Changeset 6349 for trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/indexers/MGIndexer.java

Timestamp:

2004-01-06T11:46:04+13:00 (20 years ago)

Author:

cs025

Message:

Modified indexerinterface to allow easier configuration, improved MG section handling.

File:

• trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/indexers/MGIndexer.java (modified) (14 diffs)

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/indexers/MGIndexer.java

@@ -35 +35 @@
   List         indexes;
   String       level;
-
-  public static final String MG_INDEX = "Index";
+  String       field;
 
   class MGIndex
@@ -42 +41 @@
     String field;
 
+    public MGIndex(String level, String field)
+    { this.level = level;
+      this.field = field;
+    }
+
     public MGIndex(String indexLabel)
     { int colonAt = indexLabel.indexOf(':');
+    
       if (colonAt >= 0)
       { field = indexLabel.substring(colonAt+1);
         level = indexLabel.substring(0, colonAt);
-      }
-      else
-      { 
       }
     }
@@ -91 +93 @@
   /**
    *  The output directory should be (collection)/building/text/ for
-   *  normal Greenstone builds
+   *  normal Greenstone builds.
+   *
+   *  @param <code>String</code> the label to configure
+   *  @param <code>String</code> the value...
    */ 
   public boolean configure(String label, String value)
@@ -114 +119 @@
       System.out.println("Output MG directory is " + this.textStem);
     }
-    else if (label.equals(MG_INDEX)) {
+    else if (label.equals(IndexerInterface.GS2_INDEX_LABEL)) {
       this.indexes.add(new MGIndex(value));
     }
@@ -121 +126 @@
   }
 
+  public boolean addIndex(String level, String field)
+  {
+    MGIndex index = new MGIndex(level, field);
+    this.indexes.add(index);
+    return true;
+  }
+
   private Node recurseDOM(DocumentInterface metsDoc, Node node, 
-              AbstractStructure structure, StringBuffer buffer)
+              AbstractStructure structure, StringBuffer buffer,
+              String namespace, String field)
   {
     // send out the ctrl-c...if this is
@@ -146 +159 @@
       buffer.append((char) 3);
       if (this.level != null && 
-      this.level.equals("section")) {
+      this.level.equals(IndexerInterface.SECTION_LEVEL)) {
     buffer.append((char) 2);
       }
@@ -152 +165 @@
     }
 
-    // go through our children as required...
+    // go through our children if required...
     Iterator children = structure.getChildIterator();
     while (children.hasNext()) {
       AbstractStructure child = (AbstractStructure) children.next();
-
+      
       // get xpointer for child 
       // get start position node
       Node startNode = ((HTMLDocument) metsDoc).getSectionStartNode((METSDivision) child);
-
-      // while this node isn't the child's start node, produce the node
-      while (node != startNode) {
-    XPointer.printNode(node, buffer, false);
-    // print buffer to node
-    node = XPointer.getNextNode(node, buffer);
-      }
-
+      
+      // while this node isn't the child's start node, produce the node text
+      if (field.equals("text")) {
+    while (node != startNode) {
+      XPointer.printNode(node, buffer, false);
+
+      // print buffer to node
+      node = XPointer.getNextNode(node, (field.equals("text") ? buffer : null));
+    }
+      }
+      
       // recurse to child
-      this.recurseDOM(metsDoc, node, child, buffer);
+      this.recurseDOM(metsDoc, node, child, buffer, namespace, field);
     }
 
@@ -176 +192 @@
     if (structure.getStructureType().equals(METSStructure.STRUCTURE_TYPE)) {
       while (node != null) {
-    XPointer.printNode(node, buffer, false);
-    node = XPointer.getNextNode(node, buffer);
+    if (field.equals("text")) {
+      XPointer.printNode(node, buffer, false);
+    }
+    else {
+      METSDescriptive descriptive;
+
+      if (structure.getStructureType().equals(METSDivision.DIVISION_TYPE)) {
+        METSDivision division = (METSDivision) structure;
+
+        String metadataId = division.getDefaultMetadataReference();
+        
+        descriptive = metsDoc.getDocumentMetadata().getDescriptiveById(metadataId);
+        if (descriptive != null) {
+          List values = descriptive.getMetadata(namespace, field);
+          
+          Iterator valueIter = values.iterator();
+          while (valueIter.hasNext()) {
+        String value = valueIter.next().toString();
+        
+        buffer.append(value);
+        if (valueIter.hasNext()) {
+          buffer.append((char) 3);
+        }
+          }
+        }
+      }
+    }
+    node = XPointer.getNextNode(node, (field.equals("text") ? buffer : null));
       }
       buffer.append((char) 3);
@@ -185 +227 @@
   }
 
-  private String prepareDOM(DocumentInterface metsDoc, Document document, METSStructure structure)
+  private String prepareDOM(DocumentInterface metsDoc, Document document, METSStructure structure, String namespace, String field)
   { Node node = document.getDocumentElement();
     StringBuffer textBuffer = new StringBuffer();
     
-    this.recurseDOM(metsDoc, node, structure, textBuffer);
+    this.recurseDOM(metsDoc, node, structure, textBuffer, namespace, field);
     return textBuffer.toString();
   }
@@ -217 +259 @@
       METSStructure sections = document.getDocumentStructure().getStructure("Section");
       if (sections != null) {
-    docText = this.prepareDOM(document, domDocument, sections);
+    docText = this.prepareDOM(document, domDocument, sections, "gsdl3", this.field);
     //  System.out.println(docText);
       }
@@ -316 +358 @@
 
     this.indexStem = this.outputDirectory + File.separatorChar + 
-      this.getIndexDirectory("document", "text") +
+      this.getIndexDirectory(index.getLevel(), index.getField()) +
       File.separatorChar + "index"; // TODO: modify for index
     this.level = index.getLevel();
+    this.field = index.getField();
+      }
+      else {
+    this.field = "text";
       }
       
@@ -343 +389 @@
         Process p = Runtime.getRuntime().exec("mg_perf_hash_build -f index -d " + this.indexDirectory.toString());
         p.waitFor();
-        System.out.println(p.exitValue());
+        if (p.exitValue() == 0) {
+          System.out.println("Perfect hashes completed");
+        }
         
         mg_passes = Runtime.getRuntime().exec("mg_passes " + pathParams +" -b 100000 -2 -c 3 -G -t 10 -N2");
@@ -402 +450 @@
       try {
     switch (mgPass)
-    {                   
+    { 
       case 0:
         System.out.println("Compressing dictionary");
@@ -413 +461 @@
 
       case 3:
+        System.out.println("Writing weights file");
         p = Runtime.getRuntime().exec("mg_weights_build -f " + this.indexStem + " -t " + this.textStem + " -d /");
         p.waitFor();
-        System.out.println(p.exitValue());
+        if (p.exitValue() == 0) {
+          System.out.println("Weights file successfully written");
+        }
+        else {
+          System.out.println("Unable to create weights file");
+        }
 
         p = Runtime.getRuntime().exec("mg_invf_dict -f index -d " + this.indexDirectory.toString());
         p.waitFor();
-        System.out.println(p.exitValue());
+        if (p.exitValue() == 0) {
+          System.out.println("Inverted dictionary file successfully written");
+        }
+        else {
+          System.out.println("Unable to create inverted dictionary file");
+        }
         
         p = Runtime.getRuntime().exec("mg_stem_idx -b 4096 -s1 -f index -d " + this.indexDirectory.toString());
         p.waitFor();
-        System.out.println(p.exitValue());
+        if (p.exitValue() == 0) {
+          System.out.println("Stemmed index successfully written");
+        }
+        else {
+          System.out.println("Unable to create stemmed index");
+        }
+
         p = Runtime.getRuntime().exec("mg_stem_idx -b 4096 -s2 -f index -d " + this.indexDirectory.toString());
         p.waitFor();
-        System.out.println(p.exitValue());
+        if (p.exitValue() == 0) {
+          System.out.println("Stemmed index successfully written");
+        }
+        else {
+          System.out.println("Unable to create stemmed index");
+        }
+
         p = Runtime.getRuntime().exec("mg_stem_idx -b 4096 -s3 -f index -d " + this.indexDirectory.toString());
         p.waitFor();
-        System.out.println(p.exitValue());
+        if (p.exitValue() == 0) {
+          System.out.println("Stemmed index successfully written");
+        }
+        else {
+          System.out.println("Unable to create stemmed index");
+        }
       break;
     }