1 7 8 package sample; 9 10 import org.cyberneko.html.HTMLConfiguration; 11 import org.cyberneko.html.filters.ElementRemover; 12 13 import org.apache.xerces.xni.parser.XMLDocumentFilter; 14 import org.apache.xerces.xni.parser.XMLInputSource; 15 import org.apache.xerces.xni.parser.XMLParserConfiguration; 16 17 25 public class RemoveElements { 26 27 31 32 public static void main(String [] argv) throws Exception { 33 34 ElementRemover remover = new ElementRemover(); 36 37 remover.acceptElement("b", null); 39 remover.acceptElement("i", null); 40 remover.acceptElement("u", null); 41 remover.acceptElement("a", new String [] { "href" }); 42 43 remover.removeElement("script"); 45 46 org.cyberneko.html.filters.Writer writer = 48 new org.cyberneko.html.filters.Writer(); 49 50 XMLDocumentFilter[] filters = { 52 remover, 53 writer, 54 }; 55 56 XMLParserConfiguration parser = new HTMLConfiguration(); 58 parser.setProperty("http://cyberneko.org/html/properties/filters", filters); 59 60 for (int i = 0; i < argv.length; i++) { 62 String systemId = argv[i]; 63 XMLInputSource source = new XMLInputSource(null, systemId, null); 64 parser.parse(source); 65 } 66 67 } 69 } | Popular Tags |