Mercurial > public > html2wiki

annotate src/org/nwoca/ssdt/tools/html2wiki/Html2Wiki.java @ 4:22ed6d93442c

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Start modifying transformers to Confluence wiki syntax
author smith@nwoca.org
date Tue, 25 Jan 2011 21:59:31 -0500
parents 5da2e67620f9
children d34f4d408ef9
rev   line source
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
1 package org.nwoca.ssdt.tools.html2wiki;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
2 /*
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
3 * Html2Wiki.java
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
4 *
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
5 * Created on May 9, 2006, 3:22 PM
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
6 *
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
7 */
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
8
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
9 import java.io.*;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
10 import java.util.Collection;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
11 import java.util.ArrayList;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
12 import java.util.List;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
13 import java.util.Iterator;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
14 import org.apache.commons.io.FileSystemUtils;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
15 import org.apache.commons.io.FileUtils;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
16 import java.util.regex.*;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
17 import org.apache.commons.io.FilenameUtils;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
18
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
19 /**
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
20 * Converter to convert HTML documents into MediaWiki test.
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
21 *
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
22 * Heavily customized to handle HTML produced by DEC DOCUMENT
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
23 * SOFTARE doctype. Breaks file into Chapters in the manner done
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
24 * by Document. Needs modification to work with other HTML files.
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
25 *
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
26 * @author SMITH
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
27 */
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
28 public class Html2Wiki {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
29
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
30 private StringBuffer buffer;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
31 private Collection<Transformer> transformers;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
32 private boolean converted = false;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
33 private static String category;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
34
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
35 /** Creates a new instance of Html2Wiki. */
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
36 public Html2Wiki(String html) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
37 buffer = new StringBuffer(html);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
38 transformers = new ArrayList<Transformer>();
4
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
39 // transformers.add(new PreTagTransformer());
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
40 // transformers.add(new DeleteTransformer("^\\s",true));
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
41 transformers.add(new DeleteTransformer("<html>|</html>|<body>|</body>"));
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
42 transformers.add(new DeleteTransformer("<!--.*-->(\\n|\\r)*",true));
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
43 transformers.add(new DeleteTransformer("<a .*?>|</a>"));
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
44 transformers.add(new DeleteTransformer("(?m)^\\*"));
4
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
45 // transformers.add(new DeleteTransformer("<blockquote>|</blockquote>"));
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
46 transformers.add(new DeleteTransformer("<p>"));
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
47 transformers.add(new DeleteTransformer("(?m)<br>$"));
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
48 transformers.add(new DeleteTransformer("<font .*?>|</font>"));
4
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
49 transformers.add(new CloseTagTransformer("<li>","(\n|\r)*(<li>|</ul>|</ol>|<ul>|<ol>)","</li>"));
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
50 transformers.add(new BadTableDataTransformer());
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
51 transformers.add(new BadTableRowTransformer());
4
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
52 // transformers.add(new ReplaceTransformer("</td>","\n</td>"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
53 transformers.add(new ReplaceTransformer("\\{","\\{"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
54 transformers.add(new ReplaceTransformer("\\}","\\}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
55 // transformers.add(new ReplaceTransformer("\\[","\\["));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
56 // transformers.add(new ReplaceTransformer("\\]","\\]"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
57 transformers.add(new ReplaceTransformer("<br>","\\\\"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
58 transformers.add(new ReplaceTransformer("<table.*?>|</table>","{table}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
59 transformers.add(new ReplaceTransformer("<tr>|</tr>","{tr}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
60 transformers.add(new ReplaceTransformer("<td>|</td>","{td}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
61 transformers.add(new ReplaceTransformer("<th>|</th>","{th}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
62 transformers.add(new ReplaceTransformer("<ol.*?>|</ol>","{ol}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
63 transformers.add(new ReplaceTransformer("<ul.*?>|</ul>","{ul}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
64 transformers.add(new ReplaceTransformer("<li>","{li}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
65 transformers.add(new ReplaceTransformer("</li>","{li}\n"));
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
66 transformers.add(new ChapterTransformer(category));
4
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
67 transformers.add(new TagTransformer("<pre>(.*?)</pre>", true, "{code}","{code}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
68 transformers.add(new TagTransformer("<center>(.*?)</center>", true, "{center}","{center}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
69 transformers.add(new TagTransformer("<em>(.*?)</em>", "*","*"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
70 transformers.add(new TagTransformer("<strong>(.*?)</strong>", "*","*"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
71 transformers.add(new TagTransformer("(?s)<kbd>(.*?)</kbd>", "{{", "}}"));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
72 transformers.add(new TagTransformer("<h1>(.*)</h1>", "h1. ", ""));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
73 transformers.add(new TagTransformer("<h2>(.*)</h2>", "h2. ", ""));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
74 transformers.add(new TagTransformer("<h3>(accessing the program|sample run|sample screens?|sample reports?)</[h|H]3>","h3.", ""));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
75 transformers.add(new TagTransformer("<h3>(.*)</H3>", "h3. ", ""));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
76 transformers.add(new TagTransformer("<h3>(.*)</h3>", "h3. ", ""));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
77 transformers.add(new TagTransformer("<h4>(.*)</h4>", "h4. ", ""));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
78 transformers.add(new TagTransformer("<h5>(.*)</h5>", "h5. ", ""));
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
79 transformers.add(new TagTransformer("<h6>(.*)</h6>", "h6. ", ""));
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
80 transformers.add(new DeleteTransformer("(?s)<hr.*?>"));
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
81
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
82 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
83
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
84 /**
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
85 * @param args the command line arguments
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
86 */
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
87 public static void main(String[] args) throws IOException {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
88
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
89 if (args.length == 0) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
90 System.out.println("Usage:");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
91 System.out.println(" Html2Wiki {inputDirectory} [Category]");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
92 System.out.println(" default is current directory");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
93 System.out.println(" Processes all *.html files. ");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
94 System.out.println(" Each 'chapter' written to *.wiki");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
95 return;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
96 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
97
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
98 File inputs = new File(args[0]);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
99
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
100 if (args.length > 1) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
101 category = args[1];
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
102 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
103
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
104 File[] inputFiles = inputs.listFiles(new HtmlFileFilter());
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
105 for (int i = 0; i < inputFiles.length; i++) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
106
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
107 process(inputFiles[i]);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
108
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
109 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
110
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
111 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
112
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
113 protected static void process(File input) throws IOException {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
114
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
115 System.out.println(input.getAbsoluteFile());
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
116
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
117 Html2Wiki converter = new Html2Wiki(FileUtils.readFileToString(input,null));
2
5da2e67620f9 Upgrade to Ivy configuration and begin clean up of tests. Added FreeBSD license.
smith@nwoca.org
parents: 0
diff changeset
118
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
119 WikiChapter[] chapters = converter.getWikiChapters();
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
120
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
121 System.out.format("Writing %d wiki files...\n",chapters.length);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
122
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
123 StringBuffer wikiIndex = new StringBuffer();
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
124 wikiIndex.append("Contents:\n\n");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
125
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
126 for (int i = 0; i < chapters.length; i++) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
127
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
128 wikiIndex.append("# [[" + chapters[i].getChapterName() + "]]\n");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
129 FileUtils.writeStringToFile(new File(input.getParent(),
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
130 generateFilename(chapters[i].getChapterName())+".wiki"),
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
131 chapters[i].getContents().toString(),
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
132 null);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
133
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
134 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
135 System.out.println("Writing wikiIndex...");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
136
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
137 FileUtils.writeStringToFile(new File(FilenameUtils.removeExtension(input.getPath())+".wikiIndex"),wikiIndex.toString(),null);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
138 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
139
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
140 public static String generateFilename(String input) {
4
22ed6d93442c Start modifying transformers to Confluence wiki syntax
smith@nwoca.org
parents: 2
diff changeset
141 return input.replaceAll("\\\\|/|:|\\(|\\)","-").replace("<br>", "");
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
142
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
143 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
144 public String getWikiText() {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
145 convert();
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
146 return buffer.toString();
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
147 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
148
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
149 public WikiChapter[] getWikiChapters() {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
150
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
151 convert();
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
152
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
153 List<WikiChapter> chapters = new ArrayList<WikiChapter>();
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
154
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
155 Pattern chapterPat = Pattern.compile("<chapter>");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
156 Matcher begin = chapterPat.matcher(buffer);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
157 Matcher end = chapterPat.matcher(buffer);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
158
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
159 while(begin.find()) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
160
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
161
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
162 end.find(begin.end());
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
163
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
164 Pattern chapterNamePat = Pattern.compile("<chapter>(.*?)</chapter>");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
165
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
166 Matcher chapterNameMatcher = chapterNamePat.matcher(buffer);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
167
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
168 String chapterName = chapterNameMatcher.find(begin.start()) ? chapterNameMatcher.group(1) : null;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
169
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
170 CharSequence contents = buffer.subSequence(chapterName == null ? begin.start() : chapterNameMatcher.end()
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
171 ,end.hitEnd() ? buffer.length() : end.start());
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
172
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
173 chapters.add(new WikiChapter(chapterName,contents));
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
174
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
175 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
176 return (WikiChapter[])chapters.toArray(new WikiChapter[]{});
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
177 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
178
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
179 private void convert() {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
180
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
181 if(!converted) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
182 for (Transformer t : transformers) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
183
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
184 System.out.println(".Applying: " + t);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
185 t.apply(buffer);
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
186
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
187 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
188 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
189 converted = true;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
190 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
191
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
192 private static class HtmlFileFilter implements FileFilter {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
193 public boolean accept(File pathname) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
194 return pathname.getName().toLowerCase().matches("^.*\\.html$");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
195 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
196
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
197 }
2
5da2e67620f9 Upgrade to Ivy configuration and begin clean up of tests. Added FreeBSD license.
smith@nwoca.org
parents: 0
diff changeset
198 protected static class WikiChapter {
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
199 private String chapterName;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
200 private CharSequence contents;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
201
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
202 public WikiChapter(String chapterName, CharSequence contents) {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
203 this.chapterName = chapterName.replaceAll("\\\\|/|:|\\(|\\)","-").replaceAll("\\s+"," ").replaceAll("&amp;","and");
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
204
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
205 this.contents = contents;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
206 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
207
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
208 public String getChapterName() {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
209 return chapterName;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
210 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
211
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
212 public CharSequence getContents() {
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
213 return contents;
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
214 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
215
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
216 public String toString() {
2
5da2e67620f9 Upgrade to Ivy configuration and begin clean up of tests. Added FreeBSD license.
smith@nwoca.org
parents: 0
diff changeset
217 return "Chapter: " + chapterName + " Content length: " + contents.length();
0
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
218 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
219 }
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
220
f8b1ea49d065 Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
smith@nwoca.org
parents:
diff changeset
221 }