Changeset 7188

Timestamp:

2004-04-06T09:26:49+12:00 (20 years ago)

Author:

cs025

Message:

Classifier fixes and improvements.

Location:

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

Files:

• AZListClassifier.java (modified) (6 diffs)
• AbstractHierarchyNode.java (modified) (1 diff)
• HierarchyClassifier.java (modified) (6 diffs)

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/classifier/AZListClassifier.java

@@ -170 +170 @@
   }
 
-  public int writeSQLClassifyNode(GS3SQLConnection connection, String parentId, String label, String name, String description, int noOfLeafDocs)
+  public int writeSQLClassifyNode(GS3SQLConnection connection, String parentId, int orderRef,
+                  String label, String name, String description, int noOfLeafDocs)
   { GS3SQLAction action;
     GS3SQLSelect select;
@@ -210 +211 @@
       action.addValue("Name", name);
       action.addValue("Description", description);
+      action.addValue("ClassifyOrder", Integer.toString(orderRef), GS3SQLField.INTEGER_TYPE);
       action.addValue("NumLeafDocs", Integer.toString(noOfLeafDocs), GS3SQLField.INTEGER_TYPE);
 
@@ -262 +264 @@
     }
 
-    int parentClassify = this.writeSQLClassifyNode(connection, "", prefix, "", "", leafCount);
+    int parentClassify = this.writeSQLClassifyNode(connection, "", 0, prefix, "", "", leafCount);
 
     if (parentClassify < 0) 
@@ -271 +273 @@
     
     // TODO: cope with change rather than create from scratch...
-    
+    int nodeOrder = 1;
     keys = this.alphaMap.keySet().iterator();
     while (keys.hasNext()) {
@@ -281 +283 @@
     String className = prefix + "." + key.toString();
 
-    int classifyRef = this.writeSQLClassifyNode(connection, prefix, className, "", "", childDocs.size());
+    int classifyRef = this.writeSQLClassifyNode(connection, prefix, nodeOrder, className, "", "", childDocs.size());
 
     // note the child documents...
         Iterator iterator = childDocs.iterator();
-    int order = 1;
+    int childOrder = 1;
     while (iterator.hasNext()) {
       AZDocumentItem documentItem = (AZDocumentItem) iterator.next();
@@ -293 +295 @@
       insert.addValue("ClassifyRef", Integer.toString(classifyRef), GS3SQLField.INTEGER_TYPE);
       insert.addValue("DocID", docId.toString());
-      insert.addValue("DocOrder", Integer.toString(order), GS3SQLField.INTEGER_TYPE);     
+      insert.addValue("DocOrder", Integer.toString(childOrder), GS3SQLField.INTEGER_TYPE);    
 
       connection.execute(insert.toString());
 
-      order ++;
-    }
+      childOrder ++;
+    }
+
+    nodeOrder ++;
       }
     }

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/classifier/AbstractHierarchyNode.java

@@ -296 +296 @@
       action.addValue("Name", this.name);
       action.addValue("Description", this.descriptor);
+      action.addValue("ClassifyOrder", this.id, GS3SQLField.INTEGER_TYPE);
       action.addValue("NumLeafDocs", Integer.toString(this.noOfLeafDocs()), GS3SQLField.INTEGER_TYPE);
 

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/classifier/HierarchyClassifier.java

@@ -2 +2 @@
 
 import java.io.*;
+
+import java.net.URL;
 
 import java.util.List;
@@ -7 +9 @@
 import java.util.Iterator;
 
+import org.apache.xerces.parsers.SAXParser;
 import org.xml.sax.XMLReader;
 import org.xml.sax.InputSource;
@@ -175 +178 @@
   private String           sortBy;
 
-  public HierarchyClassifier(File basefile, List fields, String sortBy)
+  public HierarchyClassifier(URL basefile, List fields, String sortBy)
   { 
     try 
-    { XMLReader reader = XMLReaderFactory.createXMLReader();
+      { SAXParser parser = new SAXParser();
       HierarchyHandler handler = new HierarchyHandler();
+      /*
+      XMLReader reader = XMLReaderFactory.createXMLReader();
       reader.setContentHandler(handler);
-      reader.setErrorHandler(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;
+
+      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());
     }
   }
@@ -218 +232 @@
   }
 
+  /**
+   *  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;
@@ -230 +249 @@
    *  @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();
@@ -270 +297 @@
   }
 
+  /**
+   *  @see ClassifierInterface:getClassifierPasses
+   */
   public int getClassifierPasses()
   { return 1;