annotate src/de/mpiwg/itgroup/eSciDoc/Tools/Html2Text.java @ 0:c6929e63b0b8

first import
author dwinter
date Wed, 24 Nov 2010 16:52:07 +0100
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c6929e63b0b8 first import
dwinter
parents:
diff changeset
1 package de.mpiwg.itgroup.eSciDoc.Tools;
c6929e63b0b8 first import
dwinter
parents:
diff changeset
2
c6929e63b0b8 first import
dwinter
parents:
diff changeset
3 import java.io.IOException;
c6929e63b0b8 first import
dwinter
parents:
diff changeset
4 import java.io.Reader;
c6929e63b0b8 first import
dwinter
parents:
diff changeset
5
c6929e63b0b8 first import
dwinter
parents:
diff changeset
6 import javax.swing.text.html.HTMLEditorKit;
c6929e63b0b8 first import
dwinter
parents:
diff changeset
7 import javax.swing.text.html.parser.ParserDelegator;
c6929e63b0b8 first import
dwinter
parents:
diff changeset
8
c6929e63b0b8 first import
dwinter
parents:
diff changeset
9 //import javax.swing.text.html.*;
c6929e63b0b8 first import
dwinter
parents:
diff changeset
10 //import javax.swing.text.html.parser.*;
c6929e63b0b8 first import
dwinter
parents:
diff changeset
11
c6929e63b0b8 first import
dwinter
parents:
diff changeset
12
c6929e63b0b8 first import
dwinter
parents:
diff changeset
13 public class Html2Text extends HTMLEditorKit.ParserCallback {
c6929e63b0b8 first import
dwinter
parents:
diff changeset
14 StringBuffer s;
c6929e63b0b8 first import
dwinter
parents:
diff changeset
15
c6929e63b0b8 first import
dwinter
parents:
diff changeset
16 public Html2Text() {}
c6929e63b0b8 first import
dwinter
parents:
diff changeset
17
c6929e63b0b8 first import
dwinter
parents:
diff changeset
18 public void parse(Reader in) throws IOException {
c6929e63b0b8 first import
dwinter
parents:
diff changeset
19 s = new StringBuffer();
c6929e63b0b8 first import
dwinter
parents:
diff changeset
20 ParserDelegator delegator = new ParserDelegator();
c6929e63b0b8 first import
dwinter
parents:
diff changeset
21 // the third parameter is TRUE to ignore charset directive
c6929e63b0b8 first import
dwinter
parents:
diff changeset
22 delegator.parse(in, this, Boolean.TRUE);
c6929e63b0b8 first import
dwinter
parents:
diff changeset
23 }
c6929e63b0b8 first import
dwinter
parents:
diff changeset
24
c6929e63b0b8 first import
dwinter
parents:
diff changeset
25 public void handleText(char[] text, int pos) {
c6929e63b0b8 first import
dwinter
parents:
diff changeset
26 s.append(text);
c6929e63b0b8 first import
dwinter
parents:
diff changeset
27 }
c6929e63b0b8 first import
dwinter
parents:
diff changeset
28
c6929e63b0b8 first import
dwinter
parents:
diff changeset
29 public String getText() {
c6929e63b0b8 first import
dwinter
parents:
diff changeset
30 return s.toString();
c6929e63b0b8 first import
dwinter
parents:
diff changeset
31 }
c6929e63b0b8 first import
dwinter
parents:
diff changeset
32
c6929e63b0b8 first import
dwinter
parents:
diff changeset
33 }
c6929e63b0b8 first import
dwinter
parents:
diff changeset
34