| 1 6 7 package org.contineo.core.text.parser; 8 9 import java.io.File ; 10 import java.io.FileReader ; 11 import java.io.Reader ; 12 import javax.swing.text.html.HTMLEditorKit ; 13 import javax.swing.text.html.parser.ParserDelegator ; 14 import org.apache.log4j.Level; 15 import org.apache.log4j.Logger; 16 import org.contineo.core.LoggingManager; 17 21 public class HTMLParser implements Parser { 22 23 26 private final StringBuffer content = new StringBuffer (); 27 28 32 private Logger logger; 33 34 35 36 public HTMLParser(File file) { 37 logger = LoggingManager.getLogger(this.getClass()); 38 init(file); 39 } 40 41 protected void init(File file) { 42 try { 43 HTMLEditorKit.ParserCallback callback = 44 new HTMLEditorKit.ParserCallback () { 45 public void handleText(char[] data, int pos) { 46 content.append(data); 47 } 48 }; 49 Reader reader = new FileReader (file); 50 new ParserDelegator ().parse(reader, callback, true); 51 } 52 catch (Exception ex) { 53 if (logger.isEnabledFor(Level.ERROR)) 54 logger.error(ex.getMessage()); 55 } 56 } 57 58 62 public StringBuffer getContent() { 63 return content; 64 } 65 66 67 public String getVersion() { 68 return ""; 69 } 70 71 74 public String getAuthor() { 75 return ""; 76 } 77 78 81 public String getSourceDate() { 82 return ""; 83 } 84 85 88 public String getKeywords() { 89 return ""; 90 } 91 92 95 public String getTitle() { 96 return ""; 97 } 98 99 } 100 | Popular Tags |