Examples with LanguageResource gate.LanguageResource used on opensource projects

Example 6 with LanguageResource

use of gate.LanguageResource in project gate-core by GateNLP.

the class LRPersistence method extractDataFromSource.

/**
 * Populates this Persistence with the data that needs to be stored from the
 * original source object.
 */
@Override
public void extractDataFromSource(Object source) throws PersistenceException {
    // check input
    if (!(source instanceof LanguageResource)) {
        throw new UnsupportedOperationException(getClass().getName() + " can only be used for " + LanguageResource.class.getName() + " objects!\n" + source.getClass().getName() + " is not a " + LanguageResource.class.getName());
    }
    super.extractDataFromSource(source);
    // LR's will have the features saved by their respective persistence
    // mechanism
    features = null;
    LanguageResource lr = (LanguageResource) source;
    if (lr.getDataStore() == null) {
        dsData = null;
    } else {
        dsData = PersistenceManager.getPersistentRepresentation(lr.getDataStore());
        persistenceID = lr.getLRPersistenceId();
    }
}

Example 7 with LanguageResource

use of gate.LanguageResource in project gate-core by GateNLP.

the class TestPersist method testMultipleLrs.

// testSimple()
/**
 * Test multiple LRs
 */
public void testMultipleLrs() throws Exception {
    // create a temporary directory; because File.createTempFile actually
    // writes the bloody thing, we need to delete it from disk before calling
    // DataStore.create
    File storageDir = File.createTempFile("TestPersist__", "__StorageDir");
    storageDir.delete();
    // create and open a serial data store
    SerialDataStore sds = new SerialDataStore(storageDir.toURI().toURL().toString());
    sds.create();
    sds.open();
    // create a document with some annotations / features on it
    String server = TestDocument.getTestServerName();
    Document doc = Factory.newDocument(new URL(server + "tests/doc0.html"));
    doc.getFeatures().put("hi there", new Integer(23232));
    doc.getAnnotations().add(new Long(5), new Long(25), "ThingyMaJig", Factory.newFeatureMap());
    // create another document with some annotations / features on it
    Document doc2 = Factory.newDocument(new URL(server + "tests/html/test1.htm"));
    doc.getFeatures().put("hi there again", new Integer(23232));
    doc.getAnnotations().add(new Long(5), new Long(25), "dog poo irritates", Factory.newFeatureMap());
    // create a corpus with the documents
    Corpus corp = Factory.newCorpus("Hamish test corpus");
    corp.add(doc);
    corp.add(doc2);
    LanguageResource persCorpus = sds.adopt(corp);
    sds.sync(persCorpus);
    // read the documents back
    List<Resource> lrsFromDisk = new ArrayList<Resource>();
    List<String> lrIds = sds.getLrIds("gate.corpora.SerialCorpusImpl");
    Iterator<String> idsIter = lrIds.iterator();
    while (idsIter.hasNext()) {
        String lrId = idsIter.next();
        FeatureMap features = Factory.newFeatureMap();
        features.put(DataStore.DATASTORE_FEATURE_NAME, sds);
        features.put(DataStore.LR_ID_FEATURE_NAME, lrId);
        Resource lr = Factory.createResource("gate.corpora.SerialCorpusImpl", features);
        lrsFromDisk.add(lr);
    }
    if (DEBUG)
        System.out.println("LRs on disk" + lrsFromDisk);
    // check that the versions we read back match the originals
    Corpus diskCorp = (Corpus) lrsFromDisk.get(0);
    Document diskDoc = diskCorp.get(0);
    if (DEBUG)
        Out.prln("Documents in corpus: " + corp.getDocumentNames());
    assertTrue("corp name != mem name", corp.getName().equals(diskCorp.getName()));
    if (DEBUG)
        Out.prln("Memory features " + corp.getFeatures());
    if (DEBUG)
        Out.prln("Disk features " + diskCorp.getFeatures());
    assertTrue("corp feat != mem feat", corp.getFeatures().equals(diskCorp.getFeatures()));
    if (DEBUG)
        Out.prln("Annotations in doc: " + diskDoc.getAnnotations());
    assertTrue("doc annotations from disk not equal to memory version", TestEqual.annotationSetsEqual(doc.getAnnotations(), diskDoc.getAnnotations()));
    assertTrue("doc from disk not equal to memory version", TestEqual.documentsEqual(doc, diskDoc));
    Iterator<Document> corpusIter = diskCorp.iterator();
    while (corpusIter.hasNext()) {
        if (DEBUG)
            Out.prln(corpusIter.next().getName());
        else
            corpusIter.next();
    }
    // assertTrue("doc2 from disk not equal to memory version",
    // doc2.equals(diskDoc2));
    // delete the datastore
    sds.delete();
}

Example 8 with LanguageResource

use of gate.LanguageResource in project gate-core by GateNLP.

the class LuceneDataStoreImpl method sync.

/**
 * Save: synchonise the in-memory image of the LR with the persistent
 * image.
 */
@Override
public void sync(LanguageResource lr) throws PersistenceException {
    if (lr.getLRPersistenceId() != null) {
        // lock the LR ID so we don't write to the file while an
        // indexer task is reading it
        Object lock = lockObjectForID(lr.getLRPersistenceId());
        synchronized (lock) {
            // we load the copy of this LR and check if any modification were done
            // if so, it should be reindexed or else it should not be synced again.
            LanguageResource copy = null;
            try {
                copy = getLr(lr.getClass().getName(), lr.getLRPersistenceId());
                // we check it only if it is an instance of Document
                if (copy instanceof Document && lr instanceof Document) {
                    Document cDoc = (Document) copy;
                    Document lrDoc = (Document) lr;
                    boolean sameDocs = false;
                    // as that's what matters from the annic perspective
                    if (cDoc.getContent().equals(lrDoc.getContent())) {
                        if (cDoc.getAnnotations().equals(lrDoc.getAnnotations())) {
                            if (cDoc.getNamedAnnotationSets().equals(lrDoc.getNamedAnnotationSets())) {
                                boolean allSetsSame = true;
                                for (String key : cDoc.getNamedAnnotationSets().keySet()) {
                                    if (!cDoc.getAnnotations(key).equals(lrDoc.getAnnotations(key))) {
                                        allSetsSame = false;
                                        break;
                                    }
                                }
                                if (allSetsSame) {
                                    sameDocs = true;
                                }
                            }
                        }
                    }
                    if (sameDocs) {
                        lock = null;
                        return;
                    }
                }
            } catch (SecurityException e) {
                e.printStackTrace();
            } finally {
                // delete the copy of this LR
                if (copy != null) {
                    Factory.deleteResource(copy);
                }
            }
            super.sync(lr);
        }
        lock = null;
    } else {
        super.sync(lr);
    }
    if (lr instanceof Document) {
        queueForIndexing(lr.getLRPersistenceId());
    }
}

Example 9 with LanguageResource

use of gate.LanguageResource in project gate-core by GateNLP.

the class SerialDataStore method getLr.

// constructPersistenceId
@Override
public LanguageResource getLr(String lrClassName, Object lrPersistenceId) throws PersistenceException, SecurityException {
    // find the subdirectory for resources of this type
    File resourceTypeDirectory = new File(storageDir, lrClassName);
    if ((!resourceTypeDirectory.exists()) || (!resourceTypeDirectory.isDirectory())) {
        throw new PersistenceException("Can't find " + resourceTypeDirectory);
    }
    // create a File to representing the resource storage file
    File resourceFile = new File(resourceTypeDirectory, lrPersistenceId.toString());
    if (!resourceFile.exists() || !resourceFile.isFile())
        throw new PersistenceException("Can't find file " + resourceFile);
    // try and read the file and deserialise it
    LanguageResource lr = null;
    try {
        InputStream is = new FileInputStream(resourceFile);
        // after 1.1 the serialised files are compressed
        if (!currentProtocolVersion.equals("1.0"))
            is = new GZIPInputStream(is);
        is = new BufferedInputStream(is);
        // Use an input stream that is aware of the GATE classloader
        ObjectInputStream ois = new GateAwareObjectInputStream(is);
        lr = (LanguageResource) ois.readObject();
        ois.close();
    } catch (IOException e) {
        throw new PersistenceException("Couldn't read file " + resourceFile + ": " + e);
    } catch (ClassNotFoundException ee) {
        throw new PersistenceException("Couldn't find class " + lrClassName + ": " + ee);
    }
    // set the dataStore property of the LR (which is transient and therefore
    // not serialised)
    lr.setDataStore(this);
    lr.setLRPersistenceId(lrPersistenceId);
    if (DEBUG)
        Out.prln("LR read in memory: " + lr);
    return lr;
}

Example 10 with LanguageResource

use of gate.LanguageResource in project gate-core by GateNLP.

the class TestCreole method testClassIndex.

// testLoading()
/**
 * Test resource indexing by class
 */
public void testClassIndex() throws Exception {
    ResourceData docRd = reg.get("gate.corpora.DocumentImpl");
    assertNotNull("couldn't find document res data", docRd);
    assertTrue("doc res data has wrong class name", docRd.getClassName().equals("gate.corpora.DocumentImpl"));
    assertTrue("doc res data has wrong interface name", docRd.getInterfaceName().equals("gate.Document"));
    Class<?> docClass = docRd.getResourceClass();
    assertNotNull("couldn't get doc class", docClass);
    LanguageResource docRes = (LanguageResource) docClass.newInstance();
    assertTrue("instance of doc is wrong type", docRes instanceof gate.Document);
    reg.clear();
}