Mercurial > public > html2wiki

comparison src/org/nwoca/ssdt/tools/html2wiki/Html2Wiki.java @ 0:f8b1ea49d065

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Initial version of crude HTML to WikiText converter. Customized for converting HTML files from DEC Document into Wiki markup.
author smith@nwoca.org
date Fri, 12 May 2006 16:45:42 -0400
parents
children 5da2e67620f9
comparison
equal deleted inserted replaced
-1:000000000000 0:f8b1ea49d065
1 package org.nwoca.ssdt.tools.html2wiki;
2 /*
3 * Html2Wiki.java
4 *
5 * Created on May 9, 2006, 3:22 PM
6 *
7 */
8
9 import java.io.*;
10 import java.util.Collection;
11 import java.util.ArrayList;
12 import java.util.List;
13 import java.util.Iterator;
14 import org.apache.commons.io.FileSystemUtils;
15 import org.apache.commons.io.FileUtils;
16 import java.util.regex.*;
17 import org.apache.commons.io.FilenameUtils;
18
19 /**
20 * Converter to convert HTML documents into MediaWiki test.
21 *
22 * Heavily customized to handle HTML produced by DEC DOCUMENT
23 * SOFTARE doctype. Breaks file into Chapters in the manner done
24 * by Document. Needs modification to work with other HTML files.
25 *
26 * @author SMITH
27 */
28 public class Html2Wiki {
29
30 private StringBuffer buffer;
31 private Collection<Transformer> transformers;
32 private boolean converted = false;
33 private static String category;
34
35 /** Creates a new instance of Html2Wiki. */
36 public Html2Wiki(String html) {
37 buffer = new StringBuffer(html);
38 transformers = new ArrayList<Transformer>();
39 transformers.add(new PreTagTransformer());
40 transformers.add(new DeleteTransformer("^\\s",true));
41 transformers.add(new DeleteTransformer("<html>|</html>|<body>|</body>"));
42 transformers.add(new DeleteTransformer("<!--.*-->(\\n|\\r)*",true));
43 transformers.add(new DeleteTransformer("<a .*?>|</a>"));
44 transformers.add(new DeleteTransformer("(?m)^\\*"));
45 transformers.add(new DeleteTransformer("<blockquote>|</blockquote>"));
46 transformers.add(new DeleteTransformer("<p>"));
47 transformers.add(new DeleteTransformer("(?m)<br>$"));
48 transformers.add(new DeleteTransformer("<font .*?>|</font>"));
49 transformers.add(new CloseTagTransformer("<li>","(\n|\r)*(<li>|</ul>|</ol>|<ul>|<ol>)","\n</li>"));
50 transformers.add(new BadTableDataTransformer());
51 transformers.add(new BadTableRowTransformer());
52 transformers.add(new ReplaceTransformer("</td>","\n</td>"));
53 transformers.add(new ChapterTransformer(category));
54 transformers.add(new TagTransformer("<em>(.*?)</em>", "''"));
55 transformers.add(new TagTransformer("<strong>(.*?)</strong>", "'''"));
56 transformers.add(new TagTransformer("(?s)<kbd>(.*?)</kbd>", "<tt>", "</tt>"));
57 transformers.add(new TagTransformer("<h1>(.*)</h1>", "== ", " =="));
58 transformers.add(new TagTransformer("<h2>(.*)</h2>", "=== ", " ==="));
59 transformers.add(new TagTransformer("<h3>(accessing the program|sample run|sample screens?|sample reports?)</[h|H]3>","=== ", " ==="));
60 transformers.add(new TagTransformer("<h3>(.*)</H3>", "", ""));
61 transformers.add(new TagTransformer("<h3>(.*)</h3>", "==== ", " ===="));
62 transformers.add(new TagTransformer("<h4>(.*)</h4>", "===== ", " ====="));
63 transformers.add(new TagTransformer("<h5>(.*)</h5>", "====== ", " ======"));
64 transformers.add(new TagTransformer("<h6>(.*)</h6>", "======= ", " ======="));
65 transformers.add(new DeleteTransformer("(?s)<hr.*?>"));
66
67 }
68
69 /**
70 * @param args the command line arguments
71 */
72 public static void main(String[] args) throws IOException {
73
74 if (args.length == 0) {
75 System.out.println("Usage:");
76 System.out.println(" Html2Wiki {inputDirectory} [Category]");
77 System.out.println(" default is current directory");
78 System.out.println(" Processes all *.html files. ");
79 System.out.println(" Each 'chapter' written to *.wiki");
80 return;
81 }
82
83 File inputs = new File(args[0]);
84
85 if (args.length > 1) {
86 category = args[1];
87 }
88
89 File[] inputFiles = inputs.listFiles(new HtmlFileFilter());
90 for (int i = 0; i < inputFiles.length; i++) {
91
92 process(inputFiles[i]);
93
94 }
95
96 }
97
98 protected static void process(File input) throws IOException {
99
100 System.out.println(input.getAbsoluteFile());
101
102 Html2Wiki converter = new Html2Wiki(FileUtils.readFileToString(input,null));
103
104
105 WikiChapter[] chapters = converter.getWikiChapters();
106
107 System.out.format("Writing %d wiki files...\n",chapters.length);
108
109 StringBuffer wikiIndex = new StringBuffer();
110 wikiIndex.append("Contents:\n\n");
111
112 for (int i = 0; i < chapters.length; i++) {
113
114 wikiIndex.append("# [[" + chapters[i].getChapterName() + "]]\n");
115 FileUtils.writeStringToFile(new File(input.getParent(),
116 generateFilename(chapters[i].getChapterName())+".wiki"),
117 chapters[i].getContents().toString(),
118 null);
119
120 }
121 System.out.println("Writing wikiIndex...");
122
123 FileUtils.writeStringToFile(new File(FilenameUtils.removeExtension(input.getPath())+".wikiIndex"),wikiIndex.toString(),null);
124 }
125
126 public static String generateFilename(String input) {
127 return input.replaceAll("\\\\|/|:|\\(|\\)","-");
128
129 }
130 public String getWikiText() {
131 convert();
132 return buffer.toString();
133 }
134
135 public WikiChapter[] getWikiChapters() {
136
137 convert();
138
139 List<WikiChapter> chapters = new ArrayList<WikiChapter>();
140
141 Pattern chapterPat = Pattern.compile("<chapter>");
142 Matcher begin = chapterPat.matcher(buffer);
143 Matcher end = chapterPat.matcher(buffer);
144
145 while(begin.find()) {
146
147
148 end.find(begin.end());
149
150 Pattern chapterNamePat = Pattern.compile("<chapter>(.*?)</chapter>");
151
152 Matcher chapterNameMatcher = chapterNamePat.matcher(buffer);
153
154 String chapterName = chapterNameMatcher.find(begin.start()) ? chapterNameMatcher.group(1) : null;
155
156 CharSequence contents = buffer.subSequence(chapterName == null ? begin.start() : chapterNameMatcher.end()
157 ,end.hitEnd() ? buffer.length() : end.start());
158
159 chapters.add(new WikiChapter(chapterName,contents));
160
161 }
162 return (WikiChapter[])chapters.toArray(new WikiChapter[]{});
163 }
164
165 private void convert() {
166
167 if(!converted) {
168 for (Transformer t : transformers) {
169
170 System.out.println(".Applying: " + t);
171 t.apply(buffer);
172
173 }
174 }
175 converted = true;
176 }
177
178 private static class HtmlFileFilter implements FileFilter {
179 public boolean accept(File pathname) {
180 return pathname.getName().toLowerCase().matches("^.*\\.html$");
181 }
182
183 }
184 private static class WikiChapter {
185 private String chapterName;
186 private CharSequence contents;
187
188 public WikiChapter(String chapterName, CharSequence contents) {
189 this.chapterName = chapterName.replaceAll("\\\\|/|:|\\(|\\)","-").replaceAll("\\s+"," ").replaceAll("&amp;","and");
190
191 this.contents = contents;
192 }
193
194 public String getChapterName() {
195 return chapterName;
196 }
197
198 public CharSequence getContents() {
199 return contents;
200 }
201
202 public String toString() {
203 return "Chapter: " + chapterName + "\nContents: " + contents;
204 }
205 }
206
207 }