Changeset 6496

Timestamp:

2004-01-14T16:26:50+13:00 (20 years ago)

Author:

cs025

Message:

Added AZList; support sorting within categories.

Location:

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

Files:

• AZListClassifier.java (added)
• AbstractHierarchyNode.java (modified) (5 diffs)
• ClassifierManager.java (modified) (1 diff)
• GS2HierarchyClassifier.java (modified) (6 diffs)
• HierarchyClassifier.java (modified) (10 diffs)

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

@@ -29 +29 @@
   AbstractHierarchyNode parent; // the parent of the node
 
+  class AbstractHierarchyDocument
+  { DocumentID documentId;
+    String     sortKey;
+
+    public AbstractHierarchyDocument(DocumentID documentId, String sortKey)
+    { this.documentId = documentId;
+      this.sortKey    = sortKey;
+    }
+
+    public DocumentID getID()
+    { return this.documentId;
+    }
+
+    public String toString()
+    { return this.sortKey;
+    }
+  } 
+
   /**
    *  Simple node
@@ -53 +71 @@
     this.matches    = new ArrayList();
     this.matches.add(this.id);
+    this.matches.add(this.name);
   }
     
@@ -88 +107 @@
   }
 
-  public void addDocument(DocumentID document)
-  { this.childDocs.add(document);
+  public void addDocument(DocumentID document, String sortKey)
+  { this.childDocs.add(new AbstractHierarchyDocument(document, sortKey));
+
+    if (sortKey != null) {
+      int first = 0;
+      int last  = this.childDocs.size() - 1;
+
+      while (first != last) {
+    int at = (first + last) / 2;
+    
+    if (this.childDocs.get(at) == null) {
+      last = at;
+      continue;
+    }
+
+    if (this.childDocs.get(at).toString().compareTo(sortKey) > 0) {
+      last = at;
+    }
+    else {
+      first = at + 1;
+    }
+      }
+
+      Object newItem = this.childDocs.get(this.childDocs.size() - 1);
+      last = this.childDocs.size() - 1;
+      while (last > first) {
+    this.childDocs.set(last, this.childDocs.get(last-1));
+    last --;
+      }
+      this.childDocs.set(first, newItem);
+    }
   }
 
@@ -185 +233 @@
    *         classifications that it fell within.
    */
-  abstract public void getClassifications(DocumentID documentID, List values, 
+  abstract public void getClassifications(DocumentID documentID, List values, String sortKey,
                       ClassifierObserverInterface observer);
 
@@ -280 +328 @@
     // note the child documents...
     iterator = this.childDocs.iterator();
+    int order = 1;
     while (iterator.hasNext()) {
-      DocumentID docId = (DocumentID) iterator.next();
+      AbstractHierarchyDocument hierarchyDoc = (AbstractHierarchyDocument) iterator.next();
+      DocumentID docId = hierarchyDoc.getID();
+
       insert = new GS3SQLInsert("classdocuments");
       insert.addValue("ClassifyRef", Integer.toString(classifyRef), GS3SQLField.INTEGER_TYPE);
       insert.addValue("DocID", docId.toString());
+      insert.addValue("DocOrder", Integer.toString(order), GS3SQLField.INTEGER_TYPE);
 
       connection.execute(insert.toString());
+
+      order ++;
     }
     return true;
   }
 }
+
+

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

@@ -82 +82 @@
         
         if (document.isModified()) {
+          //          System.out.println("Writing document " + document.getID());
           this.documents.modifiedDocument(document);
         }

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/classifier/GS2HierarchyClassifier.java

@@ -33 +33 @@
     public void recordClassification(String label)
     { this.document.addDocumentMetadata("gsdl3", "classified", label);
-      System.out.println("Assigned document " + this.document.getID().toString() + " to " + label);
+      this.document.setModified(true);
+      //      System.out.println("Assigned document " + this.document.getID().toString() + " to " + label);
     }
 
@@ -65 +66 @@
      *         classifications that it fell within.
      */
-    public void getClassifications(DocumentID documentID, List values, 
+    public void getClassifications(DocumentID documentID, List values, String sortKey,
                    ClassifierObserverInterface observer)
     { // Go through the list of values, and if this node matches, record
@@ -73 +74 @@
       while (valueList.hasNext())
       { if (this.isMatch(valueList.next().toString()))
-    { observer.recordClassification(this.name);
-      this.addDocument(documentID);
+    { observer.recordClassification(this.id);
+      this.addDocument(documentID, sortKey);
     }
       }
@@ -82 +83 @@
 
       while (childList.hasNext())
-      { ((GS2HierarchyNode) childList.next()).getClassifications(documentID, values, observer);
+      { ((GS2HierarchyNode) childList.next()).getClassifications(documentID, values, sortKey, observer);
       }
     }
@@ -238 +239 @@
     while (thisField.hasNext())
     { String fieldName = thisField.next().toString();
-            
+
       // ...get the values for that field...
       List values = document.getDocumentMetadataItem(fieldName);
@@ -253 +254 @@
       }
 
+      // 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, classifyObserver);
+      this.hierarchy.getClassifications(documentID, values, sortKey, classifyObserver);
     }
     return true;

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

@@ -21 +21 @@
 public class HierarchyClassifier implements ClassifierInterface
 { 
-  class HierarchyClassifierObserver
+  class HierarchyClassifierObserver implements ClassifierObserverInterface
   { List              classifications;
     DocumentInterface document;
@@ -38 +38 @@
   }
 
-  class HierarchyNode
-  { String descriptor;    // the textual descriptor used on screen or long-hand
-    String path;          // the index number, letter assignment or other item
-                          //   used to identify the position of the item in the
-                          //   hierarchy
-    String name;          // an identifier used by the GLI for maintenance
-                          //   purposes; this plays no active role in the
-                          //   rebuilding process (at the moment)
-    List childNodes;      // the child classification nodes of this node
-    List childDocs;       // the child documents of this node
-    List matches;         // the other metadata values that may be matched
-                          //   against the classifier
-    HierarchyNode parent; // the parent of the node
-
-    /**
+  class HierarchyNode extends AbstractHierarchyNode
+  { /**
      *  Simple node
      */
     public HierarchyNode()
-    { this.descriptor = null;
-      this.name       = null;
-      this.path       = null;
-      this.childNodes = new ArrayList();
-      this.childDocs  = new ArrayList();
-      this.parent     = null;
+    { super();
     }
     
-    public void addChild(HierarchyNode child)
-    { this.childNodes.add(child);
-      child.setParent(this);
-    }
-
-    public void addDocument(DocumentID document)
-    { this.childDocs.add(document);
-    }
-
-    public void setParent(HierarchyNode parent)
-    { this.parent = parent;
-    }
-    
-    public HierarchyNode getParent()
-    { return this.parent;
-    }
-
-    public void setDescriptor(String descriptor)
-    { this.descriptor = descriptor;
-    }
-
-    public void setPath(String path)
-    { this.path = path;
-    }
-
-    public void setName(String name)
-    { this.name = name;
-    }
-
-    public void addMatch(String match)
-    { this.matches.add(match);
-    }
-
-    public boolean isMatch (String toMatch)
-    { Iterator thisMatch = this.matches.iterator();
-            
-      while (thisMatch.hasNext())
-      { String thisMatchText = thisMatch.next().toString();
-
-        if (thisMatchText.equals(toMatch))
-    { return true;
-    }
-      }
-      return false;
-    }
-
-    public void getClassifications(DocumentID documentID, List values, 
-                   HierarchyClassifierObserver observer)
+    public void getClassifications(DocumentID documentID, List values, String sortKey,
+                   ClassifierObserverInterface observer)
     { Iterator valueList = values.iterator();
 
@@ -117 +53 @@
       { if (this.isMatch(valueList.next().toString()))
     { observer.recordClassification(this.name);
-      this.addDocument(documentID);
-    }
-      }
+      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, observer);
+      { ((HierarchyNode) childList.next()).getClassifications(documentID, values, sortKey, observer);
       }
     }
@@ -137 +75 @@
     StringBuffer match;
     boolean inElement;
-    HierarchyNode rootNode = null;
-    HierarchyNode currentNode = null;
+    AbstractHierarchyNode rootNode = null;
+    AbstractHierarchyNode currentNode = null;
     List          rootNodes = null;
 
@@ -195 +133 @@
       }
       else if (localName.equals("Path"))
-      { this.currentNode.setPath(XMLTools.cleanString(this.path.toString()));
+      { this.currentNode.setID(XMLTools.cleanString(this.path.toString()));
         this.path = null;
       }
@@ -227 +165 @@
     }
 
-    public HierarchyNode getHierarchy()
+    public AbstractHierarchyNode getHierarchy()
     { return this.rootNode;
     }
   }
-
-  private HierarchyNode hierarchy;
-  private List          fields;
-  private String        sort;
-
-  public HierarchyClassifier(File basefile, List fields, String sort)
+  
+  private GS3SQLConnection database;
+  private AbstractHierarchyNode    hierarchy;
+  private List             fields;
+  private String           sortBy;
+
+  public HierarchyClassifier(File basefile, List fields, String sortBy)
   { 
     try 
@@ -250 +189 @@
       this.hierarchy = handler.getHierarchy();
       this.fields = fields;
-      this.sort   = sort;
+      this.sortBy = sortBy;
     }
     catch (SAXException saxException)
@@ -265 +204 @@
   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);
+      }
+    }
+      }
+    }
   }
 
   public void setDatabase(GS3SQLConnection database)
-  {
+  { this.database = database;
   }
 
@@ -298 +249 @@
       }
 
+      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, classifyObserver);
+      this.hierarchy.getClassifications(documentID, values, sortKey, classifyObserver);
     }
     return true;
@@ -318 +284 @@
   public void completeClassification()
   { // TODO: store the classifications to file...
+    if (this.hierarchy != null) {
+      this.hierarchy.writeSQL(this.database);
+    }
   }
 }