source: trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/AbstractDocument.java@ 9874

package org.greenstone.gsdl3.gs3build.doctypes;

import java.util.List;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.HashMap;
import java.util.Map;
import java.util.Date;

import java.sql.SQLException;
import java.sql.Statement;
import java.sql.ResultSet;
import java.sql.Timestamp;

import java.net.URL;

import org.greenstone.gsdl3.gs3build.metadata.NamespaceFactory;
import org.greenstone.gsdl3.gs3build.metadata.StructureIdentifierFactory;
import org.greenstone.gsdl3.gs3build.metadata.GSDL3Namespace;
import org.greenstone.gsdl3.gs3build.metadata.METSDescriptiveSet;
import org.greenstone.gsdl3.gs3build.metadata.METSFile;
import org.greenstone.gsdl3.gs3build.metadata.METSFileSet;
import org.greenstone.gsdl3.gs3build.metadata.METSHeader;
import org.greenstone.gsdl3.gs3build.metadata.METSStructure;
import org.greenstone.gsdl3.gs3build.metadata.METSStructureSet;
import org.greenstone.gsdl3.gs3build.metadata.METSDivision;
import org.greenstone.gsdl3.gs3build.metadata.METSNamespace;
import org.greenstone.gsdl3.gs3build.metadata.MetadataLabel;

import org.greenstone.gsdl3.gs3build.util.MultiMap;
import org.greenstone.gsdl3.gs3build.database.GS3SQLConnection;

/**
 *  Provide a base-line functionality for the <code>DocumentInterface</code>
 *  class.
 */

public abstract class AbstractDocument implements DocumentInterface
{
  METSFileSet        fileSet;
  METSDescriptiveSet metadata;
  METSStructureSet   structureSet;
  METSHeader         header;
  DocumentID         id;
  boolean            isModified;
  StructureIdentifierFactory structureIdFactory;
  java.sql.Timestamp firstDate;
  java.sql.Timestamp indexDate;
  java.sql.Timestamp modifiedDate;

  /**
   *  <p>Create a very vanilla document with a given document identifier.</p>
   *  <p>Most commonly used in dealing with loading files using DocumentFactory
   *  or similar.</p>
   *
   *  @param <code>DocumentID</code> the document identifier
   */
  public AbstractDocument(DocumentID id)
  { this.fileSet      = new METSFileSet();
    this.metadata     = new METSDescriptiveSet();
    this.header       = new METSHeader();
    this.structureSet = new METSStructureSet();
    this.id           = id;
    this.structureIdFactory = new StructureIdentifierFactory();

    java.util.Date thisDate = new java.util.Date();
    this.firstDate = new java.sql.Timestamp(thisDate.getTime());
    this.indexDate = new java.sql.Timestamp(thisDate.getTime());
    this.modifiedDate = null; // as a signature that the modified date needs finding...
  }
  
  /**
   *  Create a basic document from a given <code>URL</code.  This is usually the form
   *  called through the recognisers.
   *
   *  @param <code>URL</code> the URL of the first file in the document package
   */
  public AbstractDocument(URL url)
  { this.fileSet      = new METSFileSet();
    METSFile metsFile = this.fileSet.addFile(url);
    this.metadata     = new METSDescriptiveSet();
    this.header       = new METSHeader();
    this.structureSet = new METSStructureSet();
    this.id           = null;

    java.util.Date thisDate = new java.util.Date();
    this.firstDate = new java.sql.Timestamp(thisDate.getTime());
    this.indexDate = new java.sql.Timestamp(thisDate.getTime());

    METSStructure structure = new METSStructure("All", "All", "Whole Document");
    METSDivision documentBody = new METSDivision("All", "All", "All", "Whole Document", "Document");
    structure.addDivision(documentBody);
    this.structureSet.addStructure(structure);
    documentBody.addFileReference(metsFile.getID());
    documentBody.addMetadataReference("DM1");
  }

  /**
   *  Set the identified for the document.  Every document should have
   *  a document number set on its accession, either through metadata
   *  placed upon it internally or externally, or by assignment through
   *  a <code>DocumentIDFactory</code>.  Each identifier should be
   *  unique.
   *
   *  @param <code>DocumentID</code> the document identifier - in XML
   *         terms, the gsdl3:id element.
   */
  public void setID(DocumentID id)
  { this.id = id;
    this.isModified = true;
  }

  /**
   *  Get the document identifier - this should be unique to the document,
   *  but care must be taken in the configuration of the collection to
   *  ensure that this is the case.
   *
   *  @return <code>DocumentID</code> the identifer
   */
  public DocumentID getID()
  { return this.id;
  }

  /**
   *  Indicate whether this document is indexed.
   *
   *  @see: DocumentInterface.isIndexed
   */
  public boolean isIndexed()
  { return true;
  }

  /**
   *  Get the date that this file was modified
   */
  public long getFilesDatestamp()
  { return this.fileSet.getModifiedDatestamp(); 
  }

  /**
   *  Get the date that this file was modified
   */
  public long getModifiedDatestamp()
  { if (this.modifiedDate == null) {
      this.setModifiedDatestamp();
    }
    return this.modifiedDate.getTime();
  }

  /**
   *  Update/set the date of the most recent file modification
   */
  public void setModifiedDatestamp()
  { this.modifiedDate = new java.sql.Timestamp(this.fileSet.getModifiedDatestamp()); 
  }

  /**
   *  Get the date that this document was first indexed
   */
  public long getAccessionDate()
  { return this.firstDate.getTime();
  }

  /**
   *  Get the date that this document was last indexed
   */
  public long getLastIndexedDate()
  { return this.indexDate.getTime();
  }

  /**
   *  Set the last indexed date for this document;
   */
  public void setLastIndexedDate(long timestamp)
  { this.indexDate = new java.sql.Timestamp(timestamp);
  }

  /**
   *  Check if this document is in the database already.
   *
   *  In this simple implementation, the first file in the document's "default"
   *  filegroup is taken to be the canonical file for this document - any document
   *  of the same type with the same canonical file is taken to be a match.
   *
   *  @return <code>boolean</code> - if a matching document is found in the
   *                                 database.
   */
  public String getDuplicateID(GS3SQLConnection connection)
  { //String query = "SELECT * FROM document INNER JOIN filegroups ON document.docId=filegroups.docId WHERE DocType=\"" + HTML_DOCUMENT_TYPE + "\"";

    // Query for documents using the same file...
    String query = "SELECT DocID FROM files INNER JOIN filegroups ON files.FileGroupRef=filegroups.FileGroupRef WHERE (filegroups.FileGroupId=\"default\" AND files.FileLocation=\"" + this.fileSet.getFile(0).getLocation().toString() + "\")";
    try {
    Statement statement = connection.createStatement();
    ResultSet results = statement.executeQuery(query);

    List docs = new ArrayList();

    if (results.first()) {
        do {
        String value = results.getString("DocID");      
        docs.add(value);
        } while (results.next());
        
        Iterator docIterator = docs.iterator();
        while (docIterator.hasNext()) {
        String docId = docIterator.next().toString();
        String innerQuery = "SELECT * FROM document WHERE DocID=\"" + docId + "\"";
        results = statement.executeQuery(innerQuery);
        if (results.first()) {
            String docType = results.getString("DocType");
            if (docType.equals(this.getDocumentType())) {
            return docId;
            }
        }
        }
    }
    statement.close();
    }
    catch (java.sql.SQLException sqlEx) {
      System.err.println("AbstractDocument.getDuplicateID(): "+sqlEx);
    }

    return "";
  }

  /**
   *  Obtain the METS header of this document
   *
   *  @return <code>METSHeader</code> the header
   */
  public METSHeader getHeader()
  { return this.header;
  } 

  /**
   *  Set the METS header for this document.
   *
   *  @param <code>METSHeader</code> the header
   */
  public void setHeader(METSHeader header)
  { this.header = header;
  }

  /**
   *  A simple implementation of the isDocumentType function that does <b>not</b> consider
   *  inheritance - it <code>must</code> be extended as required.
   */
  public boolean isDocumentType(String type)
  { return type.equals(this.getDocumentType());
  }

  public abstract String getDocumentType();
  
  public abstract String getDocumentText();

  public abstract String getSectionText(String sectionId);

  public String getMETSType()
  { return "document";
  }

  /**
   *   @see DocumentInterface:addDocumentMetadata
   */
  public void addDocumentMetadata(MetadataLabel label, String value)
  { // no need to set isModified, as the following call will do it anyway!
    this.addDocumentMetadata(label.getNamespace(), label.getLabel(), value);
  }

  /**
   *   @see DocumentInterface:addDocumentMetadata
   */
  public void setDocumentMetadata(MetadataLabel label, String value)
  { // no need to set isModified, as the following call will do it anyway!
    this.setDocumentMetadata(label.getNamespace(), label.getLabel(), value);
  }

  /**
   *   @see DocumentInterface:addDocumentMetadata
   */
  public void addDocumentMetadata(String namespace, String label, String value)
  { this.metadata.addMetadata("default", namespace, label, value);
    this.isModified = true;
  }

  /**
   *   @see DocumentInterface:addDocumentMetadata
   */
  public void setDocumentMetadata(String namespace, String label, String value)
  { this.metadata.setMetadata("default", namespace, label, value);
    this.isModified = true;
  }

  /**
   *  @see DocumentInterface:removeDocumentMetadata
   */
  public void removeDocumentMetadata(String namespace, String label)
  { this.metadata.removeMetadata("default", namespace, label);
    this.isModified = true;
  }

  /**
   *
   */
  public void removeAllMetadata(String namespace, String label)
  { this.metadata.removeAllMetadata(namespace, label);
    this.isModified = true;
  }

  /**
   *  Post metadata to a file in this document - the appropriate changes
   *  should be made...
   */
  public void postFileMetadata(URL fileLocation, String namespace, String label, String value)
  {
    // First get the list of file groups, etc. that this file is associated with...
    List fileGroups = this.fileSet.findGroups(fileLocation);

    // Next, get the METS divisions associated with each file group...
    List divisions = this.structureSet.findDivisionsForFiles(fileGroups);
    
    // Finally, post the metadata to the metadata group associated with each structure
    Iterator divisionIter = divisions.iterator();
    while (divisionIter.hasNext())
    { METSDivision division = (METSDivision) divisionIter.next();

      // get the open namespace for this division
      METSNamespace namespaceMetadata = division.findNamespace(namespace, true, this.metadata);

      // then post the metadata to it...
      namespaceMetadata.addMetadata(label, value);
    }
  }

  /**
   *  Get the metadata structure of the document
   *
   *  @return <code>METSDescriptive</code> the metadata holder for the document.
   */
  public METSDescriptiveSet getDocumentMetadata()
  { return this.metadata;
  }

  /**
   *  Set the metadata structure for this document
   *
   *  @param <code>METSDescriptive</code> the new metadata holder for the document.
   */
  public void setDocumentMetadata(METSDescriptiveSet metadata)
  { this.metadata = metadata;
    this.isModified = true;
  }

  /**
   *  Get the metadata structure of the document
   *
   *  @return <code>METSStructureSet</code> the metadata holder for the document.
   */
  public METSStructureSet getDocumentStructure()
  { return this.structureSet;
  }

  public void setDocumentStructure(METSStructureSet structureSet)
  { this.structureSet = structureSet;
  }

  /**
   *  Get the values associated with a particular metadata value.
   *
   *  @param <code>String</code> the namespace to find the values in.
   *  @param <code>String</code> the label to match to find the values.
   *
   *  @return <code>List</code> the values.
   */
  public List getDocumentMetadataItem(String namespace, String label)
  { return this.metadata.getMetadata("default", namespace, label);
  }

  /**
   *  Get the values associated with a particular metadata value.
   *
   *  @param <code>String</code> the namespace and label separated by a
   *         colon.
   *
   *  @return <code>List</code> the values.
   */
  public List getDocumentMetadataItem(String namespaceLabel)
  { String namespace, label;
  
    int colonAt = namespaceLabel.indexOf(':');
    if (colonAt < 0)
    { namespace = GSDL3Namespace.GSDL3_NAMESPACE_ID;
      label     = namespaceLabel;
    }
    else
    { namespace = namespaceLabel.substring(0, colonAt);
      label     = namespaceLabel.substring(colonAt+1);
    }
    return this.metadata.getMetadata("default", namespace, label);
  }

  /**
   *  @see DocumentInterface:getDocumentFiles
   */
  public METSFileSet getDocumentFiles()
  { return this.fileSet;
  }

  public void setDocumentFiles(METSFileSet fileSet)
  { this.fileSet = fileSet;
  }

  /**
   *  This is just a dummy function that does nothing at this level...
   */
  public org.w3c.dom.Document getDOMDocument()
  { return null;
  }
  
  /**
   *  @see DocumentInterface:isMETSCompatible
   */
  public boolean isMETSCompatible()
  { return true;
  }

  /**
   *  Use a default document writer - this may be overridden for subclasses...
   *
   *  @see DocumentInterface:writeMETSObject
   */
  public DocumentWriter getMETSWriter()
  { return new DocumentWriter();
  }

  /**
   *  Use a default SQL document writer - this may be overridden for subclasses...
   *
   */
  public DocumentSQLWriter getSQLWriter()
  { return new DocumentSQLWriter();
  }

  /**
   *  Obtain a document from the SQL database
   */
    public static AbstractDocument readSQL(GS3SQLConnection connection, ResultSet sqlResult)
  { try {
      DocumentID id = new DocumentID(sqlResult.getString("DocID"));
      String     type = sqlResult.getString("docType");

      // Use a factory method to create the correct subtype...
      AbstractDocument document = DocumentFactory.createDocument(type, id);
      // Append the document date information
      document.indexDate    = sqlResult.getTimestamp("IndexedDate");
      document.firstDate    = sqlResult.getTimestamp("AccessionDate");
      document.modifiedDate = sqlResult.getTimestamp("ModifiedDate");

      // Get the individual components of the document
      METSFileSet fileSet = METSFileSet.readSQL(document, connection);
      document.setDocumentFiles(fileSet);
      METSDescriptiveSet descriptiveSet = METSDescriptiveSet.readSQL(document, connection);
      document.setDocumentMetadata(descriptiveSet);
      METSStructureSet structureSet = METSStructureSet.readSQL(document, connection);
      document.setDocumentStructure(structureSet);

      // indicate that the document is not currently modified
      document.setChanged(false);
      return document;
    }
    catch (SQLException sqlEx) {
      System.err.println("AbstractDocument.readSQL(): Failure to load document: " + sqlEx);
    }
    return null;
  }

  /**
   *  
   */
  public boolean isChanged()
  { return this.isModified;
  }

  public void setChanged(boolean isModified)
  { this.isModified = isModified;
  }
}