|
0
|
1 package nabble.model;
|
|
|
2
|
|
|
3 import fschmidt.html.Html;
|
|
|
4 import fschmidt.html.HtmlTag;
|
|
|
5 import fschmidt.html.HtmlTextContainer;
|
|
|
6 import fschmidt.util.java.HtmlUtils;
|
|
|
7
|
|
|
8 import java.util.Arrays;
|
|
|
9 import java.util.HashSet;
|
|
|
10 import java.util.List;
|
|
|
11 import java.util.ListIterator;
|
|
|
12 import java.util.Set;
|
|
|
13 import java.util.regex.Matcher;
|
|
|
14 import java.util.regex.Pattern;
|
|
|
15
|
|
|
16
|
|
|
17 final class MessageFormatImpls {
|
|
|
18 private MessageFormatImpls() {} // never
|
|
|
19
|
|
|
20 private static final Pattern urlPtn = Pattern.compile(
|
|
|
21 "\\bhttps?://[^\\s\\p{Z}<>\"{}|\\\\^\\[\\]`]*[^\\s\\p{Z}<>\"{}|\\\\^\\[\\]`~,.?\\)]"
|
|
|
22 );
|
|
|
23
|
|
|
24
|
|
|
25 private static abstract class InputFormat extends Message.Format {
|
|
|
26
|
|
|
27 InputFormat(char code,String name) {
|
|
|
28 super(code,name);
|
|
|
29 }
|
|
|
30
|
|
|
31 protected final String getText(String msg, Message.Source source) {
|
|
|
32 return MessageUtils.htmlToText(MessageUtils.newHtml(msg));
|
|
|
33 }
|
|
|
34
|
|
|
35 protected final String getTextWithoutQuotes(String msg, Message.Source source) {
|
|
|
36 return MessageUtils.getTextWithoutQuotes(msg);
|
|
|
37 }
|
|
|
38
|
|
|
39 protected final String getEmail(String msg,int i) {
|
|
|
40 Html list = MessageUtils.newHtml(msg);
|
|
|
41 return MessageFormatImpls.getEmail(list,i);
|
|
|
42 }
|
|
|
43
|
|
|
44 }
|
|
|
45
|
|
|
46 static final Message.Format TEXT = new InputFormat('t',"text") {
|
|
|
47
|
|
|
48 protected Html parse(String msg,Message.Source source) {
|
|
|
49 Html html = MessageUtils.newHtml(msg);
|
|
|
50 nestIgnoredBlocks(html);
|
|
|
51 encodeUnallowedTags(html);
|
|
|
52 convertLinks(html);
|
|
|
53 textToHtml(html);
|
|
|
54 unnestIgnoredBlocks(html);
|
|
|
55 return html;
|
|
|
56 }
|
|
|
57
|
|
|
58 protected Html parseForMailText(String msg,Message.Source source) {
|
|
|
59 Html html = MessageUtils.newHtml(msg);
|
|
|
60 changeUnallowedTagsToStrings(html);
|
|
|
61 return html;
|
|
|
62 }
|
|
|
63
|
|
|
64 };
|
|
|
65
|
|
|
66 static void convertLinks(Html list) {
|
|
|
67 boolean convertLinks = true;
|
|
|
68 for( ListIterator<Object> i=list.listIterator(); i.hasNext(); ) {
|
|
|
69 Object o = i.next();
|
|
|
70 if( o instanceof HtmlTag ) {
|
|
|
71 HtmlTag tag = (HtmlTag)o;
|
|
|
72 String name = tag.getName().toLowerCase();
|
|
|
73 if( name.equals("a") || name.equals("nabble_a") ) {
|
|
|
74 convertLinks = false;
|
|
|
75 } else if( name.equals("/a") || name.equals("/nabble_a") ) {
|
|
|
76 convertLinks = true;
|
|
|
77 }
|
|
|
78 } else if( convertLinks && o instanceof String ) {
|
|
|
79 String s = (String)o;
|
|
|
80 i.remove();
|
|
|
81 for( Object o2 : convertLinks(s) ) {
|
|
|
82 i.add(o2);
|
|
|
83 }
|
|
|
84 }
|
|
|
85 }
|
|
|
86 }
|
|
|
87
|
|
|
88 static Html convertLinks(String s) {
|
|
|
89 Html list = new Html();
|
|
|
90 Matcher m = urlPtn.matcher(s);
|
|
|
91 int i = 0;
|
|
|
92 while( m.find() ) {
|
|
|
93 String s2 = s.substring(i,m.start());
|
|
|
94 if( s2.length() > 0 )
|
|
|
95 list.add(s2);
|
|
|
96 String url = m.group();
|
|
|
97 HtmlTag tag = new HtmlTag("a");
|
|
|
98 tag.setAttribute("href",HtmlTag.quote(url));
|
|
|
99 list.add(tag);
|
|
|
100 list.add(url);
|
|
|
101 list.add(_a);
|
|
|
102 i = m.end();
|
|
|
103 }
|
|
|
104 String s2 = s.substring(i);
|
|
|
105 if( s2.length() > 0 )
|
|
|
106 list.add(s2);
|
|
|
107 return list;
|
|
|
108 }
|
|
|
109
|
|
|
110 private static final Set<String> asBlock = new HashSet<String>(Arrays.asList(
|
|
|
111 "table",
|
|
|
112 "/table",
|
|
|
113 "thead",
|
|
|
114 "/thead",
|
|
|
115 "tbody",
|
|
|
116 "/tbody",
|
|
|
117 "tr",
|
|
|
118 "/tr",
|
|
|
119 "th",
|
|
|
120 "/th",
|
|
|
121 "td",
|
|
|
122 "/td"
|
|
|
123 ));
|
|
|
124
|
|
|
125 static void textToHtml(Html list) {
|
|
|
126 boolean isBlock = false;
|
|
|
127 int n = list.size();
|
|
|
128 for( int i=0; i<n; i++ ) {
|
|
|
129 Object o = list.get(i);
|
|
|
130 if( o instanceof String ) {
|
|
|
131 String s = (String)o;
|
|
|
132 Html l = textToHtml(s,isBlock);
|
|
|
133 list.remove(i);
|
|
|
134 list.addAll(i,l);
|
|
|
135 int d = l.size() - 1;
|
|
|
136 i += d;
|
|
|
137 n += d;
|
|
|
138 }
|
|
|
139 isBlock = false;
|
|
|
140 if( o instanceof HtmlTag ) {
|
|
|
141 HtmlTag tag = (HtmlTag)o;
|
|
|
142 String name = tag.getName().toLowerCase();
|
|
|
143 if( name.charAt(0) == '/' && Message.htmlSeparators.get(name) == "\n"
|
|
|
144 || asBlock.contains(name)
|
|
|
145 ) {
|
|
|
146 isBlock = true;
|
|
|
147 }
|
|
|
148 } else if( o instanceof HtmlTextContainer ) {
|
|
|
149 isBlock = true;
|
|
|
150 // } else if( o instanceof Html ) {
|
|
|
151 // isBlock = true;
|
|
|
152 }
|
|
|
153 }
|
|
|
154 }
|
|
|
155
|
|
|
156 private static final HtmlTag br = new HtmlTag("br/");
|
|
|
157
|
|
|
158 private static Html textToHtml(String s,boolean isAfterBlock) {
|
|
|
159 s = encodeHtml(s);
|
|
|
160 Html list = new Html();
|
|
|
161 StringBuilder buf = new StringBuilder();
|
|
|
162 final char[] a = s.toCharArray();
|
|
|
163 boolean isLineStart = true;
|
|
|
164 for (char c : a) {
|
|
|
165 switch (c) {
|
|
|
166 case '\r':
|
|
|
167 buf.append( c );
|
|
|
168 break;
|
|
|
169 case '\n':
|
|
|
170 buf.append( c );
|
|
|
171 String t = buf.toString();
|
|
|
172 if( t.trim().length() > 0 )
|
|
|
173 list.add(t);
|
|
|
174 if( isAfterBlock )
|
|
|
175 isAfterBlock = false;
|
|
|
176 else
|
|
|
177 list.add( br );
|
|
|
178 isLineStart = true;
|
|
|
179 buf.setLength(0);
|
|
|
180 break;
|
|
|
181 case '\t':
|
|
|
182 if( isLineStart ) {
|
|
|
183 buf.append( " " );
|
|
|
184 } else {
|
|
|
185 buf.append( '\t' );
|
|
|
186 }
|
|
|
187 break;
|
|
|
188 case ' ':
|
|
|
189 if( isLineStart ) {
|
|
|
190 buf.append( " " );
|
|
|
191 isLineStart = false;
|
|
|
192 } else if( buf.charAt(buf.length()-1)==' ' ) {
|
|
|
193 buf.append( " " );
|
|
|
194 } else {
|
|
|
195 buf.append( ' ' );
|
|
|
196 }
|
|
|
197 break;
|
|
|
198 default:
|
|
|
199 buf.append( c );
|
|
|
200 isLineStart = false;
|
|
|
201 isAfterBlock = false;
|
|
|
202 }
|
|
|
203 }
|
|
|
204 if( buf.length() > 0 )
|
|
|
205 list.add( buf.toString() );
|
|
|
206 return list;
|
|
|
207 }
|
|
|
208
|
|
|
209
|
|
|
210 static final Message.Format HTML = new InputFormat('h',"html") {
|
|
|
211
|
|
|
212 protected Html parse(String msg,Message.Source source) {
|
|
|
213 return MessageUtils.newHtml(msg);
|
|
|
214 }
|
|
|
215
|
|
|
216 protected Html parseForMailText(String msg,Message.Source source) {
|
|
|
217 return MessageUtils.newHtml(msg);
|
|
|
218 }
|
|
|
219
|
|
|
220 };
|
|
|
221
|
|
|
222 private static final HtmlTag _a = new HtmlTag("/a");
|
|
|
223
|
|
|
224 static String getEmail(Html list,int iEmail) {
|
|
|
225 int count = 0;
|
|
|
226 int n = list.size() - 3 + 1;
|
|
|
227 for( int i=0; i<n; i++ ) {
|
|
|
228 Object o = list.get(i);
|
|
|
229 if( !(o instanceof HtmlTag) )
|
|
|
230 continue;
|
|
|
231 HtmlTag tag = (HtmlTag)o;
|
|
|
232 if( !tag.getName().toLowerCase().equals("email") )
|
|
|
233 continue;
|
|
|
234 o = list.get(i+1);
|
|
|
235 if( !(o instanceof String) )
|
|
|
236 continue;
|
|
|
237 String email = (String)o;
|
|
|
238 o = list.get(i+2);
|
|
|
239 if( !(o instanceof HtmlTag) )
|
|
|
240 continue;
|
|
|
241 HtmlTag endTag = (HtmlTag)o;
|
|
|
242 if( !endTag.getName().toLowerCase().equals("/email") )
|
|
|
243 continue;
|
|
|
244 if( count++ == iEmail )
|
|
|
245 return email;
|
|
|
246 i += 2;
|
|
|
247 }
|
|
|
248 return null;
|
|
|
249 }
|
|
|
250
|
|
|
251
|
|
|
252 static String encodeHtml(String s) {
|
|
|
253 char[] a = s.toCharArray();
|
|
|
254 StringBuilder buf = new StringBuilder();
|
|
|
255 for( int i=0; i<a.length; i++ ) {
|
|
|
256 char c = a[i];
|
|
|
257 switch(c) {
|
|
|
258 /*
|
|
|
259 case '&':
|
|
|
260 buf.append("&");
|
|
|
261 break;
|
|
|
262 */
|
|
|
263 case '<':
|
|
|
264 buf.append("<");
|
|
|
265 break;
|
|
|
266 case '>':
|
|
|
267 buf.append(">");
|
|
|
268 break;
|
|
|
269 case '"':
|
|
|
270 buf.append(""");
|
|
|
271 break;
|
|
|
272 default:
|
|
|
273 buf.append(c);
|
|
|
274 }
|
|
|
275 }
|
|
|
276 return buf.toString();
|
|
|
277 }
|
|
|
278
|
|
|
279
|
|
|
280 static {
|
|
|
281 MailingLists.nop();
|
|
|
282 }
|
|
|
283
|
|
|
284
|
|
|
285 private static final List<String> ignoredTags = Arrays.asList(
|
|
|
286 "nabble_embed"
|
|
|
287 );
|
|
|
288
|
|
|
289 private static void nestIgnoredBlocks(Html html) {
|
|
|
290 outer:
|
|
|
291 for( String tagName : ignoredTags ) {
|
|
|
292 String endTagName = "/" + tagName;
|
|
|
293 for( ListIterator<Object> i=html.listIterator(); i.hasNext(); ) {
|
|
|
294 Object obj = i.next();
|
|
|
295 if( obj instanceof HtmlTag ) {
|
|
|
296 HtmlTag tag = (HtmlTag)obj;
|
|
|
297 if( tag.getName().equalsIgnoreCase(tagName) ) {
|
|
|
298 Html block = new Html();
|
|
|
299 i.remove();
|
|
|
300 block.add(obj);
|
|
|
301 while(true) {
|
|
|
302 if( !i.hasNext() ) { // unclosed
|
|
|
303 i.add(HtmlUtils.htmlEncode(block.toString()));
|
|
|
304 continue outer;
|
|
|
305 }
|
|
|
306 obj = i.next();
|
|
|
307 i.remove();
|
|
|
308 block.add(obj);
|
|
|
309 if( obj instanceof HtmlTag ) {
|
|
|
310 tag = (HtmlTag)obj;
|
|
|
311 if( tag.getName().equalsIgnoreCase(endTagName) ) {
|
|
|
312 i.add(block);
|
|
|
313 break;
|
|
|
314 }
|
|
|
315 }
|
|
|
316 }
|
|
|
317 }
|
|
|
318 }
|
|
|
319 }
|
|
|
320 }
|
|
|
321 }
|
|
|
322
|
|
|
323 private static void unnestIgnoredBlocks(Html html) {
|
|
|
324 for( ListIterator<Object> i=html.listIterator(); i.hasNext(); ) {
|
|
|
325 Object obj = i.next();
|
|
|
326 if( obj instanceof Html ) {
|
|
|
327 Html block = (Html)obj;
|
|
|
328 i.remove();
|
|
|
329 for( Object obj2 : block ) {
|
|
|
330 i.add(obj2);
|
|
|
331 }
|
|
|
332 }
|
|
|
333 }
|
|
|
334 }
|
|
|
335
|
|
|
336 private static final Set<String> allowedTextTags = new HashSet<String>();
|
|
|
337 static {
|
|
|
338 for( String tag : new String[]{
|
|
|
339 "b",
|
|
|
340 "i",
|
|
|
341 "quote",
|
|
|
342 "email",
|
|
|
343 "img",
|
|
|
344 "a",
|
|
|
345 "font",
|
|
|
346 "span",
|
|
|
347 "smiley",
|
|
|
348 "nabble_img",
|
|
|
349 "nabble_a",
|
|
|
350 "object",
|
|
|
351 "param",
|
|
|
352 "embed",
|
|
|
353 "raw",
|
|
|
354 "nabble_embed",
|
|
|
355 "h1",
|
|
|
356 "h2",
|
|
|
357 "h3",
|
|
|
358 "h4",
|
|
|
359 "h5",
|
|
|
360 "h6",
|
|
|
361 "table",
|
|
|
362 "thead",
|
|
|
363 "tbody",
|
|
|
364 "tr",
|
|
|
365 "th",
|
|
|
366 "td",
|
|
|
367 } ) {
|
|
|
368 allowedTextTags.add(tag);
|
|
|
369 allowedTextTags.add("/"+tag);
|
|
|
370 }
|
|
|
371 }
|
|
|
372
|
|
|
373 private static void encodeUnallowedTags(Html html) {
|
|
|
374 for( ListIterator<Object> i=html.listIterator(); i.hasNext(); ) {
|
|
|
375 Object obj = i.next();
|
|
|
376 if( obj instanceof HtmlTag ) {
|
|
|
377 HtmlTag tag = (HtmlTag)obj;
|
|
|
378 if( !allowedTextTags.contains( tag.getName().toLowerCase() ) ) {
|
|
|
379 i.remove();
|
|
|
380 i.add( HtmlUtils.htmlEncode(tag.toString()) );
|
|
|
381 }
|
|
|
382 }
|
|
|
383 }
|
|
|
384 }
|
|
|
385
|
|
|
386 private static void changeUnallowedTagsToStrings(Html html) {
|
|
|
387 for( ListIterator<Object> i=html.listIterator(); i.hasNext(); ) {
|
|
|
388 Object obj = i.next();
|
|
|
389 if( obj instanceof HtmlTag ) {
|
|
|
390 HtmlTag tag = (HtmlTag)obj;
|
|
|
391 if( !allowedTextTags.contains( tag.getName().toLowerCase() ) ) {
|
|
|
392 i.remove();
|
|
|
393 i.add( tag.toString() );
|
|
|
394 }
|
|
|
395 }
|
|
|
396 }
|
|
|
397 }
|
|
|
398 }
|
