001 /*
002 * TestHtml.java
003 *
004 * Copyright (c) 1995-2010, The University of Sheffield. See the file
005 * COPYRIGHT.txt in the software or at http://gate.ac.uk/gate/COPYRIGHT.txt
006 *
007 * This file is part of GATE (see http://gate.ac.uk/), and is free
008 * software, licenced under the GNU Library General Public License,
009 * Version 2, June 1991 (in the distribution as file licence.html,
010 * and also available at http://gate.ac.uk/gate/licence.html).
011 *
012 * Cristian URSU, 8/May/2000
013 *
014 * $Id: TestHtml.java 12006 2009-12-01 17:24:28Z thomas_heitz $
015 */
016
017 package gate.html;
018
019 import java.util.Map;
020
021 import junit.framework.*;
022
023 import gate.Gate;
024
025
026 /** Test class for HTML facilities
027 */
028 public class TestHtml extends TestCase
029 {
030 /** Debug flag */
031 private static final boolean DEBUG = false;
032
033 /** Construction */
034 public TestHtml(String name) { super(name); }
035
036 /** Fixture set up */
037 public void setUp() {
038 } // setUp
039
040 /** A test */
041 public void testUnpackMarkup() throws Exception {
042 // create the markupElementsMap map
043 Map markupElementsMap = null;
044
045 gate.Document doc = null;
046 /*
047 markupElementsMap = new HashMap();
048 // populate it
049 markupElementsMap.put ("h1","Header 1");
050 markupElementsMap.put ("H1","Header 1");
051 markupElementsMap.put ("A","link");
052 markupElementsMap.put ("a","link");
053 */
054 doc = gate.Factory.newDocument(Gate.getUrl("tests/html/test1.htm"));
055 // doc = gate.Factory.newDocument(new URL("http://www"));
056
057 // get the docFormat that deals with it.
058 gate.DocumentFormat docFormat = gate.DocumentFormat.getDocumentFormat(
059 doc, doc.getSourceUrl()
060 );
061 assertTrue( "Bad document Format was produced. HtmlDocumentFormat was expected",
062 docFormat instanceof gate.corpora.HtmlDocumentFormat
063 );
064
065
066 // set's the map
067 // Don't need to unpack markup explicitly, as it is already unpacked by
068 // default by newDocument - unpacking it twice causes exceptions
069 //docFormat.setMarkupElementsMap(markupElementsMap);
070 //docFormat.unpackMarkup (doc,"DocumentContent");
071
072 gate.corpora.TestDocument.verifyNodeIdConsistency(doc);
073 /*
074 // Save it as XML
075 File xmlFile = null;
076 xmlFile = Files.writeTempFile(null);
077
078 OutputStreamWriter writer = new OutputStreamWriter(
079 new FileOutputStream(xmlFile),"UTF-8");
080 // Write (test the toXml() method)
081 writer.write(doc.toXml());
082 writer.flush();
083 writer.close();
084 */
085 } // testUnpackMarkup()
086 //*
087 public static void main(String[] args){
088 try{
089 Gate.init();
090 TestHtml test = new TestHtml("gicu");
091 test.testUnpackMarkup();
092 } catch (Exception e){
093 e.printStackTrace(System.out);
094 }
095 }
096 //*/
097 /** Test suite routine for the test runner */
098 public static Test suite() {
099 return new TestSuite(TestHtml.class);
100 } // suite
101
102 }//class TestHtml
|