Changeset 8490

Timestamp:

2004-11-09T11:22:56+13:00 (19 years ago)

Author:

kjdon

Message:

slightly reordered the code, removed the addition of recognisers, cos this is now done from the config file <recognise><docType name='HTML'/>...</recognise>

File:

• trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/BuildManager.java (modified) (7 diffs)

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

@@ -21 +21 @@
     String            outputDir;
     String            archiveDir;
-    
-    public BuildManager(List inputRoots, String site, String collection, String outputDir)
+    
+    public BuildManager(List inputRoots, String site, String collection, String outputDir) 
     {
-    // get the collection configuration information
+    // set up the collection manager
     this.collectionManager = new CollectionManager(site, collection);
     this.collectionManager.setBuildManager(this);
+    
     // set up the list of input directories
     if (inputRoots.size() == 0) {
@@ -32 +33 @@
         System.out.println("Adding collection import directory to input list " + this.collectionManager.getImportDirectory());
     }
-    this.docList = new DocumentList(new DocumentIDFactory(this.collectionManager), 
-                    this.collectionManager.getDatabase());
     this.inputRoots = inputRoots;
-    
-    // set up the output directory
+    
+    // set up the output directories (building and archive)
     this.outputDir  = outputDir != null ? outputDir : this.collectionManager.getBuildDirectory();
     File outputDirFile = new File(this.outputDir);
-    if (!outputDirFile.exists() || !outputDirFile.isDirectory()) {
+    if (!outputDirFile.isDirectory()) {
         System.err.println("Unable to open directory " + this.outputDir + " for writing");
         System.exit(1);
     }
     System.out.println("Output directory set to " + this.outputDir);
-
+    
+    // set up the archive directory - this will definitely exist
     this.archiveDir = this.collectionManager.getArchiveDirectory();
 
-    // TODO: initialise the expanders here...
-
-    // initialise any recognisers
-    this.recogniserManager = new RecogniserManager();
-    RecogniserInterface recogniser = new HTMLRecogniser(docList);
-    recogniserManager.addRecogniser(recogniser);
-    recogniserManager.addRecogniser(new IndexRecogniser(docList));
-    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);
-    recogniser = new GS2METSRecogniser(docList);
-    recogniserManager.addRecogniser(recogniser);
-
-    // set up the extractors
+    // the list that documents get added to when they are recognised
+    this.docList = new DocumentList(new DocumentIDFactory(this.collectionManager), 
+                    this.collectionManager.getDatabase());
+    
+    // set up the various managers
+    this.recogniserManager = new RecogniserManager(this.docList);
     this.extractorManager  = new ExtractorManager(this.docList);
+    this.classifierManager = new ClassifierManager(this.docList, collectionManager.getDatabase());
+    this.indexerManager = new IndexerManager(this.docList);
+    
+    // configure the collection - this will add classifiers, indexers, recognisers to the various managers.
+    this.collectionManager.configureCollection();
+    
+    // add in expanders by hand cos they are not in the config file yet
+    // TODO
+    // add in extractors by hand cos they are not in the config file yet
     this.extractorManager.addExtractor(new IndexExtractor());
-    /*    this.extractorManager.addExtractor(new ExtXMLExtractor());*/
     this.extractorManager.addExtractor(new MetaXMLExtractor());
-    
-    // TODO: set up the enrichers...
-    
-    // set up the classifiers
-    this.classifierManager = new ClassifierManager(this.docList, collectionManager.getDatabase());
-        
-    // set up the indexers
-    this.indexerManager = new IndexerManager(this.docList);
-
-    // configure the collection
-    this.collectionManager.configureCollection();
-
+    
+    // add in enrichers by hand cos they are not in the config file yet
+    // TODO
+        
     // prepare a file crawler on the etc directory, and a crawl observer to respond to
     // the appropriate files.
@@ -105 +92 @@
     }
 
+    public RecogniserManager getRecogniserManager()
+    {
+    return this.recogniserManager;
+    }
+    
     public void addIndexer(IndexerInterface iface)
     { 
@@ -115 +107 @@
     // Initialise collection manager - note start of build
     this.collectionManager.startBuild();
-
+    
     // TODO: add expansion (e.g. Zip files)
     
@@ -121 +113 @@
     for (int i = 0; i < this.inputRoots.size(); i ++) {
         FileCrawler fileCrawler = new FileCrawler(new File((String) this.inputRoots.get(i)), recogniserManager);
-        
+
         fileCrawler.crawl();
     }
@@ -134 +126 @@
     // This should only occur at the end of building in case the build is cancelled...
     docList.updateTimestamps(this.collectionManager.getBuildDate());
-
+    
     // TODO: validation phase
-
+    
     if (this.archiveDir != null) {
         this.docList.writeDocuments(new File(this.archiveDir));
         //this.docList.writeSQLDocuments(this.collectionManager.getDatabase());
     }
-    
+    
     this.collectionManager.setCollectionMetadata("gsdl3", "documentCount", Integer.toString(this.docList.getCount()));
     
@@ -150 +142 @@
   
     public void addInputDirectory(String inputDir)
-    { 
+    {
     this.inputRoots.add(inputDir);
     }