1 | package org.greenstone.gsdl3.gs3build.util;
|
---|
2 |
|
---|
3 |
|
---|
4 |
|
---|
5 | import java.net.*;
|
---|
6 |
|
---|
7 | import java.io.*;
|
---|
8 |
|
---|
9 |
|
---|
10 |
|
---|
11 | public class HTMLLoader implements Runnable
|
---|
12 |
|
---|
13 | { URL from;
|
---|
14 |
|
---|
15 | StringBuffer HTML;
|
---|
16 |
|
---|
17 | boolean loaded;
|
---|
18 |
|
---|
19 | boolean error;
|
---|
20 |
|
---|
21 | boolean notFound;
|
---|
22 |
|
---|
23 | Thread thread;
|
---|
24 |
|
---|
25 | static int running;
|
---|
26 |
|
---|
27 |
|
---|
28 |
|
---|
29 | /**
|
---|
30 |
|
---|
31 | * Constructor to load given URL
|
---|
32 |
|
---|
33 | */
|
---|
34 |
|
---|
35 | public HTMLLoader(URL from)
|
---|
36 |
|
---|
37 | { this.from = from;
|
---|
38 |
|
---|
39 | this.HTML = new StringBuffer();
|
---|
40 |
|
---|
41 | this.loaded = false;
|
---|
42 |
|
---|
43 | this.error = false;
|
---|
44 |
|
---|
45 | this.thread = null;
|
---|
46 | }
|
---|
47 |
|
---|
48 |
|
---|
49 |
|
---|
50 | /**
|
---|
51 |
|
---|
52 | * @return <code>0</code> if finished running; any other value if
|
---|
53 |
|
---|
54 | * still executing
|
---|
55 |
|
---|
56 | */
|
---|
57 |
|
---|
58 | public static synchronized int running()
|
---|
59 |
|
---|
60 | { return running;
|
---|
61 |
|
---|
62 | }
|
---|
63 |
|
---|
64 |
|
---|
65 |
|
---|
66 | /**
|
---|
67 |
|
---|
68 | * Starts the thread; the method <code>load</code> is a synonym for this
|
---|
69 |
|
---|
70 | */
|
---|
71 |
|
---|
72 | public synchronized void start()
|
---|
73 |
|
---|
74 | { if (thread == null)
|
---|
75 |
|
---|
76 | { thread = new Thread(this);
|
---|
77 |
|
---|
78 | thread.start();
|
---|
79 |
|
---|
80 | running ++;
|
---|
81 |
|
---|
82 | }
|
---|
83 |
|
---|
84 | // System.out.println("Running: "+running);
|
---|
85 |
|
---|
86 | }
|
---|
87 |
|
---|
88 |
|
---|
89 |
|
---|
90 | /**
|
---|
91 |
|
---|
92 | * Do the actual loading; this will multi-thread with calling task
|
---|
93 |
|
---|
94 | */
|
---|
95 |
|
---|
96 | public void run()
|
---|
97 |
|
---|
98 | { URLConnection connect;
|
---|
99 |
|
---|
100 | int length, i, c;
|
---|
101 |
|
---|
102 | byte inbuffer[];
|
---|
103 |
|
---|
104 | InputStream in;
|
---|
105 |
|
---|
106 |
|
---|
107 |
|
---|
108 | inbuffer = new byte[128];
|
---|
109 |
|
---|
110 |
|
---|
111 |
|
---|
112 | this.notFound = false;
|
---|
113 | // System.out.println("Running loader");
|
---|
114 |
|
---|
115 |
|
---|
116 |
|
---|
117 | // get connection and test length
|
---|
118 |
|
---|
119 | try
|
---|
120 |
|
---|
121 | { connect = this.from.openConnection();
|
---|
122 |
|
---|
123 | }
|
---|
124 |
|
---|
125 | catch (IOException ioex)
|
---|
126 |
|
---|
127 | { this.loaded = true;
|
---|
128 |
|
---|
129 | this.error = true;
|
---|
130 |
|
---|
131 | running --;
|
---|
132 | System.out.println(ioex);
|
---|
133 |
|
---|
134 | return;
|
---|
135 |
|
---|
136 | }
|
---|
137 |
|
---|
138 |
|
---|
139 | length = connect.getContentLength();
|
---|
140 |
|
---|
141 | if (length == 0)
|
---|
142 |
|
---|
143 | { this.loaded = true;
|
---|
144 |
|
---|
145 | running --;
|
---|
146 | System.out.println("Empty");
|
---|
147 |
|
---|
148 | return;
|
---|
149 |
|
---|
150 | }
|
---|
151 |
|
---|
152 |
|
---|
153 |
|
---|
154 | // get an inputstream to the document body
|
---|
155 |
|
---|
156 | try
|
---|
157 |
|
---|
158 | { in = connect.getInputStream();
|
---|
159 |
|
---|
160 | }
|
---|
161 |
|
---|
162 | catch (IOException ioex)
|
---|
163 |
|
---|
164 | { this.error = true;
|
---|
165 |
|
---|
166 | this.loaded = true;
|
---|
167 |
|
---|
168 | running --;
|
---|
169 | System.out.println("Open failed" + ioex);
|
---|
170 |
|
---|
171 | return;
|
---|
172 |
|
---|
173 | }
|
---|
174 |
|
---|
175 |
|
---|
176 |
|
---|
177 | if (length < 0)
|
---|
178 |
|
---|
179 | { if (connect.getHeaderField(0) != null &&
|
---|
180 |
|
---|
181 | connect.getHeaderField(0).indexOf("404") >= 0)
|
---|
182 |
|
---|
183 | { this.notFound = true;
|
---|
184 |
|
---|
185 | this.error = true;
|
---|
186 |
|
---|
187 | this.loaded = true;
|
---|
188 |
|
---|
189 | running --;
|
---|
190 | System.out.println("Not found");
|
---|
191 |
|
---|
192 | return;
|
---|
193 |
|
---|
194 | }
|
---|
195 |
|
---|
196 | }
|
---|
197 |
|
---|
198 | // System.out.println("Ready to read " + length);
|
---|
199 |
|
---|
200 |
|
---|
201 |
|
---|
202 | // read in the document body
|
---|
203 |
|
---|
204 | i = 0;
|
---|
205 |
|
---|
206 | c = 0;
|
---|
207 |
|
---|
208 | while (true || i < length)
|
---|
209 |
|
---|
210 | { // read in 128 characters (abort on IO Exception)
|
---|
211 |
|
---|
212 | try
|
---|
213 |
|
---|
214 | { c = in.read(inbuffer);
|
---|
215 |
|
---|
216 | }
|
---|
217 |
|
---|
218 | catch (IOException ioex)
|
---|
219 |
|
---|
220 | { break;
|
---|
221 |
|
---|
222 | }
|
---|
223 |
|
---|
224 | // System.out.println("read " + c);
|
---|
225 |
|
---|
226 |
|
---|
227 |
|
---|
228 | // if no characters were read, end, otherwise add it to the body
|
---|
229 |
|
---|
230 | if (c != -1)
|
---|
231 |
|
---|
232 | { if (c > 0)
|
---|
233 |
|
---|
234 | { this.HTML.append(new String(inbuffer, 0, c));
|
---|
235 |
|
---|
236 | i += c;
|
---|
237 |
|
---|
238 | }
|
---|
239 |
|
---|
240 | }
|
---|
241 |
|
---|
242 | else
|
---|
243 |
|
---|
244 | { break;
|
---|
245 |
|
---|
246 | }
|
---|
247 |
|
---|
248 | // if an incomplete buffer was read (ie no characters remain), end
|
---|
249 |
|
---|
250 | /* if (c < inbuffer.length && i >= length)
|
---|
251 |
|
---|
252 | { break;
|
---|
253 |
|
---|
254 | }*/
|
---|
255 |
|
---|
256 | }
|
---|
257 |
|
---|
258 |
|
---|
259 |
|
---|
260 | // Close file
|
---|
261 |
|
---|
262 | try
|
---|
263 |
|
---|
264 | { in.close();
|
---|
265 |
|
---|
266 | }
|
---|
267 |
|
---|
268 | catch (IOException ioex)
|
---|
269 |
|
---|
270 | { //System.out.println("Closed except "+ioex);
|
---|
271 |
|
---|
272 | }
|
---|
273 |
|
---|
274 |
|
---|
275 | this.loaded = true;
|
---|
276 |
|
---|
277 | running --;
|
---|
278 |
|
---|
279 | }
|
---|
280 |
|
---|
281 |
|
---|
282 |
|
---|
283 | /**
|
---|
284 |
|
---|
285 | * Do the actual loading (calls a Start on the Thread)
|
---|
286 |
|
---|
287 | */
|
---|
288 |
|
---|
289 | public void load()
|
---|
290 |
|
---|
291 | { this.start();
|
---|
292 |
|
---|
293 | // this.run();
|
---|
294 |
|
---|
295 | }
|
---|
296 |
|
---|
297 |
|
---|
298 |
|
---|
299 | /**
|
---|
300 |
|
---|
301 | * @return HTML text as a <code>String</code>.
|
---|
302 |
|
---|
303 | */
|
---|
304 |
|
---|
305 | public synchronized String HTML()
|
---|
306 |
|
---|
307 | { if (this.error)
|
---|
308 |
|
---|
309 | { return "Error";
|
---|
310 |
|
---|
311 | }
|
---|
312 |
|
---|
313 | return this.HTML.toString();
|
---|
314 |
|
---|
315 | }
|
---|
316 |
|
---|
317 |
|
---|
318 |
|
---|
319 | /**
|
---|
320 |
|
---|
321 | * @return <code>true</code> if the file has successfully loaded
|
---|
322 |
|
---|
323 | */
|
---|
324 |
|
---|
325 | public boolean loaded()
|
---|
326 |
|
---|
327 | { return this.loaded;
|
---|
328 |
|
---|
329 | }
|
---|
330 |
|
---|
331 |
|
---|
332 |
|
---|
333 | protected void finalize() throws Throwable
|
---|
334 |
|
---|
335 | { this.thread = null;
|
---|
336 |
|
---|
337 | this.from = null;
|
---|
338 |
|
---|
339 | this.HTML = null;
|
---|
340 |
|
---|
341 | super.finalize();
|
---|
342 |
|
---|
343 | }
|
---|
344 |
|
---|
345 |
|
---|
346 |
|
---|
347 | static
|
---|
348 |
|
---|
349 | { running = 0;
|
---|
350 |
|
---|
351 | }
|
---|
352 |
|
---|
353 | } |
---|