Esempio n. 1
0
  @Override
  public boolean add(Document o) {
    if (o == null) return false;
    Document doc = o;

    // make it accept only docs from its own datastore
    if (doc.getDataStore() != null && !this.dataStore.equals(doc.getDataStore())) {
      Err.prln("Error: Persistent corpus can only accept documents " + "from its own datastore!");
      return false;
    } // if

    // add the document with its index in the docDataList
    // in this case, since it's going to be added to the end
    // the index will be the size of the docDataList before
    // the addition
    DocumentData docData =
        new DocumentData(doc.getName(), doc.getLRPersistenceId(), doc.getClass().getName());
    boolean result = docDataList.add(docData);
    documents.add(doc);
    documentAdded(doc);
    fireDocumentAdded(
        new CorpusEvent(
            SerialCorpusImpl.this,
            doc,
            docDataList.size() - 1,
            doc.getLRPersistenceId(),
            CorpusEvent.DOCUMENT_ADDED));

    return result;
  }
Esempio n. 2
0
  @Override
  public void add(int index, Document o) {
    if (o == null) return;
    Document doc = o;

    DocumentData docData =
        new DocumentData(doc.getName(), doc.getLRPersistenceId(), doc.getClass().getName());
    docDataList.add(index, docData);

    documents.add(index, doc);
    documentAdded(doc);
    fireDocumentAdded(
        new CorpusEvent(
            SerialCorpusImpl.this,
            doc,
            index,
            doc.getLRPersistenceId(),
            CorpusEvent.DOCUMENT_ADDED));
  }
Esempio n. 3
0
  /**
   * Constructor to create a SerialCorpus from a transient one. This is called by adopt() to store
   * the transient corpus and re-route the methods calls to it, until the corpus is sync-ed on disk.
   * After that, the transientCorpus will always be null, so the new functionality will be used
   * instead.
   */
  protected SerialCorpusImpl(Corpus tCorpus) {
    // copy the corpus name and features from the one in memory
    this.setName(tCorpus.getName());
    this.setFeatures(tCorpus.getFeatures());

    docDataList = new ArrayList<DocumentData>();
    // now cache the names of all docs for future use
    List<String> docNames = tCorpus.getDocumentNames();
    for (int i = 0; i < docNames.size(); i++) {
      Document doc = tCorpus.get(i);
      docDataList.add(new DocumentData(docNames.get(i), null, doc.getClass().getName()));
    }

    // copy all the documents from the transient corpus
    documents = new ArrayList<Document>();
    documents.addAll(tCorpus);

    // make sure we fire events when docs are added/removed/etc
    Gate.getCreoleRegister().addCreoleListener(this);
  }
Esempio n. 4
0
  public int findDocument(Document doc) {
    boolean found = false;
    DocumentData docData = null;

    // first try finding the document in memory
    int index = documents.indexOf(doc);
    if (index > -1 && index < docDataList.size()) return index;

    // else try finding a document with the same name and persistent ID
    Iterator<DocumentData> iter = docDataList.iterator();
    for (index = 0; iter.hasNext(); index++) {
      docData = iter.next();
      if (docData.getDocumentName().equals(doc.getName())
          && docData.getPersistentID().equals(doc.getLRPersistenceId())
          && docData.getClassType().equals(doc.getClass().getName())) {
        found = true;
        break;
      }
    }
    if (found && index < docDataList.size()) return index;
    else return -1;
  } // findDocument
Esempio n. 5
0
  public void setTransientSource(Object source) {
    if (!(source instanceof Corpus)) return;

    // the following initialisation is only valid when we're
    // constructing
    // this object from a transient one. If it has already been stored
    // in
    // a datastore, then the initialisation is done in readObject()
    // since
    // this method is the one called by serialisation, when objects
    // are restored.
    if (this.dataStore != null && this.lrPersistentId != null) return;

    Corpus tCorpus = (Corpus) source;

    // copy the corpus name and features from the one in memory
    this.setName(tCorpus.getName());
    this.setFeatures(tCorpus.getFeatures());

    docDataList = new ArrayList<DocumentData>();
    // now cache the names of all docs for future use
    List<String> docNames = tCorpus.getDocumentNames();
    for (int i = 0; i < docNames.size(); i++) {
      Document aDoc = tCorpus.get(i);
      docDataList.add(new DocumentData(docNames.get(i), null, aDoc.getClass().getName()));
    }

    // copy all the documents from the transient corpus
    documents = new ArrayList<Document>();
    documents.addAll(tCorpus);

    this.addedDocs = new Vector<Document>();
    this.removedDocIDs = new Vector<String>();
    this.changedDocs = new Vector<Document>();

    // make sure we fire events when docs are added/removed/etc
    Gate.getCreoleRegister().addCreoleListener(this);
  }