|
1
11 package org.eclipse.ant.internal.ui.editor.derived;
12
13
14 import java.io.IOException  ;
15 import java.io.PushbackReader ;
16 import java.io.Reader ;
17 import java.util.HashMap ;
18 import java.util.HashSet ;
19 import java.util.Map ;
20 import java.util.Set ;
21
22 import org.eclipse.jface.text.TextPresentation;
23 import org.eclipse.swt.SWT;
24 import org.eclipse.swt.custom.StyleRange;
25
26
32 public class HTML2TextReader extends SubstitutionTextReader {
33
34 private static final String EMPTY_STRING= ""; private static final Map fgEntityLookup;
36 private static final Set fgTags;
37
38 static {
39
40 fgTags= new HashSet ();
41 fgTags.add("b"); fgTags.add("br"); fgTags.add("h5"); fgTags.add("p"); fgTags.add("dl"); fgTags.add("dt"); fgTags.add("dd"); fgTags.add("li"); fgTags.add("ul"); fgTags.add("pre");
52 fgEntityLookup= new HashMap (7);
53 fgEntityLookup.put("lt", "<"); fgEntityLookup.put("gt", ">"); fgEntityLookup.put("nbsp", " "); fgEntityLookup.put("amp", "&"); fgEntityLookup.put("circ", "^"); fgEntityLookup.put("tilde", "~"); fgEntityLookup.put("quot", "\""); }
61
62 private int fCounter= 0;
63 private TextPresentation fTextPresentation;
64 private int fBold= 0;
65 private int fStartOffset= -1;
66 private boolean fInParagraph= false;
67 private boolean fIsPreformattedText= false;
68
69
74 public HTML2TextReader(Reader reader, TextPresentation presentation) {
75 super(new PushbackReader (reader));
76 fTextPresentation= presentation;
77 }
78
79 public int read() throws IOException {
80 int c= super.read();
81 if (c != -1)
82 ++ fCounter;
83 return c;
84 }
85
86 protected void startBold() {
87 if (fBold == 0)
88 fStartOffset= fCounter;
89 ++ fBold;
90 }
91
92 protected void startPreformattedText() {
93 fIsPreformattedText= true;
94 setSkipWhitespace(false);
95 }
96
97 protected void stopPreformattedText() {
98 fIsPreformattedText= false;
99 setSkipWhitespace(true);
100 }
101
102 protected void stopBold() {
103 -- fBold;
104 if (fBold == 0) {
105 if (fTextPresentation != null) {
106 fTextPresentation.addStyleRange(new StyleRange(fStartOffset, fCounter - fStartOffset, null, null, SWT.BOLD));
107 }
108 fStartOffset= -1;
109 }
110 }
111
112
115 protected String computeSubstitution(int c) throws IOException {
116
117 if (c == '<')
118 return processHTMLTag();
119 else if (c == '&')
120 return processEntity();
121 else if (fIsPreformattedText)
122 return processPreformattedText(c);
123
124 return null;
125 }
126
127 private String html2Text(String html) {
128
129 String tag= html;
130 if ('/' == tag.charAt(0))
131 tag= tag.substring(1);
132
133 if (!fgTags.contains(tag))
134 return EMPTY_STRING;
135
136
137 if ("pre".equals(html)) { startPreformattedText();
139 return EMPTY_STRING;
140 }
141
142 if ("/pre".equals(html)) { stopPreformattedText();
144 return EMPTY_STRING;
145 }
146
147 if (fIsPreformattedText)
148 return EMPTY_STRING;
149
150 if ("b".equals(html)) { startBold();
152 return EMPTY_STRING;
153 }
154
155 if ("h5".equals(html) || "dt".equals(html)) { startBold();
157 return EMPTY_STRING;
158 }
159
160 if ("dl".equals(html)) return LINE_DELIM;
162
163 if ("dd".equals(html)) return "\t";
166 if ("li".equals(html)) return LINE_DELIM + "\t-";
169 if ("/b".equals(html)) { stopBold();
171 return EMPTY_STRING;
172 }
173
174 if ("p".equals(html)) { fInParagraph= true;
176 return LINE_DELIM;
177 }
178
179 if ("br".equals(html)) return LINE_DELIM;
181
182 if ("/p".equals(html)) { boolean inParagraph= fInParagraph;
184 fInParagraph= false;
185 return inParagraph ? EMPTY_STRING : LINE_DELIM;
186 }
187
188 if ("/h5".equals(html) || "/dt".equals(html)) { stopBold();
190 return LINE_DELIM;
191 }
192
193 if ("/dd".equals(html)) return LINE_DELIM;
195
196 return EMPTY_STRING;
197 }
198
199
202 private String processHTMLTag() throws IOException {
203
204 StringBuffer buf= new StringBuffer ();
205 int ch;
206 do {
207
208 ch= nextChar();
209
210 while (ch != -1 && ch != '>') {
211 buf.append(Character.toLowerCase((char) ch));
212 ch= nextChar();
213 if (ch == '"'){
214 buf.append(Character.toLowerCase((char) ch));
215 ch= nextChar();
216 while (ch != -1 && ch != '"'){
217 buf.append(Character.toLowerCase((char) ch));
218 ch= nextChar();
219 }
220 }
221 if (ch == '<'){
222 unread(ch);
223 return '<' + buf.toString();
224 }
225 }
226
227 if (ch == -1)
228 return null;
229
230 int tagLen= buf.length();
231 if ((tagLen >= 3 && "!--".equals(buf.substring(0, 3))) && !(tagLen >= 5 && "--".equals(buf.substring(tagLen - 2)))) { buf.append(ch);
236 } else {
237 break;
238 }
239 } while (true);
240
241 return html2Text(buf.toString());
242 }
243
244 private String processPreformattedText(int c) {
245 if (c == '\r' || c == '\n')
246 fCounter++;
247 return null;
248 }
249
250
251 private void unread(int ch) throws IOException {
252 ((PushbackReader ) getReader()).unread(ch);
253 }
254
255 protected String entity2Text(String symbol) {
256 if (symbol.length() > 1 && symbol.charAt(0) == '#') {
257 int ch;
258 try {
259 if (symbol.charAt(1) == 'x') {
260 ch= Integer.parseInt(symbol.substring(2), 16);
261 } else {
262 ch= Integer.parseInt(symbol.substring(1), 10);
263 }
264 return EMPTY_STRING + (char)ch;
265 } catch (NumberFormatException e) {
266 }
267 } else {
268 String str= (String ) fgEntityLookup.get(symbol);
269 if (str != null) {
270 return str;
271 }
272 }
273 return "&" + symbol; }
275
276
279 private String processEntity() throws IOException {
280 StringBuffer buf= new StringBuffer ();
281 int ch= nextChar();
282 while (Character.isLetterOrDigit((char)ch) || ch == '#') {
283 buf.append((char) ch);
284 ch= nextChar();
285 }
286
287 if (ch == ';')
288 return entity2Text(buf.toString());
289
290 buf.insert(0, '&');
291 if (ch != -1)
292 buf.append((char) ch);
293 return buf.toString();
294 }
295 }
296 |
Popular Tags |
Java API By Example, From Geeks To Geeks.