source: trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/BuildManager.java@ 8408

package org.greenstone.gsdl3.gs3build;

import org.greenstone.gsdl3.gs3build.doctypes.*;
import org.greenstone.gsdl3.gs3build.indexers.*;
import org.greenstone.gsdl3.gs3build.extractor.*;
import org.greenstone.gsdl3.gs3build.classifier.*;
import org.greenstone.gsdl3.gs3build.collection.*;

import java.io.File;
import java.util.List;

public class BuildManager 
{
  CollectionManager collectionManager;
  RecogniserManager recogniserManager;
  IndexerManager    indexerManager;
  ClassifierManager classifierManager;
  ExtractorManager  extractorManager;
  DocumentList      docList;
  List              inputRoots;
  String            outputDir;
  String            archiveDir;
    
  public BuildManager(List inputRoots, String site, String collection, String outputDir)
  {
    // get the collection configuration information
    this.collectionManager = new CollectionManager(site, collection);
    this.collectionManager.setBuildManager(this);
    // set up the list of input directories
    if (inputRoots.size() == 0) {
      inputRoots.add(this.collectionManager.getImportDirectory());
      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
    this.outputDir  = outputDir != null ? outputDir : this.collectionManager.getBuildDirectory();
    File outputDirFile = new File(this.outputDir);
    if (!outputDirFile.exists() || !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);

    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);

    // set up the extractors
    this.extractorManager  = new ExtractorManager(this.docList);
    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();

    // prepare a file crawler on the etc directory, and a crawl observer to respond to
    // the appropriate files.
    FileCrawler fileCrawler = new FileCrawler(new File(this.collectionManager.getEtcDirectory()), 
                          new CollectionCrawlObserver(this));

    fileCrawler.crawl();    
  }
  
  public String getLastBuildSequence()
  { return this.collectionManager.getCollectionMetadata("gsdl3", "buildsequence");
  }

  public ClassifierManager getClassifierManager()
  { return this.classifierManager;
  }

  public IndexerManager getIndexerManager()
  { return this.indexerManager;
  }

  public void addIndexer(IndexerInterface iface)
  { iface.configure(IndexerManager.outputDir, this.outputDir);
    this.indexerManager.addIndexer(iface);
  }

  public void run()
  { // Initialise collection manager - note start of build
    this.collectionManager.startBuild();

    // TODO: add expansion (e.g. Zip files)
    
    // Crawl the file tree - will recognise documents
    for (int i = 0; i < this.inputRoots.size(); i ++)
    { FileCrawler fileCrawler = new FileCrawler(new File((String) this.inputRoots.get(i)), recogniserManager);

      fileCrawler.crawl();
    }

    // Extract phase, etc.
    this.extractorManager.extractDocuments();
    this.classifierManager.classifyDocuments();
    this.indexerManager.indexDocuments();

    // Timestamp management - update all timestamps on modified dates...
    // 
    // 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()));
    
    // TODO: write out collection configuration file (service lists, etc.)
    
    this.collectionManager.endBuild();
  }
  
  public void addInputDirectory(String inputDir)
  { this.inputRoots.add(inputDir);
  }
}