1 | /**
|
---|
2 | *#########################################################################
|
---|
3 | *
|
---|
4 | * A component of the Gatherer application, part of the Greenstone digital
|
---|
5 | * library suite from the New Zealand Digital Library Project at the
|
---|
6 | * University of Waikato, New Zealand.
|
---|
7 | *
|
---|
8 | * <BR><BR>
|
---|
9 | *
|
---|
10 | * Author: John Thompson, Greenstone Digital Library, University of Waikato
|
---|
11 | *
|
---|
12 | * <BR><BR>
|
---|
13 | *
|
---|
14 | * Copyright (C) 1999 New Zealand Digital Library Project
|
---|
15 | *
|
---|
16 | * <BR><BR>
|
---|
17 | *
|
---|
18 | * This program is free software; you can redistribute it and/or modify
|
---|
19 | * it under the terms of the GNU General Public License as published by
|
---|
20 | * the Free Software Foundation; either version 2 of the License, or
|
---|
21 | * (at your option) any later version.
|
---|
22 | *
|
---|
23 | * <BR><BR>
|
---|
24 | *
|
---|
25 | * This program is distributed in the hope that it will be useful,
|
---|
26 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
27 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
28 | * GNU General Public License for more details.
|
---|
29 | *
|
---|
30 | * <BR><BR>
|
---|
31 | *
|
---|
32 | * You should have received a copy of the GNU General Public License
|
---|
33 | * along with this program; if not, write to the Free Software
|
---|
34 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
35 | *########################################################################
|
---|
36 | */
|
---|
37 | package org.greenstone.gatherer.collection;
|
---|
38 |
|
---|
39 | import java.awt.event.*;
|
---|
40 | import java.io.*;
|
---|
41 | import java.net.*;
|
---|
42 | import java.util.*;
|
---|
43 | import javax.swing.tree.*;
|
---|
44 | import org.greenstone.gatherer.Dictionary;
|
---|
45 | import org.greenstone.gatherer.Gatherer;
|
---|
46 | import org.greenstone.gatherer.WGet;
|
---|
47 | import org.greenstone.gatherer.gui.GProgressBar;
|
---|
48 | import org.greenstone.gatherer.util.GURL;
|
---|
49 | import org.greenstone.gatherer.util.Utility;
|
---|
50 | /**
|
---|
51 | * @author John Thompson, Greenstone Digital Library, University of Waikato
|
---|
52 | * @version 2.0
|
---|
53 | */
|
---|
54 | public class Job
|
---|
55 | implements ActionListener {
|
---|
56 |
|
---|
57 | private boolean clobber;
|
---|
58 | private boolean debug;
|
---|
59 | private boolean higher_directories;
|
---|
60 | private boolean no_parents;
|
---|
61 | private boolean other_hosts;
|
---|
62 | private boolean page_requisites;
|
---|
63 | private boolean quiet;
|
---|
64 |
|
---|
65 | private GProgressBar progress;
|
---|
66 |
|
---|
67 | private GURL initial = null;
|
---|
68 | private GURL url = null;
|
---|
69 |
|
---|
70 | private TreeModel model;
|
---|
71 |
|
---|
72 | private int depth;
|
---|
73 | private int previous_state;
|
---|
74 | private int state;
|
---|
75 |
|
---|
76 | private String current_url;
|
---|
77 | private String destination;
|
---|
78 | private String proxy_pass;
|
---|
79 | private String proxy_user;
|
---|
80 |
|
---|
81 | private Vector encountered_urls;
|
---|
82 | private Vector failed_urls;
|
---|
83 |
|
---|
84 | private WGet mummy;
|
---|
85 |
|
---|
86 | public static int COMPLETE = 0;
|
---|
87 | public static int PAUSED = 1;
|
---|
88 | public static int RUNNING = 2;
|
---|
89 | public static int STOPPED = 3;
|
---|
90 |
|
---|
91 | /**
|
---|
92 | */
|
---|
93 | public Job(TreeModel model, boolean clobber, boolean debug, boolean no_parents, boolean other_hosts, boolean page_requisites, boolean quiet, URL initial, int depth, String destination, String proxy_pass, String proxy_user, WGet mummy, boolean simple) {
|
---|
94 | this.model = model;
|
---|
95 |
|
---|
96 | this.debug = debug;
|
---|
97 | this.clobber = clobber;
|
---|
98 | this.no_parents = no_parents;
|
---|
99 | this.other_hosts = other_hosts;
|
---|
100 | this.page_requisites = page_requisites;
|
---|
101 | this.quiet = quiet;
|
---|
102 | this.initial = new GURL(initial);
|
---|
103 | this.depth = depth;
|
---|
104 | this.destination = destination;
|
---|
105 | this.proxy_pass = proxy_pass;
|
---|
106 | this.proxy_user = proxy_user;
|
---|
107 | this.mummy = mummy;
|
---|
108 |
|
---|
109 | progress = new GProgressBar(this, initial.toString(), simple);
|
---|
110 |
|
---|
111 | encountered_urls = new Vector();
|
---|
112 | failed_urls = new Vector();
|
---|
113 |
|
---|
114 | previous_state = STOPPED;
|
---|
115 | state = STOPPED;
|
---|
116 | }
|
---|
117 |
|
---|
118 | /** Depending on which button on the progress bar was pushed,
|
---|
119 | * this method will affect the state of the Job and perhaps make
|
---|
120 | * calls to wget.class if necessary.
|
---|
121 | * @param event The ActionEvent fired from within the GProgressBar
|
---|
122 | * which we must respond to.
|
---|
123 | */
|
---|
124 | public void actionPerformed(ActionEvent event) {
|
---|
125 | // The action button is used to alternately start or stop the
|
---|
126 | // job. If the current state of the job is paused then this
|
---|
127 | // restart is logically equivelent to a resume.
|
---|
128 | if(event.getSource() == progress.action) {
|
---|
129 | previous_state = state;
|
---|
130 | if(state == RUNNING) {
|
---|
131 | state = PAUSED;
|
---|
132 | }
|
---|
133 | else {
|
---|
134 | state = RUNNING;
|
---|
135 | mummy.resumeThread();
|
---|
136 | }
|
---|
137 | }
|
---|
138 | else if (event.getSource() == progress.cancel) {
|
---|
139 | state = STOPPED; // Should already be stopped.
|
---|
140 | mummy.deleteJob(this);
|
---|
141 | }
|
---|
142 | }
|
---|
143 |
|
---|
144 | /** Called by the WGet native code to inform us of a new download starting.
|
---|
145 | * @param url The url that is being downloaded, as a String.
|
---|
146 | */
|
---|
147 | public void addDownload(String raw_url) {
|
---|
148 | if(!encountered_urls.contains(raw_url)) {
|
---|
149 | encountered_urls.add(raw_url);
|
---|
150 | }
|
---|
151 | // Regardless create a new GURL
|
---|
152 | current_url = raw_url;
|
---|
153 | url = new GURL(raw_url);
|
---|
154 | progress.addDownload(raw_url);
|
---|
155 | }
|
---|
156 |
|
---|
157 | /** Used to advise the Job of a newly parsed link. Its up to Job
|
---|
158 | * to decide if it already knows about this url, and if not to
|
---|
159 | * update its progress bar.
|
---|
160 | * @param url The url in question as a String.
|
---|
161 | * @param type Whether the link is an internal or external link.
|
---|
162 | * @return A boolean indicating if the url was added.
|
---|
163 | */
|
---|
164 | public boolean addLink(String raw_url, int type) {
|
---|
165 | ///ystem.out.println("addLink("+url+", "+type+")");
|
---|
166 | if(!encountered_urls.contains(raw_url)) {
|
---|
167 | // Add it to the urls we've seen.
|
---|
168 | encountered_urls.add(raw_url);
|
---|
169 | // Add it the to links for the current GURL.
|
---|
170 |
|
---|
171 | // Add it to the progress file count.
|
---|
172 | progress.increaseFileCount();
|
---|
173 | return true;
|
---|
174 | }
|
---|
175 | // Regardless add it to the children links of the current GURL
|
---|
176 | initial.addLink(raw_url);
|
---|
177 |
|
---|
178 | // We've seen it before. Don't count it again.
|
---|
179 | return false;
|
---|
180 | }
|
---|
181 |
|
---|
182 | public void callWGet() {
|
---|
183 | // Build parameter string
|
---|
184 | String command = "wget ";
|
---|
185 |
|
---|
186 | // Parse arguments into array.
|
---|
187 | // Always:
|
---|
188 | // rewrite links to be local if possible - NOOOOOO,
|
---|
189 | // output a debug file and debug messages,
|
---|
190 | // run quietly.
|
---|
191 | //command = command + "-k ";
|
---|
192 |
|
---|
193 | if(destination != null) {
|
---|
194 | command = command + "-P " + destination + " ";
|
---|
195 | }
|
---|
196 |
|
---|
197 | if(depth < 0) {
|
---|
198 | // Infinite recursion
|
---|
199 | command = command + "-r ";
|
---|
200 | }
|
---|
201 | else if (depth == 0) {
|
---|
202 | // Just this page.
|
---|
203 | }
|
---|
204 | else if (depth > 0) {
|
---|
205 | // Recursion to the specified depth.
|
---|
206 | command = command + "-r -l" + depth + " ";
|
---|
207 | }
|
---|
208 |
|
---|
209 | if(!clobber || previous_state == Job.PAUSED) {
|
---|
210 | command = command + "-nc -c ";
|
---|
211 | }
|
---|
212 |
|
---|
213 | // Determine if we have to use a proxy.
|
---|
214 | if(Gatherer.config.get("general.use_proxy", true)) {
|
---|
215 | String proxy_host = Gatherer.config.getString("general.proxy_host", true);
|
---|
216 | String proxy_port = Gatherer.config.getString("general.proxy_port", true);
|
---|
217 | // Find out whether the user has already authenticated themselves
|
---|
218 | String user_pass = null;
|
---|
219 | String address = proxy_host + ":" + proxy_port;
|
---|
220 | int count = 0;
|
---|
221 | while(count < 3 && (user_pass = (String) Gatherer.authentications.get(address)) == null) {
|
---|
222 | Authenticator.requestPasswordAuthentication(proxy_host, null, Integer.parseInt(proxy_port), "http://", Dictionary.get("WGet.Prompt"), "HTTP");
|
---|
223 | count++;
|
---|
224 | }
|
---|
225 | if(count >= 3) {
|
---|
226 | state = STOPPED;
|
---|
227 | return;
|
---|
228 | }
|
---|
229 | if(user_pass.indexOf("@") != -1) {
|
---|
230 | // Write the use proxy command
|
---|
231 | command = command + "-e httpproxy=" + proxy_host + ":" + proxy_port + "/ --proxy-user=" + user_pass.substring(0, user_pass.indexOf("@")) + " --proxy-passwd=" + user_pass.substring(user_pass.indexOf("@") + 1) + " -Y on ";
|
---|
232 |
|
---|
233 | }
|
---|
234 | else {
|
---|
235 | Gatherer.println("Unknown user/pass");
|
---|
236 | }
|
---|
237 | }
|
---|
238 |
|
---|
239 | if(page_requisites) {
|
---|
240 | command = command + "-p ";
|
---|
241 | }
|
---|
242 |
|
---|
243 | if(other_hosts) {
|
---|
244 | command = command + "-H ";
|
---|
245 | }
|
---|
246 |
|
---|
247 | // Finally tell it the site to download.
|
---|
248 | command = command + initial.toString();
|
---|
249 |
|
---|
250 | if(previous_state == Job.COMPLETE) {
|
---|
251 | progress.mirrorBegun(true, true);
|
---|
252 | }
|
---|
253 | else {
|
---|
254 | progress.mirrorBegun(false, true);
|
---|
255 | }
|
---|
256 |
|
---|
257 | // Run it
|
---|
258 | try {
|
---|
259 | Gatherer.println("Cmd: " + command);
|
---|
260 | Runtime rt = Runtime.getRuntime();
|
---|
261 | Process prcs = rt.exec(command);
|
---|
262 | InputStreamReader isr = new InputStreamReader(prcs.getErrorStream());
|
---|
263 | BufferedReader br = new BufferedReader(isr);
|
---|
264 | // Capture the standard error stream and seach for two particular occurances.
|
---|
265 | String line;
|
---|
266 | boolean ignore_for_robots = false;
|
---|
267 | while ((line = br.readLine()) != null) {
|
---|
268 | Gatherer.println(line);
|
---|
269 |
|
---|
270 | // The first magic special test is to see if we've just
|
---|
271 | // asked for the robots.txt file. If so we ignore
|
---|
272 | // the next add and then the next complete/error.
|
---|
273 | if(line.lastIndexOf("robots.txt;") != -1) {
|
---|
274 | Gatherer.println("***** Requesting robot.txt");
|
---|
275 | ignore_for_robots = true;
|
---|
276 | }
|
---|
277 | // If line contains "=> `" display text as the
|
---|
278 | // currently downloading url. Unique to add download.
|
---|
279 | else if(line.lastIndexOf("=> `") != -1) {
|
---|
280 | if(!ignore_for_robots) {
|
---|
281 | // Add download
|
---|
282 | String new_url = line.substring(line.indexOf("`") + 1, line.lastIndexOf("'"));
|
---|
283 | // Remove the destination guff
|
---|
284 | if(destination != null) {
|
---|
285 | new_url = new_url.substring(destination.length());
|
---|
286 | }
|
---|
287 | addDownload("http:/" + new_url);
|
---|
288 | }
|
---|
289 | }
|
---|
290 | // If line contains "saved [<size>]" set currently
|
---|
291 | // downloading url to "Download Complete".
|
---|
292 | else if(line.lastIndexOf(") - `") != -1) {
|
---|
293 | if(!ignore_for_robots) {
|
---|
294 | // Download complete
|
---|
295 | downloadComplete();
|
---|
296 | }
|
---|
297 | else {
|
---|
298 | ignore_for_robots = false;
|
---|
299 | }
|
---|
300 | }
|
---|
301 | // The already there line begins "File `..." However this
|
---|
302 | // is only true in english, so instead I looked and there
|
---|
303 | // are few (if any at all) other messages than those above
|
---|
304 | // and not overwriting messages that use " `" so we'll
|
---|
305 | // look for that. Note this method is not guarenteed to be
|
---|
306 | // unique like the previous two.
|
---|
307 | else if(line.lastIndexOf(" `") != -1) {
|
---|
308 | // Not Overwriting
|
---|
309 | Gatherer.println("Already there.");
|
---|
310 | String new_url =
|
---|
311 | line.substring(line.indexOf("`") + 1, line.lastIndexOf("'"));
|
---|
312 | // For some strange reason this won't compile
|
---|
313 | // src/org/greenstone/gatherer/collection/Job.java:311: cannot resolve symbol
|
---|
314 | // symbol : class CAKE
|
---|
315 | // location: class org.greenstone.gatherer.collection.Job
|
---|
316 | /* ***********************************************************
|
---|
317 | CAKE CAKE CAKE CAKE I WANT CAKE GIVE ME CAKE NOW!!
|
---|
318 | *********************************************************** */
|
---|
319 | // Remove the destination guff
|
---|
320 | if(destination != null) {
|
---|
321 | new_url = new_url.substring(destination.length());
|
---|
322 | }
|
---|
323 | addDownload("http:/" + new_url);
|
---|
324 | downloadWarning();
|
---|
325 | }
|
---|
326 | // Any other important message starts with the time in the form hh:mm:ss
|
---|
327 | else if(line.length() > 7) {
|
---|
328 | if(line.charAt(2) == ':' && line.charAt(5) == ':') {
|
---|
329 | if(!ignore_for_robots) {
|
---|
330 | Gatherer.println("Error.");
|
---|
331 | downloadFailed();
|
---|
332 | }
|
---|
333 | else {
|
---|
334 | ignore_for_robots = false;
|
---|
335 | }
|
---|
336 | }
|
---|
337 | }
|
---|
338 | }
|
---|
339 | // Now display final message based on exit value
|
---|
340 | prcs.waitFor();
|
---|
341 | }
|
---|
342 | catch (Exception ioe) {
|
---|
343 | //message(Utility.ERROR, ioe.toString());
|
---|
344 | Gatherer.printStackTrace(ioe);
|
---|
345 | }
|
---|
346 | // If we've got to here and the state isn't STOPPED then the
|
---|
347 | // job is complete.
|
---|
348 | if(state == Job.RUNNING) {
|
---|
349 | progress.mirrorComplete();
|
---|
350 | previous_state = state;
|
---|
351 | state = Job.COMPLETE;
|
---|
352 | }
|
---|
353 | }
|
---|
354 |
|
---|
355 | /** The most important part of the Job class, this method is
|
---|
356 | * responsible for calling the WGet native methods used to
|
---|
357 | * mirror the indicated url. By this stage all the variables
|
---|
358 | * necessary should be set and we need only build up the
|
---|
359 | * parameter string and make the call.
|
---|
360 | */
|
---|
361 | public void callWGetNative() {
|
---|
362 | Vector args = new Vector();
|
---|
363 |
|
---|
364 | // Let the GProgressBar know we're starting, just in case
|
---|
365 | // the user hasn't told us to. If this is the second time the
|
---|
366 | // urls downloaded and the first attempt was successful (ie
|
---|
367 | // the previous job was complete), then we have the case where
|
---|
368 | // the user is forcing us to remirror. Reset all the values etc
|
---|
369 | // if this is the case then reset the variables.
|
---|
370 | // Note that this can cause the result line to look something
|
---|
371 | // like this.
|
---|
372 | // Downloaded 12 of 12 files (8 warnings, 0 errors).
|
---|
373 | // The warnings would be something like, 'File already downloaded'
|
---|
374 | // but the total number of files and the file successfully
|
---|
375 | // downloaded will be correct.
|
---|
376 | if(previous_state == Job.COMPLETE) {
|
---|
377 | progress.mirrorBegun(true, false);
|
---|
378 | }
|
---|
379 | else {
|
---|
380 | progress.mirrorBegun(false, false);
|
---|
381 | }
|
---|
382 |
|
---|
383 | // Parse arguments into array.
|
---|
384 | args.add(Utility.BASE_DIR + "wget");
|
---|
385 | //args.add("-k");
|
---|
386 | args.add("-d");
|
---|
387 | args.add("-o");
|
---|
388 | args.add("debug.txt");
|
---|
389 |
|
---|
390 | if(destination != null) {
|
---|
391 | args.add("-P");
|
---|
392 | args.add(destination);
|
---|
393 | }
|
---|
394 |
|
---|
395 | if(depth < 0) {
|
---|
396 | // Infinite recursion
|
---|
397 | args.add("-r");
|
---|
398 | }
|
---|
399 | else if (depth == 0) {
|
---|
400 | // Just this page.
|
---|
401 | }
|
---|
402 | else if (depth > 0) {
|
---|
403 | // Recursion to the specified depth.
|
---|
404 | args.add("-r");
|
---|
405 | args.add("-l");
|
---|
406 | args.add("" + depth + ""); // Hacky
|
---|
407 | }
|
---|
408 |
|
---|
409 | if(!clobber || previous_state == PAUSED) {
|
---|
410 | args.add("-nc");
|
---|
411 | args.add("-c");
|
---|
412 | }
|
---|
413 |
|
---|
414 | if(proxy_user != null) {
|
---|
415 | args.add("--proxy-user=" + proxy_user);
|
---|
416 | args.add("--proxy-passwd=" + proxy_pass);
|
---|
417 | }
|
---|
418 |
|
---|
419 | if(page_requisites) {
|
---|
420 | args.add("-p");
|
---|
421 | }
|
---|
422 |
|
---|
423 | if(quiet) {
|
---|
424 | args.add("-q");
|
---|
425 | }
|
---|
426 |
|
---|
427 | if(other_hosts) {
|
---|
428 | args.add("-H");
|
---|
429 | }
|
---|
430 |
|
---|
431 | args.add(initial.toString());
|
---|
432 |
|
---|
433 | Gatherer.println("Calling wget ");
|
---|
434 | for(Enumeration e = args.elements(); e.hasMoreElements();) {
|
---|
435 | Gatherer.println(e.nextElement() + " ");
|
---|
436 | }
|
---|
437 | Gatherer.println("");
|
---|
438 |
|
---|
439 | // Run home to mummy.
|
---|
440 | int value = mummy.wget(args.size(), args.toArray(), debug);
|
---|
441 |
|
---|
442 | // If we've got to here and the state isn't STOPPED then the job is complete.
|
---|
443 | if(state == RUNNING) {
|
---|
444 | progress.mirrorComplete();
|
---|
445 | previous_state = state;
|
---|
446 | state = COMPLETE;
|
---|
447 | }
|
---|
448 | }
|
---|
449 |
|
---|
450 | /** Called by the WGet native code when the current download is
|
---|
451 | * completed. In turn all download listeners are informed.
|
---|
452 | */
|
---|
453 | public void downloadComplete() {
|
---|
454 | progress.downloadComplete();
|
---|
455 | /* @todo
|
---|
456 | model.add(url.getURL(), destination);
|
---|
457 | */
|
---|
458 | url = null;
|
---|
459 | current_url = null;
|
---|
460 | }
|
---|
461 |
|
---|
462 | /** Called by the WGet native code when the requested download returns
|
---|
463 | * a status code other than 200.
|
---|
464 | */
|
---|
465 | public void downloadFailed() {
|
---|
466 | ///ystem.out.println("downloadFailed("+current_url+")");
|
---|
467 | failed_urls.add(current_url); // Its the current url thats failed.
|
---|
468 | progress.downloadFailed();
|
---|
469 | }
|
---|
470 |
|
---|
471 | /**
|
---|
472 | */
|
---|
473 | public void downloadWarning() {
|
---|
474 | progress.downloadWarning();
|
---|
475 | }
|
---|
476 |
|
---|
477 | /**
|
---|
478 | * @return A String representing the currently downloading url.
|
---|
479 | */
|
---|
480 | public String getCurrent() {
|
---|
481 | return current_url;
|
---|
482 | }
|
---|
483 |
|
---|
484 | /**
|
---|
485 | * @return A String representing the initial urls host (root node
|
---|
486 | * of tree that we are mirroring).
|
---|
487 | */
|
---|
488 | public String getHost() {
|
---|
489 | return url.getHost();
|
---|
490 | }
|
---|
491 |
|
---|
492 | /**
|
---|
493 | * @return Returns the progress bar associated with this job.
|
---|
494 | */
|
---|
495 | public GProgressBar getProgressBar() {
|
---|
496 | return progress;
|
---|
497 | }
|
---|
498 |
|
---|
499 | /** Called to discover if the user wanted this thread to run or if
|
---|
500 | * it is paused.
|
---|
501 | * @return An int representing the current Job state.
|
---|
502 | */
|
---|
503 | public int getState() {
|
---|
504 | return state;
|
---|
505 | }
|
---|
506 |
|
---|
507 | /** Returns the current state of the stop flag for this job.
|
---|
508 | * @return A boolean representing whether the user has requested to
|
---|
509 | * stop.
|
---|
510 | */
|
---|
511 | public boolean hasSignalledStop() {
|
---|
512 | if(state == Job.STOPPED || state == Job.PAUSED ||
|
---|
513 | state == Job.COMPLETE) {
|
---|
514 | return true;
|
---|
515 | }
|
---|
516 | return false;
|
---|
517 | }
|
---|
518 |
|
---|
519 | /** A convinence call.
|
---|
520 | * @return A String representing the url of the initial url (root node of the mirrored tree).
|
---|
521 | */
|
---|
522 | public String toString() {
|
---|
523 | return initial.toString();
|
---|
524 | }
|
---|
525 |
|
---|
526 | /** Called by the WGet native code to signal the current progress of
|
---|
527 | * downloading.
|
---|
528 | * @param current A long representing the number of bytes that have
|
---|
529 | * been downloaded since the last update.
|
---|
530 | * @param expected A long representing the total number of bytes
|
---|
531 | * expected for this download.
|
---|
532 | */
|
---|
533 | public void updateProgress(long current, long expected) {
|
---|
534 | progress.updateProgress(current, expected);
|
---|
535 | }
|
---|
536 | }
|
---|