Examples with Document gate.Document used on opensource projects

Example 26 with Document

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

the class TestXml method testUnpackMarkup.

// runCompleteTestWithAFormat
/**
 * A test
 */
public void testUnpackMarkup() throws Exception {
    // create the markupElementsMap map
    // Map markupElementsMap = null;
    gate.Document doc = null;
    /*
    markupElementsMap = new HashMap();
    // populate it
    markupElementsMap.put ("S","Sentence");
    markupElementsMap.put ("s","Sentence");
    */
    // Create the element2String map
    Map<String, String> anElement2StringMap = new HashMap<String, String>();
    // Populate it
    anElement2StringMap.put("S", "\n");
    anElement2StringMap.put("s", "\n");
    doc = gate.Factory.newDocument(new URL(TestDocument.getTestServerName() + "tests/xml/xces.xml"), workingEncoding);
    AnnotationSet annotSet = doc.getAnnotations(GateConstants.ORIGINAL_MARKUPS_ANNOT_SET_NAME);
    assertEquals("For " + doc.getSourceUrl() + " the number of annotations" + " should be:758", 758, annotSet.size());
    gate.corpora.TestDocument.verifyNodeIdConsistency(doc);
    // Verifies if the maximum annotation ID on the GATE doc is less than the
    // Annotation ID generator of the document.
    verifyAnnotationIDGenerator(doc);
}

Example 27 with Document

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

the class TestAnnic method testAnnicIndexing.

/**
 * Testing the annic indexing functionalities
 *
 * @throws Exception
 */
public void testAnnicIndexing() throws Exception {
    // lets create a corpus
    testCorpus = Factory.newCorpus("TestAnnic");
    File directory = new File(new File(new File(new File(new File(new File(new File(Gate.getGateHome(), "src"), "test"), "gate"), "resources"), "gate.ac.uk"), "tests"), "annic");
    File[] files = directory.listFiles();
    for (int i = 0; i < files.length; i++) {
        if (files[i].isFile()) {
            Document doc = Factory.newDocument(files[i].toURI().toURL(), "ISO-8859-1");
            testCorpus.add(doc);
        }
    }
    AnnicIndexing annicPR = new AnnicIndexing();
    LanguageAnalyser splitter = (LanguageAnalyser) Factory.createResource("gate.creole.splitter.SentenceSplitter");
    splitter.setParameterValue("inputASName", "Key");
    splitter.setParameterValue("outputASName", "Key");
    // splitter.setOutputASName("Key");
    for (int i = 0; i < testCorpus.size(); i++) {
        splitter.setDocument(testCorpus.get(i));
        splitter.execute();
    }
    // index
    annicPR.setAnnotationSetName("Key");
    annicPR.setBaseTokenAnnotationType("Token");
    annicPR.setCorpus(testCorpus);
    annicPR.setIndexUnitAnnotationType("Sentence");
    annicPR.setIndexOutputDirectoryLocation(indexURL.toURI().toURL());
    annicPR.execute();
    Factory.deleteResource(testCorpus);
}

Example 28 with Document

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

the class DynamicRegistrationTest method testDynamicRegistration.

public void testDynamicRegistration() throws Exception {
    Gate.getCreoleRegister().registerPlugin(new Plugin.Component(TestResource.class));
    SerialAnalyserController controller = (SerialAnalyserController) Factory.createResource("gate.creole.SerialAnalyserController", Factory.newFeatureMap(), Factory.newFeatureMap(), "basicRun");
    ProcessingResource testResource = (ProcessingResource) Factory.createResource(TestResource.class.getName());
    controller.add(testResource);
    Corpus corpus = Factory.newCorpus("basicTestCorpus");
    String engText = "This is the cereal shot from gnus.";
    Document doc = Factory.newDocument(engText);
    corpus.add(doc);
    controller.setCorpus(corpus);
    controller.setDocument(doc);
    controller.execute();
}

Example 29 with Document

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

the class TestPersist method testSimple.

// testSaveRestore()
/**
 * Simple test
 */
public void testSimple() 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
    DataStore sds = Factory.createDataStore("gate.persist.SerialDataStore", storageDir.toURI().toURL().toString());
    // check we can get empty lists from empty data stores
    @SuppressWarnings("unused") List<String> lrTypes = sds.getLrTypes();
    // 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());
    // save the document
    Document persDoc = (Document) sds.adopt(doc);
    sds.sync(persDoc);
    // remember the persistence ID for reading back
    // (in the normal case these ids are obtained by DataStore.getLrIds(type))
    Object lrPersistenceId = persDoc.getLRPersistenceId();
    // read the document back
    FeatureMap features = Factory.newFeatureMap();
    features.put(DataStore.LR_ID_FEATURE_NAME, lrPersistenceId);
    features.put(DataStore.DATASTORE_FEATURE_NAME, sds);
    Document doc2 = (Document) Factory.createResource("gate.corpora.DocumentImpl", features);
    // parameters should be different
    // check that the version we read back matches the original
    assertTrue(TestEqual.documentsEqual(persDoc, doc2));
    // delete the datastore
    sds.delete();
}

Example 30 with Document

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

the class TestPersist method testDelete.

// testMultipleLrs()
/**
 * Test LR deletion
 */
public void testDelete() 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");
    if (DEBUG)
        Out.prln("Corpus stored to: " + storageDir.getAbsolutePath());
    storageDir.delete();
    // create and open a serial data store
    SerialDataStore sds = new SerialDataStore();
    sds.setStorageUrl(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());
    // save the document
    Document persDoc = (Document) sds.adopt(doc);
    sds.sync(persDoc);
    // remember the persistence ID for reading back
    // (in the normal case these ids are obtained by DataStore.getLrIds(type))
    Object lrPersistenceId = persDoc.getLRPersistenceId();
    // delete document back
    sds.delete("gate.corpora.DocumentImpl", lrPersistenceId);
    // check that there are no LRs left in the DS
    assertTrue(sds.getLrIds("gate.corpora.DocumentImpl").size() == 0);
    // delete the datastore
    sds.delete();
}