|
1
11 package org.eclipse.jdt.internal.ui.text;
12
13 import java.io.IOException  ;
14 import java.io.PushbackReader ;
15 import java.io.Reader ;
16 import java.util.HashMap ;
17 import java.util.HashSet ;
18 import java.util.Map ;
19 import java.util.Set ;
20
21 import org.eclipse.swt.SWT;
22 import org.eclipse.swt.custom.StyleRange;
23
24 import org.eclipse.jface.text.TextPresentation;
25
26 import org.eclipse.jdt.internal.ui.JavaUIMessages;
27
28
29
33 public class HTML2TextReader extends SubstitutionTextReader {
34
35 private static final String EMPTY_STRING= ""; private static final Map fgEntityLookup;
37 private static final Set fgTags;
38
39 static {
40
41 fgTags= new HashSet ();
42 fgTags.add("b"); fgTags.add("br"); fgTags.add("br/"); fgTags.add("div"); fgTags.add("h1"); fgTags.add("h2"); fgTags.add("h3"); fgTags.add("h4"); fgTags.add("h5"); fgTags.add("p"); fgTags.add("dl"); fgTags.add("dt"); fgTags.add("dd"); fgTags.add("li"); fgTags.add("ul"); fgTags.add("pre"); fgTags.add("head");
60 fgEntityLookup= new HashMap (7);
61 fgEntityLookup.put("lt", "<"); fgEntityLookup.put("gt", ">"); fgEntityLookup.put("nbsp", " "); fgEntityLookup.put("amp", "&"); fgEntityLookup.put("circ", "^"); fgEntityLookup.put("tilde", "~"); fgEntityLookup.put("quot", "\""); }
69
70 private int fCounter= 0;
71 private TextPresentation fTextPresentation;
72 private int fBold= 0;
73 private int fStartOffset= -1;
74 private boolean fInParagraph= false;
75 private boolean fIsPreformattedText= false;
76 private boolean fIgnore= false;
77
78
85 public HTML2TextReader(Reader reader, TextPresentation presentation) {
86 super(new PushbackReader (reader));
87 fTextPresentation= presentation;
88 }
89
90 public int read() throws IOException {
91 int c= super.read();
92 if (c != -1)
93 ++ fCounter;
94 return c;
95 }
96
97 protected void startBold() {
98 if (fBold == 0)
99 fStartOffset= fCounter;
100 ++ fBold;
101 }
102
103 protected void startPreformattedText() {
104 fIsPreformattedText= true;
105 setSkipWhitespace(false);
106 }
107
108 protected void stopPreformattedText() {
109 fIsPreformattedText= false;
110 setSkipWhitespace(true);
111 }
112
113 protected void stopBold() {
114 -- fBold;
115 if (fBold == 0) {
116 if (fTextPresentation != null) {
117 fTextPresentation.addStyleRange(new StyleRange(fStartOffset, fCounter - fStartOffset, null, null, SWT.BOLD));
118 }
119 fStartOffset= -1;
120 }
121 }
122
123
126 protected String computeSubstitution(int c) throws IOException {
127
128 if (c == '<')
129 return processHTMLTag();
130 else if (fIgnore)
131 return EMPTY_STRING;
132 else if (c == '&')
133 return processEntity();
134 else if (fIsPreformattedText)
135 return processPreformattedText(c);
136
137 return null;
138 }
139
140 private String html2Text(String html) {
141
142 if (html == null || html.length() == 0)
143 return EMPTY_STRING;
144
145 html= html.toLowerCase();
146
147 String tag= html;
148 if ('/' == tag.charAt(0))
149 tag= tag.substring(1);
150
151 if (!fgTags.contains(tag))
152 return EMPTY_STRING;
153
154
155 if ("pre".equals(html)) { startPreformattedText();
157 return EMPTY_STRING;
158 }
159
160 if ("/pre".equals(html)) { stopPreformattedText();
162 return EMPTY_STRING;
163 }
164
165 if (fIsPreformattedText)
166 return EMPTY_STRING;
167
168 if ("b".equals(html)) { startBold();
170 return EMPTY_STRING;
171 }
172
173 if ((html.length() > 1 && html.charAt(0) == 'h' && Character.isDigit(html.charAt(1))) || "dt".equals(html)) { startBold();
175 return EMPTY_STRING;
176 }
177
178 if ("dl".equals(html)) return LINE_DELIM;
180
181 if ("dd".equals(html)) return "\t";
184 if ("li".equals(html)) return LINE_DELIM + JavaUIMessages.HTML2TextReader_listItemPrefix;
187
188 if ("/b".equals(html)) { stopBold();
190 return EMPTY_STRING;
191 }
192
193 if ("p".equals(html)) { fInParagraph= true;
195 return LINE_DELIM;
196 }
197
198 if ("br".equals(html) || "br/".equals(html) || "div".equals(html)) return LINE_DELIM;
200
201 if ("/p".equals(html)) { boolean inParagraph= fInParagraph;
203 fInParagraph= false;
204 return inParagraph ? EMPTY_STRING : LINE_DELIM;
205 }
206
207 if ((html.startsWith("/h") && html.length() > 2 && Character.isDigit(html.charAt(2))) || "/dt".equals(html)) { stopBold();
209 return LINE_DELIM;
210 }
211
212 if ("/dd".equals(html)) return LINE_DELIM;
214
215 if ("head".equals(html)) { fIgnore= true;
217 return EMPTY_STRING;
218 }
219
220 if ("/head".equals(html)) { fIgnore= false;
222 return EMPTY_STRING;
223 }
224
225 return EMPTY_STRING;
226 }
227
228
231 private String processHTMLTag() throws IOException {
232
233 StringBuffer buf= new StringBuffer ();
234 int ch;
235 do {
236
237 ch= nextChar();
238
239 while (ch != -1 && ch != '>') {
240 buf.append(Character.toLowerCase((char) ch));
241 ch= nextChar();
242 if (ch == '"'){
243 buf.append(Character.toLowerCase((char) ch));
244 ch= nextChar();
245 while (ch != -1 && ch != '"'){
246 buf.append(Character.toLowerCase((char) ch));
247 ch= nextChar();
248 }
249 }
250 if (ch == '<'){
251 unread(ch);
252 return '<' + buf.toString();
253 }
254 }
255
256 if (ch == -1)
257 return null;
258
259 int tagLen= buf.length();
260 if ((tagLen >= 3 && "!--".equals(buf.substring(0, 3))) && !(tagLen >= 5 && "--".equals(buf.substring(tagLen - 2)))) { buf.append(ch);
265 } else {
266 break;
267 }
268 } while (true);
269
270 return html2Text(buf.toString());
271 }
272
273 private String processPreformattedText(int c) {
274 if (c == '\r' || c == '\n')
275 fCounter++;
276 return null;
277 }
278
279
280 private void unread(int ch) throws IOException {
281 ((PushbackReader ) getReader()).unread(ch);
282 }
283
284 protected String entity2Text(String symbol) {
285 if (symbol.length() > 1 && symbol.charAt(0) == '#') {
286 int ch;
287 try {
288 if (symbol.charAt(1) == 'x') {
289 ch= Integer.parseInt(symbol.substring(2), 16);
290 } else {
291 ch= Integer.parseInt(symbol.substring(1), 10);
292 }
293 return EMPTY_STRING + (char)ch;
294 } catch (NumberFormatException e) {
295 }
296 } else {
297 String str= (String ) fgEntityLookup.get(symbol);
298 if (str != null) {
299 return str;
300 }
301 }
302 return "&" + symbol; }
304
305
308 private String processEntity() throws IOException {
309 StringBuffer buf= new StringBuffer ();
310 int ch= nextChar();
311 while (Character.isLetterOrDigit((char)ch) || ch == '#') {
312 buf.append((char) ch);
313 ch= nextChar();
314 }
315
316 if (ch == ';')
317 return entity2Text(buf.toString());
318
319 buf.insert(0, '&');
320 if (ch != -1)
321 buf.append((char) ch);
322 return buf.toString();
323 }
324 }
325 |
Popular Tags |
Java API By Example, From Geeks To Geeks.