this.getSubsectionXML(docPID, sectionID);
            
            // copy-and-convert that structure into a structure format for GS3
            Element docNode = getStructure(docNodeList.getOwnerDocument(), 
                    docIDs[i], docPID, srcDocElement);
            
            // add it to our list of documentNodes
            docNodeList.appendChild(docNode);
        }
    }
        
    /** 
     * Takes the portion of the XML document outlining the structure of the 
     * document (section)--in the format this is stored in Fedora--and returns 
     * Greenstone 3 DOM XML format for outlining document structure.  
     * @return a &lt;documentNode&gt; element that contains a greenstone3 
     * DocumentStructureRetrieve XML corresponding to the parameter Element section
     * (which is in fedora XML), for the document indicated by docID.
     * @param requestingDocID is the identifier of the document for which the 
     * structure was requested. It's this document's children or descendants that
     * will be returned. Note that this is not always the same as (clear from)  
     * parameter docID.
     * @param docID is the documentID for which the section's structure is 
     * returned where docID = "docPID-sectionNumber".
     * @param section - the fedora section XML that is being mirrored in
     * greenstone3 format.
    */
    protected Element getStructure(Document doc, String requestingDocID, 
            String docID, Element section)
    {
        // we want to mirror the section's DOM (given in fedora XML) in 
        // greenstone3's XML for a DocumentStructureRetrieve response.
        
        // <documentNode nodeID="docID"> - the docNode on which a structure retrieve  
        // is being performed
        Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
        Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
        attribute.setValue(requestingDocID); //requestingDocID.replace(HYPHEN+SECTION, "")
        docNode.setAttributeNode(attribute);
        
        // <nodeStructure>
        Element nodeStructure = doc.createElement(GSXML.NODE_STRUCTURE_ELEM);
        
        // <documentNode nodeID="docID" docType="hierarchy" nodeType="root">
        Element rootNode = createDocNodeFromSubsection(doc, section, docID);
        
        // fills in the subtree of the rootNode in our nodeStructure element
        createDocStructure(doc, section, rootNode, docID); 
            //where section represents the root section
        
        nodeStructure.appendChild(rootNode);
        docNode.appendChild(nodeStructure);
        return docNode;     
    }
    
    /** Recursive method that creates a documentStructure mirroring parameter
     * section, starting from parameter parent down to all descendants
     * @param section is the XML &lt;Section&gt; in the fedora repository's TOC 
     * for the docPID whose substructure is to be mirrored
     * @param parent is the XML documentNode in the greenstone repository whose
     * descendants created by this method will correspond to the descendants of 
     * parameter section.  
     * @param doc is the document containing the parent; 
     * @param docPID is the prefix of all nodeIDs in the parent's structure
    */
    protected void createDocStructure(
            Document doc, Element section, Element parent, String docPID)
    {
        // get the section's children (if any)
        NodeList children = section.getChildNodes();
        for(int i = 0; i < children.getLength(); i++) {
            Node n = children.item(i);
            
            if(n.getNodeName().equals(SECTION_ELEMENT)) {
                //then we know it's an element AND that its tagname is "Section"
                Element subsection = (Element)n;
                Element child = createDocNodeFromSubsection(doc, subsection, docPID);
                parent.appendChild(child);
                
                // recursion call on newly found child-element and subsection
                createDocStructure(doc, subsection, child, docPID);
            }
        }
    }
    
    /** Given a particular subsection element, this method creates a 
     * Greenstone3 DocumentNode element that mirrors it.
     * @param doc is the document that will contain the created DocumentNode 
     * @param docID is the prefix of all nodeIDs in the parent's structure 
     * @param subSection is the XML &lt;Section&gt; in the fedora repository's 
     * TOC for the docPID which will be mirrored in the greenstone XML 
     * documentNode that will be returned.
     * @return a greenstone &lt;documentNode&gt; that represents the fedora TOC's 
     * &lt;Section&gt; element passed as parameter subSection. */
    protected Element createDocNodeFromSubsection(
            Document doc, Element subSection, String docID) 
    {
        Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
        Attr docType = doc.createAttribute(GSXML.DOC_TYPE_ATT);
        docType.setValue(GSXML.DOC_TYPE_HIERARCHY);
        docNode.setAttributeNode(docType);
        
        Attr nodeID = doc.createAttribute(GSXML.NODE_ID_ATT);
        String sectionID = subSection.hasAttribute(ID) ?
                subSection.getAttribute(ID) : "";
        nodeID.setValue(docID + HYPHEN + sectionID);
        docNode.setAttributeNode(nodeID);
        
        Attr nodeType = doc.createAttribute(GSXML.NODE_TYPE_ATT);
        if(sectionID.equals("1")) { // root case
            nodeType.setValue(GSXML.NODE_TYPE_ROOT);
            // reset the attribute without the section number
            docNode.setAttribute(GSXML.NODE_ID_ATT, docID);
        }
        else if(subSection.getElementsByTagName(SECTION_ELEMENT).getLength() > 0)
            // this section has further <Section> children, so it's an internal node 
            nodeType.setValue(GSXML.NODE_TYPE_INTERNAL);
        else if(subSection.hasAttribute(TYPE)) 
            nodeType.setValue(GSXML.NODE_TYPE_INTERNAL);
        else // leaf
            nodeType.setValue(GSXML.NODE_TYPE_LEAF);
        docNode.setAttributeNode(nodeType);
        return docNode;
    }
    
    
    /** Given an identifier that is either a docPID or a concatenation of 
     * docPID+sectionID, this method works out the fedora assigned docPID and 
     * sectionID and then calls getContentBody(docPID, sectionID) with those. 
     * @param docID is expected to be of the form 
     * "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-&lt;sectionNumber&gt;" or
     * "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;"
     * If it is "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;", then the content for
     * "greenstone:&lt;collectionName&gt;-1" ("greenstone:&lt;collectionName&gt;-Section1")
     * is returned! */
    public String getContent(String docID) {
        return this.getContent(new String[]{docID});
    }
    
    /** Given an identifier that is a concatenation of docID+sectionID, this
     * method works out the fedora assigned docPID and sectionID and then calls 
     * getContentBody(docPID, sectionID) with those. 
     * @param docIDs is an array of document identifiers of the form 
     * "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;-&lt;sectionNumber&gt;" 
     * If it is "greenstone:&lt;collectionName&gt;-&lt;docPID&gt;", then the content for
     * "greenstone:&lt;collectionName&gt;-Section1" is returned! */
    public String getContent(String[] docIDs) {
        Document doc = builder.newDocument();
        FedoraGS3RunException ex = null; 
        
        //<documentNodeList>
        Element docNodeList = doc.createElement(
                GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
        
        try{
            for(int i = 0; i < docIDs.length; i++) {
                // get the sectionID and docPID from the docID
                String sectionID = this.removePrefix(
                        getSectionIDFromDocID(docIDs[i]), SECTION);
                String docPID = getDocPIDFromDocID(docIDs[i]);
                if(sectionID.equals("")) // if no section is specified, get 
                    sectionID = "1"; // get the content for Section id="1"
                            
                // Get the contents for the requested section of document docPID 
                String sectionContent = this.getContentBody(docPID, sectionID);
                
                // set the nodeID attribute
                Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
                Attr nodeId = doc.createAttribute(GSXML.NODE_ID_ATT);
                
                nodeId.setValue(docIDs[i]); // just set the docID which will contain
                            // the docPID (and sectionID if already present)
                
                docNode.setAttributeNode(nodeId);
                // set the text content to what was retrieved
                Element nodeContent = doc.createElement(GSXML.NODE_CONTENT_ELEM);
                Text textNode = doc.createTextNode(sectionContent.trim());
                
                nodeContent.appendChild(textNode);
                docNode.appendChild(nodeContent);
                //add the documentNode to the docNodeList 
                docNodeList.appendChild(docNode);
            }
        } catch(Exception e) { 
            ex = new FedoraGS3RunException(e);
            ex.setSpecifics("requested doc Section datastream");
        }
        Element responseMsg = createResponseMessage(doc, docNodeList, ex,
                GSXML.REQUEST_TYPE_PROCESS, "DocumentContentRetrieve");
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        } catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
        
    /** Gets the contents of a textNode from a section.
     * @return the text content of a section.
     * @param docPID the pid of the document from which a section's text is to 
     * be retrieved.
     * @param sectionID is the section identifier of the document denoted by 
     * docPID whose text is to be returned.
    */
    protected String getContentBody(String docPID, String sectionID) 
        throws RemoteException, UnsupportedEncodingException, 
            SAXException, IOException
    {   
        String section = this.getSection(docPID, sectionID);
        
        // the content is nested inside a <Section> element, 
        // we extract it from there:
        InputSource source = new InputSource(new StringReader(section));
        Document doc = builder.parse(source);
            
        // The document Element is the <Section> we want. 
        // Get its text contents:
        section = FedoraCommons.getValue(doc.getDocumentElement());
        
        // we are going to remove all occurrences of "_httpdocimg_/"
        // that precede associated filenames, because that's a GS3 
        // defined macro for resolving relative urls. It won't help
        // with documents stored in fedora.
        section = section.replaceAll(GS3FilePathMacro+"/", "");
        return section;
    }
        
    /** Here we create the greenstone's response message element: 
     * &lt;message&lg;&lt;response&gt;&lt;content&gt;&lt;/response&gt;&lt;/message&gt; 
     * @return a greenstone response-message element.
     * @param doc - the Document object which should me used to create the
     * &lt;message&gt; and &lt;response&gt; elements 
     * @param content - the element that is to be nested inside &lt;response&gt;
     * @param ex - any exception that occurred when trying to create
     * the content parameter
     * @param responseType - the value for the type attribute of &lt;response&gt;,
     * such as "describe", "retrieve", "browse", "query"... 
     * @param originator - indiates the collectionName or service (like 
     * DocumentContentRetrieve) from where this response message originates
    */
    protected Element createResponseMessage(Document doc, Element content, 
            Exception ex, String responseType, String originator) 
    {
        Element response = doc.createElement(GSXML.RESPONSE_ELEM);
        // from = "FedoraGS3"
        Attr attribute = doc.createAttribute(GSXML.FROM_ATT);
        String from = originator.equals("") ? FEDORA_GS3
                : FEDORA_GS3+"/"+originator;
        
        attribute.setValue(from);
        response.setAttributeNode(attribute);
        
        // type = "describe" or "process" - whatever's given in requestType: 
        attribute = doc.createAttribute(GSXML.TYPE_ATT);
        attribute.setValue(responseType);
        response.setAttributeNode(attribute);
        
        if(content != null)
            response.appendChild(content);
        
        // we'll create an error element for RemoteExceptions (web service problems)  
        // and UnsupportedEncodingExceptions and 
        if(ex != null) {
            Element error = doc.createElement(GSXML.ERROR_ELEM);
            error.appendChild(doc.createTextNode(ex.getMessage()));
            // now append the error to the <response> element (after
            // the content element whatever that was)
            response.appendChild(error);
        }
        
        Element message = doc.createElement(GSXML.MESSAGE_ELEM);
        message.appendChild(response);
        doc.appendChild(message);
        return message;
    }
    
    /** @return a &lt;serviceList&gt; Element as defined by GS3: containing all the
     * services (denoted by &lt;service&gt; elements) that are supported by FedoraGS3. 
     * At present these are: DocumentContentRetrieve, DocumentMetadataRetrieve,
     * DocumentStructureRetrieve, TextQuery, FieldQuery, ClassifierBrowse, 
     * ClassifierBrowseMetadataRetrieve (as indicated by member var serviceNames). 
     * @param doc - the Document object which should me used to create the
     * &lt;serviceList&gt; element */
    protected Element createServiceList(Document doc) 
    { 
        Element serviceList = doc.createElement(
                GSXML.SERVICE_ELEM+GSXML.LIST_MODIFIER);
        
        for(int i = 0; i < serviceNames.length; i++) {
            // create the <service name="serviceName[i]" type="servicetype" />
            Element service = doc.createElement(GSXML.SERVICE_ELEM);
            
            Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
            attribute.setValue(serviceNames[i]);
            service.setAttributeNode(attribute);
            
            attribute = doc.createAttribute(GSXML.TYPE_ATT);
            if(serviceNames[i].equals("ClassifierBrowse")) //browseTitlesByLetter 
                attribute.setValue(GSXML.SERVICE_TYPE_BROWSE);
            else if(serviceNames[i].contains("Query")) // search services 
                attribute.setValue(GSXML.SERVICE_TYPE_QUERY);
            else 
                attribute.setValue(GSXML.SERVICE_TYPE_RETRIEVE);
            service.setAttributeNode(attribute);
            
            // add the service element to the serviceList element
            // <serviceList><service /></serviceList>
            serviceList.appendChild(service);
        }
        return serviceList;
    }
    
    /** @return a GS3 response message for a describe services request: 
     * indicating the list of services supported by the Fedora-Greenstone 
     * interface. These are DocumentContentRetrieve, DocumentMetadataRetrieve,
     * DocumentStructureRetrieve, ClassifierBrowse, TextQuery, FieldQuery, 
     * ClassifierBrowseMetadataRetrieve - as indicated by member variable
     * serviceNames. */
    public String getServiceList() 
    {
        Document doc = builder.newDocument();
        Element serviceList = createServiceList(doc);
        // make <serviceList> the body of the responseMessage: 
        // <message><response><serviceList></response></message>
        Element responseMsg = createResponseMessage(doc, serviceList, null,
                GSXML.REQUEST_TYPE_DESCRIBE, "");
        try {
            return FedoraCommons.elementToFormattedString(responseMsg);
        }catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    /** @return a GS3 describe response message listing the collections and  
     * collection-specific metadata stored in the Fedora-Greenstone repository. */
    public String getCollectionList()
    {
        Document doc = builder.newDocument();
        FedoraGS3RunException ex = null; // any RemoteException
        
        // create the <collectionList /> element
        Element collectionList = doc.createElement(
                GSXML.COLLECTION_ELEM+GSXML.LIST_MODIFIER);
        try{
            String[] collectionNames = this.getCollectionNames(
                    this.getCollections()); // this line could throw RemoteException 
            for(int i = 0; i < collectionNames.length; i++) {
                // create the <collection name="somename" /> element
                Element collection = doc.createElement(GSXML.COLLECTION_ELEM);
                Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
                attribute.setValue(collectionNames[i]);
                collection.setAttributeNode(attribute);
                
                // append the <collection> element as child of <collectionList>
                collectionList.appendChild(collection);
                
                //if(collection.hasAttribute(GSXML.NAME_ATT))
                    //LOG.debug(collection.getAttribute(GSXML.NAME_ATT));
            }
        } catch(RemoteException e) { // if this happens, perhaps it's because it 
            // can't find Greenstone collections in fedora repository?
            ex = new FedoraGS3RunException(e);
            ex.setSpecifics(
                "greenstone collections in fedora repository");
        } 
        
        // make <collectionList> the body of the responseMessage: 
        // <message><response><collectionList></response></message>
        Element responseMsg = createResponseMessage(doc, collectionList, ex,
                GSXML.REQUEST_TYPE_DESCRIBE, "");
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        }catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    /** @return a GS3 describe response message for a collection in the 
     * Fedora-Greenstone repository. 
     * @param collectionName - the name of the collection that is to be described.  
     * It will be converted to a fedora collection pid, which is of the form 
     * "greenstone:&lt;collectionName&gt;-collection". */
    public String describeCollection(String collectionName)
    {
        Document doc = builder.newDocument();
        FedoraGS3RunException ex = null;
        
        Element collection = doc.createElement(GSXML.COLLECTION_ELEM);
        Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(collectionName);
        collection.setAttributeNode(attribute);
        
        //<displayItem assigned="true" lang="en" name="name">
        //"some display name"</displayItem>
        Element displayItem = doc.createElement(GSXML.DISPLAY_TEXT_ELEM);
        
        attribute = doc.createAttribute(GSXML.LANG_ATT);
        attribute.setValue(this.lang);
        displayItem.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(GSXML.DISPLAY_TEXT_NAME);
        displayItem.setAttributeNode(attribute);
        
        try{
            Text textNode = doc.createTextNode(
                this.getCollectionTitle(getCollectionPID(collectionName)));
            displayItem.appendChild(textNode);
        } catch(Exception e) { 
            // can't find Greenstone collections in fedora repository or problem
            // getting their titles from their metadata datastream?
            ex = new FedoraGS3RunException(e);
            ex.setSpecifics("greenstone collections or their metadata" 
                    + "in the fedora repository");
        } 
        // now append the displayItem element as child of the collection element
        collection.appendChild(displayItem);
        // get the <serviceList> and add it into the collection description.
        // Services for all collections in the FedoraGS3 repository are the 
        // same, offering a ClassifierBrowse to browse titles by starting letter 
        // and DocRetrieve services: Content, Metadata and Structure.
        
        Element serviceList = createServiceList(doc);
        collection.appendChild(serviceList);
        
        Element responseMsg = createResponseMessage(doc, collection, ex,
                GSXML.REQUEST_TYPE_DESCRIBE, collectionName);
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        }catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    /** @return a GS3 describe response message for the services of a collection 
     * in the Fedora-Greenstone repository. So far, these services are the same for
     * all fedora collections: they are the services given in member variable
     * serviceNames: DocumentContent/Metadata/StructureRetrieve, ClassifierBrowse,
     * ClassifierBrowseMetadataRetrieve.   
     * All collections in this Digital Library (Fedora Repository) share the
     * same services, so this method returns the same services as getServiceList();
     * @param collectionName - the name of the collection whose services are to
     * be described. It will be converted to a fedora collection pid, which is of 
     * the form "greenstone:&lt;collectionName&gt;-collection". */
    public String describeCollectionServices(String collectionName)
    {
        Document doc = builder.newDocument();
        
        Element collection = doc.createElement(GSXML.COLLECTION_ELEM);
        Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(collectionName);
        collection.setAttributeNode(attribute);
        
        Element serviceList = createServiceList(doc);
        collection.appendChild(serviceList);
        
        Element responseMsg = createResponseMessage(doc, collection, null,
                GSXML.REQUEST_TYPE_DESCRIBE, collectionName);
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        }catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    /** All collections in this Digital Library (Fedora Repository) share 
     * the same services, so this method returns the same as 
     * describeCollectionService(collName, serviceName). 
     * @return a GS3 describe response message for the requested service 
     * of the given collection. DocumentContent/Metadata/StructureRetrieve 
     * return nothing special except their names; browse (and any query)
     * return more complex XML responses.
     * @param serviceName - the name of the service in the collection which is to 
     * be described.*/
    public String describeService(String serviceName)
    {
        // For all the *retrieve* services (incl ClassifierBrowseMetadataRetrieve)
        // we return: 
        // <message><response from="<name>Retrieve" type="describe">
        // <service name="<name>Retrieve" type="retrieve" /></response></message>
        // But for browse (and any query) service, we return the data necessary
        // for displaying it
        
        Document doc = this.builder.newDocument();
        Element service = doc.createElement(GSXML.SERVICE_ELEM);
        Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(serviceName);
        service.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.TYPE_ATT);
        
        if(serviceName.toLowerCase().endsWith("retrieve")) {
            attribute.setValue(GSXML.SERVICE_TYPE_RETRIEVE);
        } 
        else if(serviceName.toLowerCase().contains("browse")) {
            attribute.setValue(GSXML.SERVICE_TYPE_BROWSE);
            
            // we need name and description <displayItem> elements
            Element displayItem 
                = createNameValuePairElement(doc, 
                        GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, "Browse");
            service.appendChild(displayItem);
            
            displayItem = createNameValuePairElement(doc, 
                    GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_DESCRIPTION, 
                    "Browse pre-defined classification hierarchies");
            service.appendChild(displayItem);
            
            // now need a classifierList
            Element classifierList = doc.createElement(
                    GSXML.CLASSIFIER_ELEM+GSXML.LIST_MODIFIER);
            
            int classifierNum = 1;
            // append a <classifier content="some letter" name="CL#">
            // for each letter of the alphabet:
            Element classifier = createClassifierElement(doc, "TitleByLetter", 
                classifierNum++, "titles by letter", "Browse titles by letter");
            // now add this <classifier> to the <classifierList>
            classifierList.appendChild(classifier);
            
            // ANY MORE CLASSIFIERS? ADD THEM HERE
            
            service.appendChild(classifierList);
        } // ELSE check for whether it is a query service
        else if(serviceName.toLowerCase().contains("query")) {
            attribute.setValue(GSXML.SERVICE_TYPE_QUERY);
            if(serviceName.equals("TextQuery")) {
                describeTextQueryService(service);
            } else if(serviceName.equals("FieldQuery")) {
                describeFieldQueryService(service);
            }
        }
        
        // don't forget to add the type attribute to the service!
        service.setAttributeNode(attribute);
        
        String from = serviceName;
        
        Element responseMsg = createResponseMessage(doc, service, null,
                GSXML.REQUEST_TYPE_DESCRIBE, from);
        try{ 
            return FedoraCommons.elementToFormattedString(responseMsg);
        }catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    /** Appends children to the parameter service Element that make the 
     * final service Element into a describe response XML for FedoraGS3's 
     * TextQuery service.
     * @param service is the service Element that is being filled out. */
    protected void describeTextQueryService(Element service) {
        Document doc = service.getOwnerDocument();
        // we need name, submit (button) and description <displayItem> elements
        Element displayItem = createNameValuePairElement(doc, 
            GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, 
            "Text Search");
        service.appendChild(displayItem);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_SUBMIT, "Search");
        service.appendChild(displayItem);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_DESCRIPTION, 
                "Title and full-text search service");
        service.appendChild(displayItem);
        
        //create the <paramList> 
        Element paramList = doc.createElement(
            GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
        
        // we ignore granularity to search at: it will always be
        // document and section level
        // we ignore casefolding: always on (that is, case is irrelevant)
        // we ignore document display order: always ranked
        
        // Constructing the following:
        // <param default="100" name="maxDocs" type="integer">
        // <displayItem name="name">Maximum hits to return</displayItem>
        // </param>
        Element param = doc.createElement(GSXML.PARAM_ELEM);
        
        Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(MAXDOCS);
        param.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.DEFAULT_ATT);
        attribute.setValue("100");
        param.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.TYPE_ATT);
        attribute.setValue(GSXML.PARAM_TYPE_INTEGER);
        param.setAttributeNode(attribute);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, 
                "Maximum hits to return");
        param.appendChild(displayItem);
        
        paramList.appendChild(param);
        
        // Constructing the following:
        // <param name="query" type="string">
        // <displayItem name="name">Query string</displayItem>
        // </param> 
        param = doc.createElement(GSXML.PARAM_ELEM);
        
        attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(QUERY);
        param.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.TYPE_ATT);
        attribute.setValue(GSXML.PARAM_TYPE_STRING);
        param.setAttributeNode(attribute);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, 
                "Query string");
        param.appendChild(displayItem);
        
        paramList.appendChild(param);
        
        service.appendChild(paramList);
    }
    
    /** Appends children to the parameter service Element that make the 
     * final service Element into a describe response XML for FedoraGS3's 
     * FieldQuery service.
     * @param service is the service Element that is being filled out. */
    protected void describeFieldQueryService(Element service) {
        Document doc = service.getOwnerDocument();
        // we need name, submit (button) and description <displayItem> elements
        Element displayItem = createNameValuePairElement(doc, 
            GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, 
            "Form Search");
        service.appendChild(displayItem);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_SUBMIT, "Search");
        service.appendChild(displayItem);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_DESCRIPTION, 
                "Simple fielded search");
        service.appendChild(displayItem);
        
        //create the <paramList> 
        Element paramList = doc.createElement(
            GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
        
        // we ignore granularity to search at: it will always be
        // document and section level
        // we ignore casefolding: always on (that is, case is irrelevant)
        // we ignore document display order: always ranked
        
        // Constructing the following:
        // <param default="100" name="maxDocs" type="integer">
        // <displayItem name="name">Maximum hits to return</displayItem>
        // </param>
        Element param = doc.createElement(GSXML.PARAM_ELEM);
        
        Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(MAXDOCS);
        param.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.DEFAULT_ATT);
        attribute.setValue("100");
        param.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.TYPE_ATT);
        attribute.setValue(GSXML.PARAM_TYPE_INTEGER);
        param.setAttributeNode(attribute);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, 
                "Maximum hits to return");
        param.appendChild(displayItem);
        
        paramList.appendChild(param);
        
        // Constructing the following:
        // <param name="simpleField" occurs="4" type="multi">
        // <displayItem name="name"></displayItem>
        //
        // <param name="query" type="string">
        //   <displayItem name="name">Word or phrase </displayItem>
        // </param>
        //
        // <param default="allFields" name="fieldname" type="enum_single">
        //   <displayItem name="name">in field</displayItem>
        //   
        //   <option name="docTitles">
        //     <displayItem name="name">document titles</displayItem>
        //   </option>
        //   <option name="allTitles">
        //     <displayItem name="name">document and section titles</displayItem>
        //   </option>
        //   <option name="fullText">
        //     <displayItem name="name">full text</displayItem>
        //   </option>
        //   <option name="all">
        //     <displayItem name="name">titles and full text</displayItem>
        //   </option>
        //   <option name="">
        //     <displayItem name="name"></displayItem>
        //       </option>
        //  </param>
        // </param>
        Element rowOfParams = doc.createElement(GSXML.PARAM_ELEM);  
        attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(SIMPLEFIELD_ATT);
        rowOfParams.setAttributeNode(attribute);
        
        // we want the row of controls to occur multiple times
        attribute = doc.createAttribute(GSXML.TYPE_ATT);
        attribute.setValue(GSXML.PARAM_TYPE_MULTI);
        rowOfParams.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(OCCURS_ATT);
        attribute.setValue("4"); // we want this row to occur 4 times
        rowOfParams.setAttributeNode(attribute);
        
        // <param name="query" type="string">
        //   <displayItem name="name">Word or phrase </displayItem>
        // </param>
        param = doc.createElement(GSXML.PARAM_ELEM);
        
        attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(QUERY);
        param.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.TYPE_ATT);
        attribute.setValue(GSXML.PARAM_TYPE_STRING);
        param.setAttributeNode(attribute);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, 
                "Word or phrase");
        param.appendChild(displayItem);
        rowOfParams.appendChild(param);
        
        // <param default="allFields" name="fieldName" type="enum_single">
        //   <displayItem name="name">in field</displayItem>
        param = doc.createElement(GSXML.PARAM_ELEM);
        attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(FIELDNAME_ATT);
        param.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.TYPE_ATT);
        attribute.setValue(GSXML.PARAM_TYPE_ENUM_SINGLE);
        param.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.DEFAULT_ATT);
        attribute.setValue(ALL_FIELDS);
        param.setAttributeNode(attribute);
        
        displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, 
                "in field");
        param.appendChild(displayItem);
        
        String[] searchFieldNames 
            = {ALL_FIELDS, DOC_TITLES, ALL_TITLES, FULLTEXT};
        String[] searchFieldDisplay  = {"all titles and full-text", 
            "document titles only", "document and section titles", 
            "full-text only"};
        
        // for each fieldName create an option element and insert
        // the option into the enum_multi drop-down param:
        // <option name="fieldName">
        //   <displayItem name="name">fieldName</displayItem>
        // </option>
        for(int i = 0; i < searchFieldNames.length; i++) {
            Element option = doc.createElement(GSXML.PARAM_OPTION_ELEM);
            attribute = doc.createAttribute(GSXML.NAME_ATT);
            attribute.setValue(searchFieldNames[i]);
            option.setAttributeNode(attribute);
            
            displayItem = createNameValuePairElement(doc, 
                    GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, 
                    searchFieldDisplay[i]);
            option.appendChild(displayItem);
            param.appendChild(option); // add option to the drop-down box
        }
        
        rowOfParams.appendChild(param);
        paramList.appendChild(rowOfParams);
        service.appendChild(paramList);
    }
    
    /** 
     * @return a GS3 describe response message for the requested service 
     * of the given collection. DocumentContent/Metadata/StructureRetrieve 
     * return nothing special except their names; browse (and any query)
     * return more complex XML responses.
     * All collections in this Digital Library (Fedora Repository) share 
     * the same services, so this method returns the same as 
     * describeService(serviceName).   
     * @param collectionName - the name of the collection whose service is to
     * be described. It will be converted to a fedora collection pid, which is of 
     * the form "greenstone:&lt;collectionName&gt;-collection". 
     * @param serviceName - the name of the service in the collection which is to 
     * be described. */
    public String describeCollectionService(String collectionName, 
            String serviceName) {
        // collectionName can be ignored, because all services are FedoraGS3 
        // services and are not unique to any particular (greenstone) collection. 
        return describeService(serviceName);
    }
    
    /** This method performs the implemented browse operation: allowing the
     * user to browse the titles of documents in the given collection by letter
     * and returning the results.
     * @param classifierIDs are the ids of the classifiers on which to browse. In
     * this case, the classifier indicates whether we browse titles by letter, or
     * browse (documents) by collection; and it is of the form &lt;CL(letter)&gt;.
     * @param collectionName is the name of the collection whose documents 
     * starting with the given letter will be returned.
     * @return a GS3 DocumentStructureRetrieve response message which lists all 
     * the documents that start with the letter indicated by parameter classifier.
    */
    public String browse(String collectionName, String[] classifierIDs)
    {
        Document doc = builder.newDocument();
        FedoraGS3RunException ex = null; //any RemoteException or UnsupportedEncodingException
        
        // <classifierNodeList>
        Element classifierNodeList = doc.createElement(
                GSXML.CLASS_NODE_ELEM+GSXML.LIST_MODIFIER);
        
    for(int i = 0; i < classifierIDs.length; i++) {
        // <classifierNode nodeID="classifierNum">
        Element requestedClassifierNode = doc.createElement(
                GSXML.CLASS_NODE_ELEM);
        Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
        attribute.setValue(classifierIDs[i]);
        requestedClassifierNode.setAttributeNode(attribute);
        classifierNodeList.appendChild(requestedClassifierNode);
        
        // <nodeStructure>
        Element nodeStructure = doc.createElement(GSXML.NODE_STRUCTURE_ELEM);
        requestedClassifierNode.appendChild(nodeStructure);
        
        // And one more time, the top level classifierNode:
        Element classifierNode = doc.createElement(GSXML.CLASS_NODE_ELEM);
        attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
        attribute.setValue(classifierIDs[i]);
        classifierNode.setAttributeNode(attribute);
        nodeStructure.appendChild(classifierNode);
        
        // Work out what we're browsing base on the classifierID's number
        // classifier CL1 = browse titles by letter;
        // classifier CL2 = browse by collection;
        //remove the CL prefix and obtain the number from the id:
        int classifierNum = Integer.parseInt(classifierIDs[i].replace("CL", ""));
        switch(classifierNum) {
            case 1:
                // we're going to loop to the end of the alphabet
                int num = 1;
                for(char ch = 'A'; ch <= 'Z'; ch++, num++) {
                    // Retrieve the document structure for each subClassifierID: 
                    // all the documents that begin with its letter.
                    String letter = String.valueOf(ch);
                    try {
                    String[] docPIDs = this.browseTitlesByLetter(
                             collectionName, letter);
                    if(docPIDs.length == 0) {
                        continue; // skip letters that don't have any kids
                    }
                    
                    // <classifierNode nodeID="CL3.1">
                    Element subClassifier = doc.createElement(
                          GSXML.CLASS_NODE_ELEM);
                    attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
                    attribute.setValue(classifierIDs[i]+"."+num);
                    subClassifier.setAttributeNode(attribute);
                    classifierNode.appendChild(subClassifier);
                    
                    // append the <docNodes> for the docPIDs found as children 
                    // of subclassifier 
                    getStructureElement(subClassifier, docPIDs, DESCENDANTS);
                        //CHILDREN); // for testing
                    } catch(Exception e) { 
                    ex = new FedoraGS3RunException(e);
                    ex.setSpecifics("requested portion of TOC file or "
                            + "trouble with fielded search ");
                    }
                }
                // No titles in this collection that start with a letter at all 
                if(!classifierNode.hasChildNodes()) {
                    // <classifierNode nodeID="CL1.0"> which we will equate with A-Z
                    Element subClassifier = doc.createElement(
                            GSXML.CLASS_NODE_ELEM);
                    attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
                    attribute.setValue(classifierIDs[i]+"."+0); //
                    subClassifier.setAttributeNode(attribute);
                    classifierNode.appendChild(subClassifier);
                }
                break;
            case 2:
                break;
            default:
                ex = new FedoraGS3RunException( // cause is regular exception
                    new Exception("Unknown classifier ID: " + classifierIDs[i]));
        }
    }
        Element responseMsg = createResponseMessage(doc, classifierNodeList, ex,
                GSXML.REQUEST_TYPE_DESCRIBE, collectionName+"/ClassifierBrowse");
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        }catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    /** This method performs something equivalent to a greenstone3 
     * ClassifierBrowseMetadataRetrieve on the classifierNodeIDs
     * @param classNodeIDs array of classifierNode IDs of for which the metadata
     * needs to be returned.
     * @return a GS3 ClassifierBrowseMetadataRetrieve response message which 
     * lists the metadata for all the classifierNodes passed as parameter.*/
    public String browseMetadataRetrieve(String[] classNodeIDs)
    {
        Document doc = this.builder.newDocument();
        // <classifierNodeList>
        Element classifierNodeList = doc.createElement(
                GSXML.CLASS_NODE_ELEM+GSXML.LIST_MODIFIER);
        
        // create <classifierNode><metadataList><metadata>s
        // </metadataList></classifierNode> for all letters of alphabet
        for(int i = 0; i < classNodeIDs.length; i++) {
            // strip ID of everything before the first '.' (i.e. remove "CL#.")
            int index = classNodeIDs[i].indexOf('.');
            String subClassifierNumber = classNodeIDs[i].substring(index+1);
            int subClassifierNum = Integer.parseInt(subClassifierNumber);
            String classifierName = "";
            if(subClassifierNum == 0) { // no document titles started with a letter
                classifierName = "A-Z";
            } else {
                char letter = (char)('A' + subClassifierNum - 1); // A = 1
                classifierName = String.valueOf(letter);
            }
            
            // <classifierNode nodeID="CL#.subNum">
            Element classifierNode = doc.createElement(GSXML.CLASS_NODE_ELEM);
            Attr attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
            attribute.setValue(classNodeIDs[i]);
            classifierNode.setAttributeNode(attribute);
            
            // <metadataList>
            Element metadataList = doc.createElement(
                    GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
            
            // at least one metadata element: that of the title of this
            // classifierNode:
            // <metadata name="Title">letter</metadata>
            Element metadata = this.createNameValuePairElement(doc, 
                    GSXML.METADATA_ELEM, "Title", classifierName);
            
            // now connect up everything
            metadataList.appendChild(metadata);
            classifierNode.appendChild(metadataList);
            classifierNodeList.appendChild(classifierNode);
        }
        
        Element responseMsg = createResponseMessage(doc, classifierNodeList, null,
                GSXML.REQUEST_TYPE_PROCESS, //collName + 
                "ClassifierBrowseMetadataRetrieve");
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        }catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    /** @return a newly created element of the following format:
     * &lt;classifier content="somecontent" name="CL+num"&gt;
     *      &lt;displayItem name="name"&gt;someClassifierName&lt;/displayItem&gt;
     *      &lt;displayItem name="description"&gt;Browse by classifier name&lt;/displayItem&gt;
     * &lt;/classifier&gt;
     * @param doc - the document used to create the element
     * @param content - value of the content attribute
     * @param classifierNum - the number suffixed to the CL, together forming
     * the classifier Node's ID 
     * @param displayNameVal is the bodytext of a named displayItem element 
     * @param displayDescrVal is the bodytext of a displayItem element with
     * description */
    protected Element createClassifierElement(Document doc, String content, 
            int classifierNum, String displayNameVal, String displayDescrVal) 
    {
        final String CL = "CL";
        Element classifier = doc.createElement(GSXML.CLASSIFIER_ELEM);
        // content attribute
        Attr att = doc.createAttribute(GSXML.CLASSIFIER_CONTENT_ATT);
        att.setValue(content);
        classifier.setAttributeNode(att);
        // name attribute
        att = doc.createAttribute(GSXML.NAME_ATT);
        att.setValue(CL + classifierNum);
        classifier.setAttributeNode(att);
        
        // now create the displayItem children for classifier:
        // <displayItem name="name">#letter</displayItem>
        // <displayItem name="description">
        //Browse titles starting with #letter</displayItem>
        Element displayItem = createNameValuePairElement(doc, 
                GSXML.DISPLAY_TEXT_ELEM, GSXML.DISPLAY_TEXT_NAME, displayNameVal);
        classifier.appendChild(displayItem);
        displayItem = createNameValuePairElement(doc, GSXML.DISPLAY_TEXT_ELEM, 
                GSXML.DISPLAY_TEXT_DESCRIPTION, displayDescrVal);
        classifier.appendChild(displayItem);
        
        return classifier;
    }
    
        
    /** @return a newly created element of the following format:
     * &lt;elementName name="somename"&gt;"some display value"&lt;/elementName&gt;
     * @param doc - the document used to create the element
     * @param elementName - the tag name
     * @param name - value of attribute name
     * @param value - the body text of the element */
    protected Element createNameValuePairElement(Document doc, String elementName, 
            String name, String value) {
        // <elementName name="somename">"some display value"</elementName>
        Element element = doc.createElement(elementName);
        Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(name);
        element.setAttributeNode(attribute);
        
        element.appendChild(doc.createTextNode(value));
        return element;
    }
    
    /** 
     * @param collection is the collection to search in
     * @param query is the query term to search for. It won't specify the
     * indexed field to search in, which will mean that GSearch will
     * search all default indexed fields.
     * @param maxDocs is the maximum number of results to return (which
     * at present we consider equivalent to FedoraGSearch's hitpageSize).   
    */
    public String[] textQuery(String collection, String query, 
            int maxDocs) 
        throws Exception
    {
        // no need to search there is no query or query is empty spaces
        if(query.trim().equals(""))
            return new String[]{};
        
        // QUERY value won't specify indexed field to search, Fedora 
        // Gsearch will take that as meaning all default indexed fields. 
        // Params to search() method below: string of fielded query terms;  
        // hitpageStart, hitpageEnd, snippetsMax (leave that 0)
        query = query + " " + "PID" + COLON + GREENSTONE;
        
        String searchResult = this.fedoraGSearch.search(query, 1, maxDocs, 0);
        // now we have the XML returned by FedoraGSearch, get the pids
        // of the documents returned (if any)
        String[] pids = this.fedoraGSearch.getPIDsFromSearchResult(
                collection, searchResult);
        return pids;
    }
    
    /** 
     * This method performs a fieldquery, searching for x number of phrases
     * in each of the 4 indexed fields.  
     * @param collection is the collection to search in
     * @param nameValParamsMap is a Map of several(key, value) entries,
     * 4 of which we're concerned with here:
     * - the keys are ALL_FIELDS, DOC_TITLES, ALL_TITLES, FULLTEXT
     * - the values are a comma separated list of terms (phrases or single
     * words) to search that field in. There may be more than 1 or
     * there may be none (in which case there may be N empty values or
     * spaces separated by commas).
     * @param maxDocs is the maximum number of results to return (which
     * at present we consider equivalent to FedoraGSearch's hitpageSize).   
     * */
    public String[] fieldQuery(String collection, Map nameValParamsMap, 
            int maxDocs) 
        throws Exception
    {
        // we're going to maintain a list of UNIQUE pids that were returned
        // in search results. Hence we use Set:
        java.util.Set set = new java.util.HashSet();
        
        // (1) Use Fedora's search to search document titles, if they were
        // specified:
        String[] docTitlepids = {};
        
        String docTitleTerms = (String)nameValParamsMap.get(DOC_TITLES);
        if(docTitleTerms != null) { // no doc titles may have been specified
            String[] phrases = docTitleTerms.split(COMMA);
            
            // search the individual phrases first:
            for(int i = 0; i < phrases.length; i++) {
                if(phrases.equals("") || phrases.equals(" "))
                    continue; //skip when there are no terms
                docTitlepids = this.searchDocumentTitles(
                        collection, phrases[i], false);
                for(int j = 0; j < docTitlepids.length; j++)
                    set.add(docTitlepids[j]);
            }
        }
        // (2) use FedoraGSearch to search doc AND section titles, and 
        // fulltext (in case these were specified in nameValParamsMap):
        String searchResult = this.fedoraGSearch.search(
                nameValParamsMap, 1, maxDocs);
        
        String[] pids = this.fedoraGSearch.getPIDsFromSearchResult(
                collection, searchResult);
        
        for(int i = 0; i < pids.length; i++)
            set.add(pids[i]);
        
        pids = null;
        pids = new String[set.size()];
        set.toArray(pids); // unique pids
        return pids;
    }
    
    /** @return a String representing Greenstone3 XML for a query process
     * response returning the results for the query denoted by parameter
     * nameValParamsMap.
     * @param nameValParamsMap is a Hashmap of name and value pairs for all the
     * query field data values. The names match the field names that
     * describeCollectionService() would have returned for the query service. 
     * @param collection is the name of the collection 
     * @param service is the name of the query service
     * This method is only ever called when any of the services in the digital 
     * library described themselves as type=query. Therefore any digital
     * libraries that have no query services, can just return emtpy message 
     * strings (or even "") since this method will never be called on them 
     * anyway. */
    public String query(String collection, String service, 
            Map nameValParamsMap) 
    {
        FedoraGS3RunException ex = null; 
        // (1) obtain the requested number of maximum result documents
        int maxDocs = 100;
        try{
            maxDocs = Integer.parseInt((String)nameValParamsMap.get(MAXDOCS));
        } catch(NumberFormatException e) {
            maxDocs = 100;
        } 
        
        String pids[] = {};
        // (2) for Textquery, we simply search ALL_FIELDS using FedoraGSearch
        if(service.equals("TextQuery")) {
            try {
                // get the Query field:
                String query = (String)nameValParamsMap.get(QUERY);
                pids = textQuery(collection, query, maxDocs);
            }
            catch(Exception e) {
                LOG.error("Error in TextQuery processing: " + e);
                ex = new FedoraGS3RunException(
                    "When trying to use FedoraGenericSearch for a TextQuery", e);
                
            }
        } else { // (3) FieldQuery
            // first get the comma-separated lists
            String listOfFieldNames = (String)nameValParamsMap.get(FIELDNAME_ATT);
            String listOfSearchTerms = (String)nameValParamsMap.get(QUERY);
            // both are comma separated lists, so split both on 'comma'
            String[] fieldNames = listOfFieldNames.split(COMMA);
            String[] searchTerms = listOfSearchTerms.split(COMMA);
            
            // In the fieldNames and searchTerms lists of nameValParamsMap, 
            // each searchTerm element was matched with its correspondingly
            // indexed fieldName. 
            // A new map is going to reorganise this, by putting all terms
            // for a particular fieldName together in a comma separated list
            // and associating that with the fieldName. I.e. (key, value) ->
            // (fieldName, comma-separated list of all terms in that field)
            Map map = new HashMap();
            for(int i = 0; i < searchTerms.length; i++) {
                // there may be fewer searchTerms than fieldNames (since some
                // fieldNames may have been left empty), so loop on searchTerms
                if(map.containsKey(fieldNames[i])) { // fieldName is already 
                    // in the list, so append comma with new value
                    String termsList = (String)map.get(fieldNames[i]);
                    termsList = termsList + COMMA + searchTerms[i];
                    map.put(fieldNames[i], termsList);
                } else { // this is the first time this fieldName occurred
                    // just put the fieldName with searchTerm as-is 
                    map.put(fieldNames[i], searchTerms[i]);
                }
            }
            
            try {
                // For fieldquery, we search on all the fieldNames specified 
                // - if DOC_TITLES is specified then we use Fedora's search
                // - for all other fieldNames specified, we use FedoraGSearch
                pids = fieldQuery(collection, map, maxDocs);
            }
            catch(Exception e) {
                LOG.error("Error in FieldQuery processing: " + e);
                ex = new FedoraGS3RunException(
                    "When trying to use FedoraGenericSearch for a FieldQuery", e);
            }
        }
        
        // Build Greenstone XML Query response message for from 
        // the pids (which should be document identifiers)
        Document doc = builder.newDocument();
        // <metadataList><metadata name="numDocsMatched" value="n" />
        // </metadataList>
        Element metadataList = doc.createElement(
                GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
        Element metadata = doc.createElement(GSXML.METADATA_ELEM);
        
        Attr attribute = doc.createAttribute(GSXML.NAME_ATT);
        attribute.setValue(NUM_DOCS_MATCHED);
        metadata.setAttributeNode(attribute);
        
        attribute = doc.createAttribute(GSXML.VALUE_ATT);
        attribute.setValue(Integer.toString(pids.length));
        metadata.setAttributeNode(attribute);
        
        metadataList.appendChild(metadata);
        
        // <documentNodeList>
        // <documentNode nodeID="HASHac0a04dd14571c60d7fbfd.4.2"
        // docType='hierarchy' nodeType="leaf" />
        // ...
        // ...
        // </documentNodeList>
        Element docNodeList = doc.createElement(
                GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
        // for each 
        for(int i = 0; i < pids.length; i++) {
            Element docNode = doc.createElement(GSXML.DOC_NODE_ELEM);
            attribute = doc.createAttribute(GSXML.NODE_ID_ATT);
            attribute.setValue(pids[i]);
            docNode.setAttributeNode(attribute);
            
            attribute = doc.createAttribute(GSXML.DOC_TYPE_ATT);
            attribute.setValue("hierarchy");
            docNode.setAttributeNode(attribute);
            
            attribute = doc.createAttribute(GSXML.NODE_TYPE_ATT);
            attribute.setValue("root");
            docNode.setAttributeNode(attribute);
            docNodeList.appendChild(docNode);
        }
        
        Element responseMsg = createResponseMessage(doc, docNodeList, ex,
                GSXML.REQUEST_TYPE_PROCESS, service);
        try{
            return FedoraCommons.elementToFormattedString(responseMsg);
        }catch(TransformerException e) {
            return FedoraGS3RunException.xmlToStringConversionFailureResponseMsg
                + " " + e;
        }
    }
    
    public static void main(String args[]) {
        try{
            // testing default constructor
            //FedoraGS3Connection con = new FedoraGS3Connection();
            
            // testing constructor that takes properties file to show initial
            // fedora server values
            java.io.File propertyFilename 
                = new java.io.File("fedoraGS3.properties");
            FedoraGS3Connection con = new FedoraGS3Connection(propertyFilename);
            
            // DESCRIBE: serviceList, collectionList
            System.out.println("serviceList:\n" + con.getServiceList());
            
            System.out.println("collectionList:\n" + con.getCollectionList());
            
            String[] colPIDs = con.getCollections();
            String[] collectionNames = con.getCollectionNames(con.getCollections());
            
            
            for(int i = 0; i < collectionNames.length; i++) {
                System.out.println("Describing collections:\n");
                System.out.println(con.describeCollection(collectionNames[i]));
                System.out.println("Describing collection services:\n" 
                    + con.describeCollectionServices(collectionNames[i]));
            }
            
            String[] serviceNames = con.getServiceNames();
            for(int i = 0; i < serviceNames.length; i++) {
                System.out.println("Describing " + serviceNames[i] + ":\n"
                    + con.describeCollectionService("demo", serviceNames[i]));
            }
            
                        
            // TRYING OUT SPECIAL top-level document metadata retrieval (DLS, DC)
            // along with EX of the top-level document:
            System.out.println("\nGET META for greenstone:gs2mgdemo-HASH01d667303fe98545f03c14ae:");
            System.out.println(con.getDocumentMetadata(new String[]{"greenstone:gs2mgdemo-HASH01d667303fe98545f03c14ae"}));
            
                        
            String[] docIDs = con.getCollectionDocs(colPIDs[0]);
            System.out.println("\nGET CONTENT:");
            for(int i = 0; i < docIDs.length; i++) {
                System.out.println(con.getContent(docIDs[i]));
            }
            
            System.out.println("\nGET META:");
            for(int i = 0; i < docIDs.length; i++) {
                System.out.println(con.getDocumentMetadata(docIDs[i]));
            }
            
            String[] getTitlesFor = {
                    "greenstone:gs2mgdemo-HASH01d667303fe98545f03c14ae",
                    "greenstone:gs2mgdemo-HASHa568bac1d8d7bd12a0938b",
                    "greenstone:gs2mgdemo-HASHa568bac1d8d7bd12a0938b-1",
                    "greenstone:gs2mgdemo-HASH01d667303fe98545f03c14ae-1.7",
                    "greenstone:gs2mgdemo-HASHa568bac1d8d7bd12a0938b-1.5.1"
            };
            
            // first let's display the regular meta for top-level docs and 
            // their sections
            for(int i = 0; i < getTitlesFor.length; i++) {
                System.out.println(con.getDocumentMetadata(getTitlesFor[i]));
            }
            
            System.out.println("\nTitles are:");
            System.out.println(con.getTitleMetadata(getTitlesFor));
            
            System.out.println("\nGET STRUCTURE:");
            for(int i = 0; i < docIDs.length; i++) {
                System.out.println(con.getChildren(docIDs[i]));
                System.out.println(con.getDocumentStructure(docIDs[i]));
            }
            
            // TEST ERROR CASES:
            System.out.println("\nTESTING ERROR CASES");
            System.out.println(con.getContent("greenstone:demo-pinky"));
            String[] errorCases = { "greenstone:demo-HASH23d1019b589e2ef6a680e3-1.5.1.5", 
            "greenstone:demo-pinky" };
            System.out.println(con.getContent(errorCases));
            System.out.println(con.getDocumentMetadata(errorCases));
            System.out.println(con.getDocumentStructure(errorCases));
            
            System.out.println("\nCLASSIFIER BROWSE");
            System.out.println(con.browse("gs2mgdemo", //"ClassifierBrowse", 
                              new String[]{"CL1"}));
            
            System.out.println("\nCLASSIFIER BROWSE METADATA RETRIEVE");
            String[] classNodeIDs = new String[26];
            for(int i = 0; i < classNodeIDs.length; i++) {
                int subClassifierNum = i + 1;
                classNodeIDs[i] = "CL1." + subClassifierNum;
            }
            System.out.println(con.browseMetadataRetrieve(//"gs2mgdemo", 
                    classNodeIDs));
            
            System.out.println("Testing query services");
            System.out.println("TEXT QUERY:");
            Map formControlValsMap = new HashMap();
            formControlValsMap.put(MAXDOCS, "100");
            formControlValsMap.put(QUERY, "snails");
            String searchResponse 
                = con.query("gs2mgdemo", "TextQuery", formControlValsMap);
            System.out.println(searchResponse);
            
            System.out.println("FIELD QUERY:");
            formControlValsMap.clear();
            formControlValsMap.put(MAXDOCS, "100");
            formControlValsMap.put(QUERY, "interview,Gender equality,cyclone");
            formControlValsMap.put(FIELDNAME_ATT, 
                    "allFields,docTitles,allFields,allFields");
            searchResponse 
                = con.query("gs2mgdemo", "FieldQuery", formControlValsMap);
            System.out.println(searchResponse);
            
            System.exit(0);
        }catch(Exception e) {
            JOptionPane.showMessageDialog(
                    null, e, "Error", JOptionPane.ERROR_MESSAGE);
            //System.err.println("ERROR: " + e);
            e.printStackTrace();
        }
    }
}