Changeset 6697

Timestamp:

2004-02-02T16:07:09+13:00 (20 years ago)

Author:

cs025

Message:

Additions to cope with handling of date of accession, etc.

Location:

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes

Files:

• AbstractDocument.java (modified) (7 diffs)
• DocumentInterface.java (modified) (2 diffs)
• DocumentList.java (modified) (1 diff)
• DocumentSQLWriter.java (modified) (1 diff)
• HTMLDocument.java (modified) (1 diff)

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/AbstractDocument.java

@@ -6 +6 @@
 import java.util.HashMap;
 import java.util.Map;
+import java.util.Date;
 
 import java.sql.SQLException;
 import java.sql.ResultSet;
+//import java.sql.Date;
 
 import java.net.URL;
@@ -35 +37 @@
 public abstract class AbstractDocument implements DocumentInterface
 {
-  METSFileSet      fileSet;
+  METSFileSet        fileSet;
   METSDescriptiveSet metadata;
-  METSStructureSet structureSet;
-  METSHeader       header;
-  DocumentID       id;
-  boolean          isModified;
+  METSStructureSet   structureSet;
+  METSHeader         header;
+  DocumentID         id;
+  boolean            isModified;
   StructureIdentifierFactory structureIdFactory;
+  java.sql.Date      firstDate;
+  java.sql.Date      indexDate;
 
   /**
@@ -57 +61 @@
     this.id           = id;
     this.structureIdFactory = new StructureIdentifierFactory();
+
+    java.util.Date thisDate = new java.util.Date();
+    this.firstDate = new java.sql.Date(thisDate.getTime());
+    this.indexDate = new java.sql.Date(thisDate.getTime());
   }
   
@@ -73 +81 @@
     this.id           = null;
 
+    java.util.Date thisDate = new java.util.Date();
+    this.firstDate = new java.sql.Date(thisDate.getTime());
+    this.indexDate = new java.sql.Date(thisDate.getTime());
+
     METSStructure structure = new METSStructure("All", "All", "Whole Document");
     METSDivision documentBody = new METSDivision("All", "All", "All", "Whole Document", "Document");
@@ -113 +125 @@
    */
   public boolean isIndexed()
-  {
-    return true;
+  { return true;
+  }
+
+  /**
+   *  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 boolean hasDuplicate(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() + "\")";
+    connection.execute(query);
+
+    List docs = new ArrayList();
+    ResultSet results = connection.getResultSet();
+
+    try {
+      if (results != null &&
+      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 + "\"";
+      connection.execute(innerQuery);
+      ResultSet innerSet = connection.getResultSet();
+      if (innerSet != null && innerSet.first()) {
+        String docType = innerSet.getString("DocType");
+        if (docType.equals(this.getDocumentType())) {
+          return true;
+        }
+      }
+    }
+      }
+    }
+    catch (java.sql.SQLException sqlEx) {
+      System.err.println(sqlEx);
+    }
+
+    return false;
   }
 
@@ -330 +392 @@
       AbstractDocument document = DocumentFactory.createDocument(type, id);
 
+      // Append the document date information
+      document.indexDate = sqlResult.getDate("IndexedDate");
+      document.firstDate = sqlResult.getDate("AccessionDate");
+
       // Get the individual components of the document
       METSFileSet fileSet = METSFileSet.readSQL(document, connection);
@@ -343 +409 @@
     }
     catch (SQLException sqlEx) {
+      System.out.println("Failure to load document: " + sqlEx);
     }
     return null;

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/DocumentInterface.java

@@ -9 +9 @@
 
 import org.greenstone.gsdl3.gs3build.util.MultiMap;
+import org.greenstone.gsdl3.gs3build.util.GS3SQLConnection;
 
 public interface DocumentInterface
@@ -49 +50 @@
   public boolean isIndexed();
 
+  /**
+   *  Check if the document matches another in the database
+   */
+  public boolean hasDuplicate(GS3SQLConnection connection);
+    
   /**
    *  The plain text of the document.

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/DocumentList.java

@@ -280 +280 @@
    */
   public void addDocument(DocumentInterface document)
-  { // first cache it...
+  { // initially, test if the document has a duplicate...
+    if (document.hasDuplicate(this.connection)) {
+      System.out.println("Found duplicate document ");
+      return;
+    }
+
+    // first cache it...
     this.cacheDocument(document);
 

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/DocumentSQLWriter.java

@@ -39 +39 @@
       insert.addValue("DocID", document.getID().toString());
       insert.addValue("DocType", document.getDocumentType());
+      // TODO: avoid this terrible cast
+      insert.addDate("AccessionDate", ((AbstractDocument) document).firstDate);
 
       connection.execute(insert.toString());

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/HTMLDocument.java

@@ -5 +5 @@
 import java.net.MalformedURLException;
 
+import java.util.Iterator;
 import java.util.List;
 import java.util.Map;