Changeset 7186

Timestamp:

2004-04-06T09:24:53+12:00 (20 years ago)

Author:

cs025

Message:

Merged changes; slow widening of configuration options.

Location:

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

Files:

• BuildManager.java (modified) (2 diffs)
• CollectionManager.java (modified) (7 diffs)

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/BuildManager.java

@@ -56 +56 @@
     recogniser = new TextRecogniser(docList);
     recogniserManager.addRecogniser(recogniser);
+    recogniser = new JPEGRecogniser(docList);
+    recogniserManager.addRecogniser(recogniser);
+    recogniser = new ExtXMLRecogniser(docList);
+    recogniserManager.addRecogniser(recogniser);
     recogniser = new MetadataRecogniser(docList);
     recogniserManager.addRecogniser(recogniser);
@@ -62 +66 @@
     this.extractorManager  = new ExtractorManager(this.docList);
     this.extractorManager.addExtractor(new IndexExtractor());
+    this.extractorManager.addExtractor(new ExtXMLExtractor());
     this.extractorManager.addExtractor(new MetaXMLExtractor());
     

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/CollectionManager.java

@@ -67 +67 @@
    
   class CollectionClassifier
-  { File file;
+  { URL    file;
     String type;
     List fields;
     String sort;
 
-    public CollectionClassifier(String type, Node node)
+    public CollectionClassifier(URL parentURL, String type, Node node)
     { this.type = type;
       this.fields = new ArrayList();
@@ -90 +90 @@
           continue;
 
-        this.file = new File(urlString);
+        try {
+          URL url = new URL(parentURL, urlString);
+          this.file = url;
+          System.out.println(url);
+        }
+        catch (java.net.MalformedURLException malEx) {
+          System.out.println(malEx);
+        }
       }
       else if (name.equals("field")) {
         String fieldName = DOMUtils.getNodeChildText(children.item(c));
+        System.out.println("Adding field " + fieldName);
         this.fields.add(fieldName.toString());
       }
@@ -109 +117 @@
     return null;
       }
+      System.out.println(this.type.toLowerCase());
       if (this.type.toLowerCase().equals("hierarchy")) {
     return new HierarchyClassifier(this.file, this.fields, this.sort);
@@ -180 +189 @@
     this.buildDocNo = 1;
   }
-    public void setBuildManager(BuildManager build_man) {
-    this.buildManager = build_man;
-    }
-
-  private void configureBrowsers(Node node)
+
+  public void setBuildManager(BuildManager build_man) 
+  { this.buildManager = build_man;
+  }
+
+  private void configureBrowsers(Node node, File etcFile)
   { CollectionClassifier classifier = null;
+    URL etcURL = null;
+
+    try {
+      etcURL = new URL("file://"+etcFile.toString());
+    }
+    catch (java.net.MalformedURLException malEx) {
+      System.out.println(malEx);
+    }
 
     NodeList children = node.getChildNodes();
@@ -206 +224 @@
 
     String type = attribute.getNodeValue();
-    classifier = new CollectionClassifier(type, children.item(c));
+    classifier = new CollectionClassifier(etcURL, type, children.item(c));
     
     System.out.println("Found classifier " + type);
@@ -251 +269 @@
 
     // the name is a plugin element
-    if (name.equals(GSXML.SEARCH_ELEM)) {
+
+    if (name.equals("doctype")) {
+      System.out.println("document type");
+      NamedNodeMap typeAttributes = children.item(c).getAttributes();
+      Node typeAttribute = typeAttributes.getNamedItem("type");
+      String documentType = null;
+
+      NodeList childNodes = children.item(c).getChildNodes();
+      for (int n = 0; n < childNodes.getLength(); n ++)
+      { if (childNodes.item(n).getNodeType() == org.w3c.dom.Node.TEXT_NODE)
+        { String label = childNodes.item(n).getNodeValue();
+              label.trim();
+          if (label.length() > 0) {
+        documentType = label;
+        System.out.println("Document type " + documentType);
+          }
+        }
+      }
+    }
+    else if (name.equals(GSXML.SEARCH_ELEM)) {
       // pick up attributes from the <search> tag now...
       NamedNodeMap searchAttributes = children.item(c).getAttributes();
@@ -283 +320 @@
     }
     else if (name.equals(GSXML.BROWSE_ELEM))
-    { this.configureBrowsers(children.item(c));
+    { this.configureBrowsers(children.item(c), collectionConfig);
     }
     // TODO: other elements - make a factory-method approach here...