source: trunk/greenstone3-extensions/gs3build/src/org/greenstone/gsdl3/gs3build/classifier/HierarchyClassifier.java@ 12188

package org.greenstone.gsdl3.gs3build.classifier;

import java.io.*;

import java.net.URL;

import java.util.List;
import java.util.ArrayList;
import java.util.Iterator;

import org.apache.xerces.parsers.SAXParser;
import org.xml.sax.XMLReader;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
import org.xml.sax.Attributes;
import org.xml.sax.helpers.XMLReaderFactory;
import org.xml.sax.helpers.DefaultHandler;
import org.w3c.dom.Element;
import org.greenstone.gsdl3.gs3build.doctypes.DocumentID;
import org.greenstone.gsdl3.gs3build.doctypes.DocumentInterface;
import org.greenstone.gsdl3.gs3build.util.XMLTools;
import org.greenstone.gsdl3.gs3build.database.GS3SQLConnection;
import org.greenstone.gsdl3.util.GSXML;


public class HierarchyClassifier extends AbstractClassifier
{ 
  class HierarchyClassifierObserver implements ClassifierObserverInterface
  { List              classifications;
    DocumentInterface document;

    public HierarchyClassifierObserver(DocumentInterface document)
    { this.document = document;
    }

    public void recordClassification(String label)
    { this.document.addDocumentMetadata("gsdl3", "classified", label);
    }

    public List getClassifications()
    { return this.classifications;
    }
  }

  class HierarchyNode extends AbstractHierarchyNode
  { /**
     *  Simple node
     */
    public HierarchyNode()
    { super();
    }

    public HierarchyNode(String prefix, String name, String id, String descriptor)
    { super(prefix, name, id, descriptor);
    }
    
    public void getClassifications(DocumentID documentID, List values, String sortKey,
                   ClassifierObserverInterface observer)
    { Iterator valueList = values.iterator();

      while (valueList.hasNext())
      { if (this.isMatch(valueList.next().toString()))
    { observer.recordClassification(this.name);
      this.addDocument(documentID, sortKey);
    }
      }
     
      // recurse into the child nodes for them to do the same...
      Iterator childList = this.childNodes.iterator();

      while (childList.hasNext())
      { ((HierarchyNode) childList.next()).getClassifications(documentID, values, sortKey, observer);
      }
    }
  }

  /**
   *  An inner class to handle Classifier files
   */
  class HierarchyHandler extends DefaultHandler
  { StringBuffer description;
    StringBuffer name;
    StringBuffer path;
    StringBuffer match;
    String       prefix;
    boolean inElement;
    AbstractHierarchyNode rootNode = null;
    AbstractHierarchyNode currentNode = null;

    HierarchyHandler(String classifierName)
    { super();
 
      this.name        = null;
      this.description = null;
      this.path        = null;
      this.match       = null;
      this.prefix      = classifierName;
    }

    public void startElement(String URI, String localName, String qName, Attributes attributes)
    { if (localName.equals("Hierarchy"))
      { // create a new node in the hierarchy
    HierarchyNode node = new HierarchyNode(prefix, "", "", "Classifier");

    this.rootNode = node;
    this.currentNode = node;
      }
      else if (localName.equals("Classification"))
      { // create a new node in the hierarchy
    HierarchyNode node = new HierarchyNode(this.prefix, "", "", "");

    // Initialise the tree
    if (this.currentNode == null)
    { this.rootNode = node;
    }
        // Add the node
    else
    { this.currentNode.addChild(node);
    }
    this.currentNode = node;
      }
      else if (localName.equals("Name"))
      { this.name = new StringBuffer();
      }
      else if (localName.equals("Path"))
      { this.path   = new StringBuffer();
      }
      else if (localName.equals("Description"))
      { this.description = new StringBuffer();
      }
      else if (localName.equals("Matches"))
      { this.match = new StringBuffer();
      }
      else if (localName.equals("Children"))
      { 
      }
    }

    public void endElement(String URI, String localName, String qName)
    { if (localName.equals("Classification"))
      { if (this.currentNode.getParent() == null) {
      // TODO: Error state
          // this.rootNodes.add(this.currentNode);
        }
    this.currentNode = this.currentNode.getParent();
      }
      else if (localName.equals("Name"))
      { this.currentNode.setName(XMLTools.cleanString(this.name.toString()));
      System.out.println("Name is " + this.name);
        this.name = null;
      }
      else if (localName.equals("Path"))
      { this.currentNode.setID(XMLTools.cleanString(this.path.toString()));
        this.path = null;
      }
      else if (localName.equals("Description"))
      { this.currentNode.setDescriptor(XMLTools.cleanString(this.description.toString()));
        this.description = null;
      }
      else if (localName.equals("Matches"))
      { this.currentNode.addMatch(XMLTools.cleanString(this.match.toString()));
        this.match = null;
      }
      else if (localName.equals("Children"))
      { 
      }
    }

    public void characters(char c[], int start, int length)
    { String string = new String(c, start, length);
      if (this.path != null)
      { this.path.append(string);
      }
      else if (this.name != null)
      { this.name.append(string);
      }
      else if (this.description != null)
      { this.description.append(string);
      }
      else if (this.match != null)
      { this.match.append(string);
      }
    }

    public AbstractHierarchyNode getHierarchy()
    { return this.rootNode;
    }
  }
  
  private GS3SQLConnection database;
  private AbstractHierarchyNode    hierarchy;
  private List             fields;
  private String           sortBy;
  private String           name;

  public HierarchyClassifier(String name, URL basefile, List fields, String sortBy)
  { 
    try 
    { SAXParser parser = new SAXParser();
      HierarchyHandler handler = new HierarchyHandler(name);
      /*
      XMLReader reader = XMLReaderFactory.createXMLReader();
      reader.setContentHandler(handler);
      reader.setErrorHandler(handler);*/
      parser.setContentHandler(handler);

      // A Classifier consists of one file only - get it.
      /*
      FileReader fileReader = new FileReader(basefile);
      reader.parse(new InputSource(fileReader));
      */
      parser.parse(basefile.toString());

      this.hierarchy = handler.getHierarchy();
      this.fields = fields;
      this.sortBy = sortBy;
      this.name = name;
      System.out.println("Counting " + this.fields.size() + " fields");
    }
    catch (SAXException saxException)
    { // TODO: log error
    System.out.println("unable to process hierarchy " + saxException.toString());
    }
    catch (java.io.FileNotFoundException fileException)
    {
    System.out.println("unable to process hierarchy " + fileException.toString());
    }
    catch (java.io.IOException ioException)
    {
    System.out.println("unable to process hierarchy " + ioException.toString());
    }
  }

  public void configure(List parameters)
  {
    Iterator iterator = parameters.iterator();
    while (iterator.hasNext()) {
      String param = iterator.next().toString();
      if (param.equals("-metadata")) {
    if (iterator.hasNext()) {
      String field = iterator.next().toString();
      if (field != null && field.length() > 0) {
        this.fields.add(field);
      }
    }
      }
    }
  }

  /**
   *  Indicate the database to be used for saving the hierarchy to.
   *
   *  @param <code>GS3SQLConnection</code> the database
   */
  public void setDatabase(GS3SQLConnection database)
  { this.database = database;
  }

  /**
   *  Classify a particular document - the document id will be used to record its
   *  membership of any classifier nodes, and the document interface will also be
   *  used to assign back knowledge of any classifiers to which the document is
   *  assigned.
   *
   *  @param <code>DocumentID</code> the document identifier
   *  @param <code>DocumentInterface</code> the document interface
   *
   *  @return <code>boolean</code> a return value - currently always 
   *          <code>true</code> for success
   */
  public boolean classifyDocument(DocumentID documentID, DocumentInterface document)
  { // the observer records the assignment of documents to classifications
    HierarchyClassifierObserver classifyObserver = new HierarchyClassifierObserver(document);

    // if we have no fields, there is nothing to do...
    if (this.fields == null)
    { return true;
    }

    // get the metadata item from the document
    Iterator thisField = this.fields.iterator();

    // for every field listed as being a scannable field...
    while (thisField.hasNext())
    { String fieldName = thisField.next().toString();
            
      // ...get the values for that field...
      List values = document.getDocumentMetadataItem(fieldName);

      if (values == null)
      { continue;
      }

      if (documentID == null) {
    System.out.println("Bad documentID");
    continue;
      }

      // get the sort key for the metadata item if possible
      String sortKey = null;

      if (this.sortBy != null) {
    List sortKeys = document.getDocumentMetadataItem(this.sortBy);
    if (sortKeys != null && sortKeys.size() > 0) {
      sortKey = sortKeys.get(0).toString();
    }
      }

      // ...and send them to the classifier
      this.hierarchy.getClassifications(documentID, values, sortKey, classifyObserver);
    }
    return true;
  }

  /**
   *  @see ClassifierInterface:getClassifierPasses
   */
  public int getClassifierPasses()
  { return 1;
  }

  public void startClassifierPass(int pass)
  {
  }
  
  public void endClassifierPass(int pass)
  {
  }
  
  public void completeClassification()
  { // TODO: store the classifications to file...
    if (this.hierarchy != null) {
      this.hierarchy.writeSQL(this.database);
    }
  }

    public boolean addClassifierDescription(Element classifier_list) {
    // TODO check that there are some docs in the classification
    Element classifier = classifier_list.getOwnerDocument().createElement(GSXML.CLASSIFIER_ELEM);
    classifier.setAttribute(GSXML.NAME_ATT, this.name);
    classifier_list.appendChild(classifier);
    return true;
    }
}