Examples with Document gate.Document used on opensource projects

Example 31 with Document

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

the class TestAnnotationMerging method loadDocument.

// tearDown
private Document loadDocument(String path, String name) throws Exception {
    Document doc = Factory.newDocument(new URL(TestDocument.getTestServerName() + path), "UTF-8");
    doc.setName(name);
    return doc;
}

Example 32 with Document

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

the class TestTikaFormats method doTest.

private void doTest(String ext) throws Exception {
    String base = TestDocument.getTestServerName();
    URL url = new URL(base + "tests/tika/tika-test." + ext);
    Document doc = Factory.newDocument(url);
    assertNotNull(doc);
    assertTrue(doc.getContent().toString().indexOf("Testing Tika Format Parsers") != -1);
}

Example 33 with Document

use of gate.Document 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();
}

Example 34 with Document

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

the class TestSgml method testSgmlLoading.

// setUp
public void testSgmlLoading() throws Exception {
    assertTrue(true);
    // create the markupElementsMap map
    Map<String, String> markupElementsMap = null;
    gate.Document doc = null;
    /*
    markupElementsMap = new HashMap();
    // populate it
    markupElementsMap.put ("S","Sentence");
    markupElementsMap.put ("s","Sentence");
    markupElementsMap.put ("W","Word");
    markupElementsMap.put ("w","Word");
    */
    FeatureMap params = Factory.newFeatureMap();
    params.put(Document.DOCUMENT_URL_PARAMETER_NAME, new URL(TestDocument.getTestServerName() + "tests/sgml/Hds.sgm"));
    params.put(Document.DOCUMENT_MARKUP_AWARE_PARAMETER_NAME, "false");
    doc = (Document) Factory.createResource("gate.corpora.DocumentImpl", params);
    // get the docFormat that deals with it.
    // the parameter MimeType doesn't affect right now the behaviour
    // *
    gate.DocumentFormat docFormat = gate.DocumentFormat.getDocumentFormat(doc, doc.getSourceUrl());
    assertTrue("Bad document Format was produced. SgmlDocumentFormat was expected", docFormat instanceof gate.corpora.SgmlDocumentFormat);
    // set's the map
    docFormat.setMarkupElementsMap(markupElementsMap);
    docFormat.unpackMarkup(doc, "DocumentContent");
    AnnotationSet annotSet = doc.getAnnotations(GateConstants.ORIGINAL_MARKUPS_ANNOT_SET_NAME);
    assertEquals("For " + doc.getSourceUrl() + " the number of annotations" + " should be:1022", 1022, annotSet.size());
    // Verfy if all annotations from the default annotation set are consistent
    gate.corpora.TestDocument.verifyNodeIdConsistency(doc);
}

Example 35 with Document

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

the class TestXml method runCompleteTestWithAFormat.

// testGateDocumentToAndFromXmlWithDifferentKindOfFormats
private void runCompleteTestWithAFormat(URL url, String urlDescription) throws Exception {
    // Load the xml Key Document and unpack it
    gate.Document keyDocument = null;
    FeatureMap params = Factory.newFeatureMap();
    params.put(Document.DOCUMENT_URL_PARAMETER_NAME, url);
    params.put(Document.DOCUMENT_MARKUP_AWARE_PARAMETER_NAME, "false");
    keyDocument = (Document) Factory.createResource("gate.corpora.DocumentImpl", params);
    assertTrue("Coudn't create a GATE document instance for " + url.toString() + " Can't continue.", keyDocument != null);
    gate.DocumentFormat keyDocFormat = null;
    keyDocFormat = gate.DocumentFormat.getDocumentFormat(keyDocument, keyDocument.getSourceUrl());
    assertTrue("Fail to recognize " + url.toString() + " as being " + urlDescription + " !", keyDocFormat != null);
    // Unpack the markup
    keyDocFormat.unpackMarkup(keyDocument);
    // Verfy if all annotations from the default annotation set are consistent
    gate.corpora.TestDocument.verifyNodeIdConsistency(keyDocument);
    // Verifies if the maximum annotation ID on the GATE doc is less than the
    // Annotation ID generator of the document.
    verifyAnnotationIDGenerator(keyDocument);
    // Save the size of the document and the number of annotations
    long keyDocumentSize = keyDocument.getContent().size().longValue();
    int keyDocumentAnnotationSetSize = keyDocument.getAnnotations().size();
    // Export the Gate document called keyDocument as  XML, into a temp file,
    // using the working encoding
    File xmlFile = null;
    xmlFile = Files.writeTempFile(keyDocument.toXml(), workingEncoding);
    assertTrue("The temp GATE XML file is null. Can't continue.", xmlFile != null);
    // Load the XML Gate document form the tmp file into memory
    gate.Document gateDoc = null;
    gateDoc = gate.Factory.newDocument(xmlFile.toURI().toURL(), workingEncoding);
    assertTrue("Coudn't create a GATE document instance for " + xmlFile.toURI().toURL().toString() + " Can't continue.", gateDoc != null);
    gate.DocumentFormat gateDocFormat = null;
    gateDocFormat = DocumentFormat.getDocumentFormat(gateDoc, gateDoc.getSourceUrl());
    assertTrue("Fail to recognize " + xmlFile.toURI().toURL().toString() + " as being a GATE XML document !", gateDocFormat != null);
    gateDocFormat.unpackMarkup(gateDoc);
    // Verfy if all annotations from the default annotation set are consistent
    gate.corpora.TestDocument.verifyNodeIdConsistency(gateDoc);
    // Save the size of the document snd the number of annotations
    long gateDocSize = keyDocument.getContent().size().longValue();
    int gateDocAnnotationSetSize = keyDocument.getAnnotations().size();
    assertTrue("Exporting as GATE XML resulted in document content size lost." + " Something went wrong.", keyDocumentSize == gateDocSize);
    assertTrue("Exporting as GATE XML resulted in annotation lost." + " No. of annotations missing =  " + Math.abs(keyDocumentAnnotationSetSize - gateDocAnnotationSetSize), keyDocumentAnnotationSetSize == gateDocAnnotationSetSize);
    // Verifies if the maximum annotation ID on the GATE doc is less than the
    // Annotation ID generator of the document.
    verifyAnnotationIDGenerator(gateDoc);
    // Don't need tmp Gate XML file.
    xmlFile.delete();
}