src/net/pterodactylus/arachne/parser/HtmlEditorKitParser.java

   1 /*
   2  * © 2009 David ‘Bombe’ Roden
   3  */
   4 package net.pterodactylus.arachne.parser;
   5
   6 import java.io.BufferedReader;
   7 import java.io.IOException;
   8 import java.io.InputStream;
   9 import java.io.InputStreamReader;
  10
  11 import javax.swing.text.MutableAttributeSet;
  12 import javax.swing.text.html.HTML;
  13 import javax.swing.text.html.HTMLEditorKit;
  14 import javax.swing.text.html.HTML.Tag;
  15 import javax.swing.text.html.HTMLEditorKit.ParserCallback;
  16 import javax.swing.text.html.parser.ParserDelegator;
  17
  18 import de.ina.util.io.Closer;
  19
  20 /**
  21  * {@link Parser} implementation based on Swing’s {@link HTMLEditorKit}. This
  22  * parser can be re-used but is not thread-safe!
  23  *
  24  * @author David ‘Bombe’ Roden <bombe@pterodactylus.net>
  25  */
  26 public class HtmlEditorKitParser extends HTMLEditorKit.ParserCallback implements Parser {
  27
  28         /** The parser listener. */
  29         private ParserListener parserListener;
  30
  31         /** The current input stream. */
  32         private InputStream inputStream;
  33
  34         /** Whether we’re currently parsing the title. */
  35         private boolean inTitle;
  36
  37         /** The title text. */
  38         private String titleText;
  39
  40         /** Whether we’re currently parsing a link. */
  41         private boolean inLink;
  42
  43         /** The target of the link. */
  44         private String linkTarget;
  45
  46         /** The title attribute of the link. */
  47         private String linkTitle;
  48
  49         /** The text of the link. */
  50         private StringBuilder linkText;
  51
  52         /**
  53          * {@inheritdoc}
  54          *
  55          * @see net.pterodactylus.arachne.parser.Parser#parse(ParserListener,
  56          *      InputStream, String)
  57          */
  58         public void parse(ParserListener parserListener, InputStream inputStream, String charset) throws IOException {
  59                 this.parserListener = parserListener;
  60                 this.inputStream = inputStream;
  61                 InputStreamReader inputStreamReader = null;
  62                 BufferedReader htmlReader = null;
  63                 try {
  64                         inputStreamReader = new InputStreamReader(inputStream, charset);
  65                         htmlReader = new BufferedReader(inputStreamReader);
  66                         new ParserDelegator().parse(htmlReader, this, true);
  67                 } finally {
  68                         Closer.close(htmlReader);
  69                         Closer.close(inputStreamReader);
  70                 }
  71         }
  72
  73         //
  74         // METHODS FROM ParserCallback
  75         //
  76
  77         /**
  78          * {@inheritDoc}
  79          */
  80         @Override
  81         public void handleStartTag(Tag tag, MutableAttributeSet attributeSet, int position) {
  82                 if (tag == Tag.TITLE) {
  83                         if (!attributeSet.containsAttribute(ParserCallback.IMPLIED, Boolean.TRUE)) {
  84                                 inTitle = true;
  85                         }
  86                 } else if (tag == Tag.A) {
  87                         String href = (String) attributeSet.getAttribute(HTML.Attribute.HREF);
  88                         if (href != null) {
  89                                 linkTarget = href;
  90                                 inLink = true;
  91                                 linkText = new StringBuilder();
  92                                 linkTitle = (String) attributeSet.getAttribute(HTML.Attribute.TITLE);
  93                         }
  94                 }
  95                 if (inLink) {
  96                         if (tag.breaksFlow()) {
  97                                 linkText.append(' ');
  98                         }
  99                 }
 100         }
 101
 102         /**
 103          * {@inheritDoc}
 104          */
 105         @Override
 106         public void handleText(char[] data, int pos) {
 107                 if (inTitle) {
 108                         titleText = new String(data);
 109                 } else if (inLink) {
 110                         linkText.append(data);
 111                 }
 112         }
 113
 114         /**
 115          * {@inheritDoc}
 116          */
 117         @Override
 118         public void handleEndTag(Tag tag, int position) {
 119                 if (tag == Tag.TITLE) {
 120                         inTitle = false;
 121                         parserListener.parsedTitle(inputStream, titleText);
 122                 } else if (tag == Tag.A) {
 123                         inLink = false;
 124                         parserListener.parsedLink(inputStream, linkTarget, linkTitle, linkText.toString());
 125                 }
 126         }
 127
 128 }