[nekohtml-user] ElementRemover removing too much?

[Alain G. <AGi...@ta...>]

Hi,

I have a test case showing that the ElementRemover seems to get out of sync after having removed an element.

Can someone check that I am using this class correctly?

Regards,
Alain Gilbert

Follows my JUnit test class:

package org.cyberneko.html;

import java.io.IOException;
import java.io.StringReader;
import java.io.StringWriter;

import junit.framework.TestCase;

import org.apache.xerces.xni.parser.XMLInputSource;
import org.apache.xerces.xni.parser.XMLParserConfiguration;
import org.cyberneko.html.filters.ElementRemover;
import org.cyberneko.html.filters.Writer;

public class ElementRemoverTest extends TestCase
{
    private static final String DOCUMENT_FRAGMENT =
        "http://cyberneko.org/html/features/balance-tags/document-fragment";

    private static final String IGNORE_OUTSIDE_CONTENT =
        "http://cyberneko.org/html/features/balance-tags/ignore-outside-content";

    protected static final String NAMES_ELEMS = "http://cyberneko.org/html/properties/names/elems";

    private static final String FILTERS = "http://cyberneko.org/html/properties/filters";
    
    public void testRemoveClosedElement() throws IOException
    {
        String src = "<input type=\"text\">foo</input><span>bof</span>";
        String expected = "<span>bof</span>";
        assertEquals(expected, parseFragment(src));
    }
    
    public void testRemoveEmptyElement() throws IOException
    {
        String src = "<input type=\"text\"/><span>bof</span>";
        String expected = "<span>bof</span>";
        assertEquals(expected, parseFragment(src));
    }
    
    public void testRemoveUnclosedElement() throws IOException
    {
        String src = "<input type=\"text\"><span>bof</span>";
        String expected = "<span>bof</span>";
        assertEquals(expected, parseFragment(src));
    }
    
    private String parseFragment(String pHTMLFragment) throws IOException
    {
        XMLParserConfiguration parser = new HTMLConfiguration();

        // Enable HTML fragments.
        parser.setFeature(DOCUMENT_FRAGMENT, true);
        parser.setFeature(IGNORE_OUTSIDE_CONTENT, true);
        // Element names to lowercase.
        parser.setProperty(NAMES_ELEMS, "lower");
        StringWriter out = new StringWriter();
        ElementRemover remover = new ElementRemover();
        remover.removeElement("input");
        remover.acceptElement("span", null);
        org.apache.xerces.xni.parser.XMLDocumentFilter[] filters =
            {remover, new Writer(out, "UTF-8") };
        parser.setProperty(FILTERS, filters);
        XMLInputSource source = new XMLInputSource(null, "ElementRemoval", null, 
            new StringReader(pHTMLFragment), "UTF-8");
        parser.parse(source);
        return out.toString();
    }

}