Mercurial > projects > dwt2

annotate org.eclipse.jface.text/src/org/eclipse/jface/internal/text/html/HTML2TextReader.d @ 16:dbfb303e8fb0

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
first complete successful compile (win-only)
author Frank Benoit <benoit@tionex.de>
date Wed, 18 Mar 2009 08:56:47 +0100
parents bc29606a740c
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
12
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
1 /*******************************************************************************
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
2 * Copyright (c) 2000, 2008 IBM Corporation and others.
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
3 * All rights reserved. This program and the accompanying materials
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
4 * are made available under the terms of the Eclipse Public License v1.0
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
5 * which accompanies this distribution, and is available at
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
6 * http://www.eclipse.org/legal/epl-v10.html
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
7 *
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
8 * Contributors:
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
9 * IBM Corporation - initial API and implementation
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
10 * Port to the D programming language:
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
11 * Frank Benoit <benoit@tionex.de>
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
12 *******************************************************************************/
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
13 module org.eclipse.jface.internal.text.html.HTML2TextReader;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
14
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
15 import org.eclipse.jface.internal.text.html.HTMLPrinter; // packageimport
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
16 import org.eclipse.jface.internal.text.html.BrowserInformationControl; // packageimport
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
17 import org.eclipse.jface.internal.text.html.SubstitutionTextReader; // packageimport
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
18 import org.eclipse.jface.internal.text.html.HTMLTextPresenter; // packageimport
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
19 import org.eclipse.jface.internal.text.html.BrowserInput; // packageimport
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
20 import org.eclipse.jface.internal.text.html.SingleCharReader; // packageimport
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
21 import org.eclipse.jface.internal.text.html.BrowserInformationControlInput; // packageimport
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
22 import org.eclipse.jface.internal.text.html.HTMLMessages; // packageimport
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
23
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
24 import java.lang.all;
16
dbfb303e8fb0 first complete successful compile (win-only)
Frank Benoit <benoit@tionex.de>
parents: 12
diff changeset
25 import java.io.Reader;
12
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
26 import java.util.Map;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
27 import java.util.HashMap;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
28 import java.util.Set;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
29 import java.util.HashSet;
16
dbfb303e8fb0 first complete successful compile (win-only)
Frank Benoit <benoit@tionex.de>
parents: 12
diff changeset
30 import java.io.PushbackReader;
12
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
31 static import tango.text.convert.Utf;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
32
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
33 import org.eclipse.swt.SWT;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
34 import org.eclipse.swt.custom.StyleRange;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
35 import org.eclipse.jface.text.TextPresentation;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
36
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
37
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
38 /**
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
39 * Reads the text contents from a reader of HTML contents and translates
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
40 * the tags or cut them out.
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
41 * <p>
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
42 * Moved into this package from <code>org.eclipse.jface.internal.text.revisions</code>.</p>
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
43 */
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
44 public class HTML2TextReader : SubstitutionTextReader {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
45
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
46 private static const String EMPTY_STRING= ""; //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
47 private static Map fgEntityLookup_;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
48 private static Set fgTags_;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
49 private static Map fgEntityLookup(){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
50 if( fgEntityLookup_ is null ){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
51 synchronized(HTML2TextReader.classinfo ){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
52 if( fgEntityLookup_ is null ){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
53 fgEntityLookup_= new HashMap(7);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
54 fgEntityLookup_.put("lt", "<"); //$NON-NLS-1$ //$NON-NLS-2$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
55 fgEntityLookup_.put("gt", ">"); //$NON-NLS-1$ //$NON-NLS-2$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
56 fgEntityLookup_.put("nbsp", " "); //$NON-NLS-1$ //$NON-NLS-2$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
57 fgEntityLookup_.put("amp", "&"); //$NON-NLS-1$ //$NON-NLS-2$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
58 fgEntityLookup_.put("circ", "^"); //$NON-NLS-1$ //$NON-NLS-2$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
59 fgEntityLookup_.put("tilde", "~"); //$NON-NLS-2$ //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
60 fgEntityLookup_.put("quot", "\""); //$NON-NLS-1$ //$NON-NLS-2$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
61 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
62 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
63 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
64 return fgEntityLookup_;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
65 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
66 private static Set fgTags(){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
67 if( fgTags_ is null ){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
68 synchronized(HTML2TextReader.classinfo ){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
69 if( fgTags_ is null ){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
70 fgTags_= new HashSet();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
71 fgTags_.add("b"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
72 fgTags_.add("br"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
73 fgTags_.add("br/"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
74 fgTags_.add("div"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
75 fgTags_.add("h1"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
76 fgTags_.add("h2"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
77 fgTags_.add("h3"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
78 fgTags_.add("h4"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
79 fgTags_.add("h5"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
80 fgTags_.add("p"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
81 fgTags_.add("dl"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
82 fgTags_.add("dt"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
83 fgTags_.add("dd"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
84 fgTags_.add("li"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
85 fgTags_.add("ul"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
86 fgTags_.add("pre"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
87 fgTags_.add("head"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
88 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
89 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
90 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
91 return fgTags_;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
92 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
93
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
94 private int fCounter= 0;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
95 private TextPresentation fTextPresentation;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
96 private int fBold= 0;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
97 private int fStartOffset= -1;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
98 private bool fInParagraph= false;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
99 private bool fIsPreformattedText= false;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
100 private bool fIgnore= false;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
101 private bool fHeaderDetected= false;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
102
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
103 /**
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
104 * Transforms the HTML text from the reader to formatted text.
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
105 *
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
106 * @param reader the reader
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
107 * @param presentation If not <code>null</code>, formattings will be applied to
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
108 * the presentation.
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
109 */
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
110 public this(Reader reader, TextPresentation presentation) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
111 super(new PushbackReader(reader));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
112 fTextPresentation= presentation;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
113 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
114
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
115 public int read() {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
116 int c= super.read();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
117 if (c !is -1)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
118 ++ fCounter;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
119 return c;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
120 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
121
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
122 protected void startBold() {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
123 if (fBold is 0)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
124 fStartOffset= fCounter;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
125 ++ fBold;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
126 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
127
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
128 protected void startPreformattedText() {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
129 fIsPreformattedText= true;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
130 setSkipWhitespace(false);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
131 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
132
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
133 protected void stopPreformattedText() {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
134 fIsPreformattedText= false;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
135 setSkipWhitespace(true);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
136 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
137
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
138 protected void stopBold() {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
139 -- fBold;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
140 if (fBold is 0) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
141 if (fTextPresentation !is null) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
142 fTextPresentation.addStyleRange(new StyleRange(fStartOffset, fCounter - fStartOffset, null, null, SWT.BOLD));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
143 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
144 fStartOffset= -1;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
145 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
146 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
147
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
148 /*
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
149 * @see org.eclipse.jdt.internal.ui.text.SubstitutionTextReader#computeSubstitution(int)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
150 */
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
151 protected String computeSubstitution(int c) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
152
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
153 if (c is '<')
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
154 return processHTMLTag();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
155 else if (fIgnore)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
156 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
157 else if (c is '&')
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
158 return processEntity();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
159 else if (fIsPreformattedText)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
160 return processPreformattedText(c);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
161
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
162 return null;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
163 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
164
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
165 private String html2Text(String html) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
166
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
167 if (html is null || html.length() is 0)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
168 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
169
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
170 html= html.toLowerCase();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
171
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
172 String tag= html;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
173 if ('/' is tag.charAt(0))
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
174 tag= tag.substring(1);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
175
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
176 if (!fgTags.contains(tag))
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
177 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
178
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
179
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
180 if ("pre".equals(html)) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
181 startPreformattedText();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
182 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
183 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
184
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
185 if ("/pre".equals(html)) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
186 stopPreformattedText();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
187 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
188 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
189
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
190 if (fIsPreformattedText)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
191 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
192
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
193 if ("b".equals(html)) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
194 startBold();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
195 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
196 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
197
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
198 if ((html.length() > 1 && html.charAt(0) is 'h' && Character.isDigit(html.charAt(1))) || "dt".equals(html)) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
199 startBold();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
200 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
201 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
202
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
203 if ("dl".equals(html)) //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
204 return LINE_DELIM;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
205
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
206 if ("dd".equals(html)) //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
207 return "\t"; //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
208
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
209 if ("li".equals(html)) //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
210 // FIXME: this hard-coded prefix does not work for RTL languages, see https://bugs.eclipse.org/bugs/show_bug.cgi?id=91682
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
211 return LINE_DELIM ~ HTMLMessages.getString("HTML2TextReader.listItemPrefix"); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
212
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
213 if ("/b".equals(html)) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
214 stopBold();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
215 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
216 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
217
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
218 if ("p".equals(html)) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
219 fInParagraph= true;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
220 return LINE_DELIM;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
221 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
222
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
223 if ("br".equals(html) || "br/".equals(html) || "div".equals(html)) //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
224 return LINE_DELIM;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
225
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
226 if ("/p".equals(html)) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
227 bool inParagraph= fInParagraph;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
228 fInParagraph= false;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
229 return inParagraph ? EMPTY_STRING : LINE_DELIM;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
230 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
231
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
232 if ((html.startsWith("/h") && html.length() > 2 && Character.isDigit(html.charAt(2))) || "/dt".equals(html)) { //$NON-NLS-1$ //$NON-NLS-2$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
233 stopBold();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
234 return LINE_DELIM;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
235 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
236
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
237 if ("/dd".equals(html)) //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
238 return LINE_DELIM;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
239
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
240 if ("head".equals(html) && !fHeaderDetected) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
241 fHeaderDetected= true;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
242 fIgnore= true;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
243 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
244 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
245
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
246 if ("/head".equals(html) && fHeaderDetected && fIgnore) { //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
247 fIgnore= false;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
248 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
249 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
250
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
251 return EMPTY_STRING;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
252 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
253
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
254 /*
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
255 * A '<' has been read. Process a html tag
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
256 */
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
257 private String processHTMLTag() {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
258
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
259 StringBuffer buf= new StringBuffer();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
260 int ch;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
261 do {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
262
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
263 ch= nextDChar();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
264
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
265 while (ch !is -1 && ch !is '>') {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
266 buf.append(dcharToString(Character.toLowerCase(cast(dchar) ch)));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
267 ch= nextDChar();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
268 if (ch is '"'){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
269 buf.append(dcharToString(Character.toLowerCase(cast(dchar) ch)));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
270 ch= nextDChar();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
271 while (ch !is -1 && ch !is '"'){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
272 buf.append(dcharToString(Character.toLowerCase(cast(dchar) ch)));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
273 ch= nextDChar();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
274 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
275 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
276 if (ch is '<' && !isInComment(buf)) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
277 unreadDChar(ch);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
278 return '<' ~ buf.toString();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
279 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
280 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
281
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
282 if (ch is -1)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
283 return null;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
284
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
285 if (!isInComment(buf) || isCommentEnd(buf)) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
286 break;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
287 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
288 // unfinished comment
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
289 buf.append(dcharToString(cast(dchar) ch));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
290 } while (true);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
291
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
292 return html2Text(buf.toString());
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
293 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
294
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
295 private static bool isInComment(StringBuffer buf) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
296 return buf.length() >= 3 && "!--".equals(buf.slice().substring(0, 3)); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
297 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
298
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
299 private static bool isCommentEnd(StringBuffer buf) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
300 int tagLen= buf.length();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
301 return tagLen >= 5 && "--".equals(buf.slice().substring(tagLen - 2)); //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
302 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
303
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
304 private String processPreformattedText(int c) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
305 if (c is '\r' || c is '\n')
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
306 fCounter++;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
307 return null;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
308 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
309
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
310
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
311 private void unreadDChar(dchar ch) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
312 char[4] buf;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
313 dchar[1] ibuf;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
314 ibuf[0] = ch;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
315 foreach( char c; tango.text.convert.Utf.toString( ibuf[], buf[] )){
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
316 (cast(PushbackReader) getReader()).unread(c);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
317 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
318 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
319
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
320 protected String entity2Text(String symbol) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
321 if (symbol.length() > 1 && symbol.charAt(0) is '#') {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
322 int ch;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
323 try {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
324 if (symbol.charAt(1) is 'x') {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
325 ch= Integer.parseInt(symbol.substring(2), 16);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
326 } else {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
327 ch= Integer.parseInt(symbol.substring(1), 10);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
328 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
329 return dcharToString( cast(dchar)ch);
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
330 } catch (NumberFormatException e) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
331 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
332 } else {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
333 String str= stringcast( fgEntityLookup.get(symbol));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
334 if (str !is null) {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
335 return str;
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
336 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
337 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
338 return "&" ~ symbol; // not found //$NON-NLS-1$
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
339 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
340
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
341 /*
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
342 * A '&' has been read. Process a entity
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
343 */
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
344 private String processEntity() {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
345 StringBuffer buf= new StringBuffer();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
346 int ch= nextDChar();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
347 while (Character.isLetterOrDigit(cast(dchar)ch) || ch is '#') {
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
348 buf.append(dcharToString(cast(dchar) ch));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
349 ch= nextDChar();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
350 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
351
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
352 if (ch is ';')
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
353 return entity2Text(buf.toString());
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
354
16
dbfb303e8fb0 first complete successful compile (win-only)
Frank Benoit <benoit@tionex.de>
parents: 12
diff changeset
355 buf.insert(0, "&");
12
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
356 if (ch !is -1)
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
357 buf.append(dcharToString(cast(dchar) ch));
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
358 return buf.toString();
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
359 }
bc29606a740c Added dwt-addons in original directory structure of eclipse.org
Frank Benoit <benoit@tionex.de>
parents:
diff changeset
360 }