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.download;
|
---|
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.Configuration;
|
---|
45 | import org.greenstone.gatherer.DebugStream;
|
---|
46 | import org.greenstone.gatherer.Dictionary;
|
---|
47 | import org.greenstone.gatherer.Gatherer;
|
---|
48 | import org.greenstone.gatherer.GAuthenticator;
|
---|
49 | import org.greenstone.gatherer.greenstone.LocalGreenstone;
|
---|
50 | import org.greenstone.gatherer.file.WorkspaceTree;
|
---|
51 | import org.greenstone.gatherer.util.AppendLineOnlyFileDocument;
|
---|
52 | import org.greenstone.gatherer.util.Utility;
|
---|
53 | import org.greenstone.gatherer.cdm.Argument;
|
---|
54 | import org.greenstone.gatherer.collection.*;
|
---|
55 | /**
|
---|
56 | * @author John Thompson, Greenstone Digital Library, University of Waikato
|
---|
57 | * @version 2.0
|
---|
58 | */
|
---|
59 | public class DownloadJob
|
---|
60 | implements ActionListener {
|
---|
61 |
|
---|
62 | private boolean debug;
|
---|
63 | private boolean higher_directories;
|
---|
64 | private boolean no_parents;
|
---|
65 | private boolean other_hosts;
|
---|
66 | private boolean page_requisites;
|
---|
67 | private boolean quiet;
|
---|
68 |
|
---|
69 | private AppendLineOnlyFileDocument download_log;
|
---|
70 |
|
---|
71 | private DownloadProgressBar progress;
|
---|
72 |
|
---|
73 | private int depth;
|
---|
74 | private int previous_state;
|
---|
75 | private int state;
|
---|
76 |
|
---|
77 | private String download_url = "";
|
---|
78 |
|
---|
79 | // private String current_url;
|
---|
80 | // private String destination;
|
---|
81 | private String proxy_pass;
|
---|
82 | private String proxy_user;
|
---|
83 |
|
---|
84 | private Vector encountered_urls;
|
---|
85 | private Vector failed_urls;
|
---|
86 | private Download download;
|
---|
87 | private DownloadScrollPane mummy;
|
---|
88 | private HashMap download_option;
|
---|
89 |
|
---|
90 | public static int COMPLETE = 0;
|
---|
91 | public static int PAUSED = 1;
|
---|
92 | public static int RUNNING = 2;
|
---|
93 | public static int STOPPED = 3;
|
---|
94 |
|
---|
95 | public static int UNKNOWN_MAX = 0;
|
---|
96 | public static int DEFINED_MAX = 1;
|
---|
97 | public static int UNDEFINED_MAX = 2;
|
---|
98 |
|
---|
99 | private String mode = null;
|
---|
100 |
|
---|
101 | private String proxy_url;
|
---|
102 |
|
---|
103 | /**
|
---|
104 | */
|
---|
105 | public DownloadJob(Download download, String proxy_pass, String proxy_user, DownloadScrollPane mummy, String mode, String proxy_url) {
|
---|
106 | URL url = null;
|
---|
107 | int folder_hash;
|
---|
108 |
|
---|
109 | this.proxy_url = proxy_url;
|
---|
110 |
|
---|
111 | download_option = downloadToHashMap(download);
|
---|
112 | if (!mode.equals("Z3950") && !mode.equals("SRW")) {
|
---|
113 | Argument url_arg = (Argument)download_option.get((String)"url");
|
---|
114 | download_url = url_arg.getValue();
|
---|
115 |
|
---|
116 | }
|
---|
117 | else {
|
---|
118 | Argument host_arg = (Argument)download_option.get((String)"host");
|
---|
119 | Argument port_arg = (Argument)download_option.get((String)"port");
|
---|
120 | download_url = host_arg.getValue() + ":" +port_arg.getValue();
|
---|
121 | }
|
---|
122 |
|
---|
123 | folder_hash = download_url.hashCode();
|
---|
124 | String log_filename = Utility.getLogDir(null) + "download-"+ mode + folder_hash + ".log";
|
---|
125 | File log_file = new File(log_filename);
|
---|
126 | if(log_file.exists()) {
|
---|
127 | log_file.delete();
|
---|
128 | }
|
---|
129 |
|
---|
130 | File parent_log_file = log_file.getParentFile();
|
---|
131 | parent_log_file.mkdirs();
|
---|
132 | parent_log_file = null;
|
---|
133 | log_file = null;
|
---|
134 |
|
---|
135 | this.download_log = new AppendLineOnlyFileDocument(log_filename, false);
|
---|
136 |
|
---|
137 | this.proxy_pass = proxy_pass;
|
---|
138 | this.proxy_user = proxy_user;
|
---|
139 | this.mummy = mummy;
|
---|
140 | this.mode = mode;
|
---|
141 | this.download = download;
|
---|
142 |
|
---|
143 | progress = new DownloadProgressBar(this,download_url, true);
|
---|
144 | encountered_urls = new Vector();
|
---|
145 | failed_urls = new Vector();
|
---|
146 |
|
---|
147 | previous_state = STOPPED;
|
---|
148 | state = STOPPED;
|
---|
149 | }
|
---|
150 |
|
---|
151 | private HashMap downloadToHashMap(Download download)
|
---|
152 | {
|
---|
153 | HashMap download_option = new HashMap();
|
---|
154 | ArrayList arguments = download.getArguments(true, false);
|
---|
155 | for(int i = 0; i < arguments.size(); i++) {
|
---|
156 | Argument argument = (Argument) arguments.get(i);
|
---|
157 | download_option.put(argument.getName(), argument);
|
---|
158 | }
|
---|
159 | return download_option;
|
---|
160 | }
|
---|
161 |
|
---|
162 | /** Depending on which button on the progress bar was pushed,
|
---|
163 | * this method will affect the state of the DownloadJob and perhaps make
|
---|
164 | * calls to wget.class if necessary.
|
---|
165 | * @param event The ActionEvent fired from within the DownloadProgressBar
|
---|
166 | * which we must respond to.
|
---|
167 | */
|
---|
168 | public void actionPerformed(ActionEvent event) {
|
---|
169 | // The stop_start_button is used to alternately start or stop the
|
---|
170 | // job. If the current state of the job is paused then this
|
---|
171 | // restart is logically equivelent to a resume.
|
---|
172 | if(event.getSource() == progress.stop_start_button) {
|
---|
173 | previous_state = state;
|
---|
174 | if (state == RUNNING) {
|
---|
175 | state = STOPPED;
|
---|
176 | } else {
|
---|
177 | //previous_state = state;
|
---|
178 | state = RUNNING;
|
---|
179 | mummy.resumeThread();
|
---|
180 | }
|
---|
181 | }
|
---|
182 | else if (event.getSource() == progress.close_button) {
|
---|
183 | if(state == RUNNING) {
|
---|
184 | previous_state = state;
|
---|
185 | state = STOPPED; // do we need to do anything else to stop this?
|
---|
186 | }
|
---|
187 | mummy.deleteDownloadJob(this);
|
---|
188 | }
|
---|
189 | }
|
---|
190 |
|
---|
191 |
|
---|
192 | public void callDownload() {
|
---|
193 |
|
---|
194 | ArrayList command_list = new ArrayList();
|
---|
195 | if (Utility.isWindows()) {
|
---|
196 | command_list.add(Configuration.perl_path);
|
---|
197 | command_list.add("-S");
|
---|
198 | }
|
---|
199 | command_list.add(LocalGreenstone.getBinScriptDirectoryPath()+"downloadfrom.pl");
|
---|
200 | command_list.add("-download_mode");
|
---|
201 | command_list.add(mode);
|
---|
202 | command_list.add("-cache_dir");
|
---|
203 | command_list.add(Gatherer.getGLIUserCacheDirectoryPath());
|
---|
204 |
|
---|
205 | ArrayList all_arg = download.getArguments(true,false);
|
---|
206 | for(int i = 0; i < all_arg.size(); i++) {
|
---|
207 | Argument argument = (Argument) all_arg.get(i);
|
---|
208 | if(argument.isAssigned()) {
|
---|
209 | command_list.add("-" + argument.getName());
|
---|
210 | if(argument.getType() != Argument.FLAG) {
|
---|
211 | command_list.add(argument.getValue());
|
---|
212 | }
|
---|
213 | }
|
---|
214 | }
|
---|
215 |
|
---|
216 | String [] cmd = (String []) command_list.toArray(new String[0]);
|
---|
217 | DebugStream.println("Download job, "+command_list);
|
---|
218 |
|
---|
219 | if (previous_state == DownloadJob.COMPLETE) {
|
---|
220 | progress.mirrorBegun(true, true);
|
---|
221 | }
|
---|
222 | else {
|
---|
223 | progress.mirrorBegun(false, true);
|
---|
224 | }
|
---|
225 |
|
---|
226 | try {
|
---|
227 | Runtime rt = Runtime.getRuntime();
|
---|
228 |
|
---|
229 | String [] env = null;
|
---|
230 |
|
---|
231 | Process prcs = null;
|
---|
232 |
|
---|
233 |
|
---|
234 | if (Utility.isWindows()) {
|
---|
235 | prcs = rt.exec(cmd);
|
---|
236 | }
|
---|
237 | else {
|
---|
238 | if (proxy_url != null && !proxy_url.equals("")) {
|
---|
239 | // Specify proxies as environment variables
|
---|
240 | // Need to manually specify GSDLHOME and GSDLOS also
|
---|
241 | env = new String[4];
|
---|
242 | proxy_url = proxy_url.replaceAll("http://","");
|
---|
243 | env[0] = "http_proxy=http://"+proxy_url;
|
---|
244 | env[1] = "ftp_proxy=ftp://"+proxy_url;
|
---|
245 | env[2] = "GSDLHOME=" + Configuration.gsdl_path;
|
---|
246 | env[3] = "GSDLOS=" + Gatherer.client_operating_system;
|
---|
247 | prcs = rt.exec(cmd, env);
|
---|
248 | }
|
---|
249 | else {
|
---|
250 | // Will inherit the GLI's environment, with GSDLHOME and GSDLOS set
|
---|
251 | prcs = rt.exec(cmd);
|
---|
252 | }
|
---|
253 | }
|
---|
254 |
|
---|
255 | //System.out.println(newcmd);
|
---|
256 |
|
---|
257 | InputStreamReader isr = new InputStreamReader(prcs.getErrorStream());
|
---|
258 | BufferedReader br = new BufferedReader(isr);
|
---|
259 | // Capture the standard error stream and seach for two particular occurances.
|
---|
260 | String line="";
|
---|
261 | boolean ignore_for_robots = false;
|
---|
262 | int max_download = DownloadJob.UNKNOWN_MAX;
|
---|
263 |
|
---|
264 |
|
---|
265 | while ((line = br.readLine()) != null && !line.trim().equals("<<Finished>>") && state != STOPPED) {
|
---|
266 |
|
---|
267 | if ( max_download == DownloadJob.UNKNOWN_MAX) {
|
---|
268 | if(line.lastIndexOf("<<Defined Maximum>>") != -1) {
|
---|
269 | max_download = DownloadJob.DEFINED_MAX;
|
---|
270 | }
|
---|
271 | else if (line.lastIndexOf("<<Undefined Maximum>>") != -1) {
|
---|
272 | max_download = DownloadJob.UNDEFINED_MAX;
|
---|
273 | }
|
---|
274 | }
|
---|
275 | else if(max_download == DownloadJob.UNDEFINED_MAX) {
|
---|
276 | DebugStream.println(line);
|
---|
277 | download_log.appendLine(line);
|
---|
278 | // The first magic special test is to see if we've just
|
---|
279 | // asked for the robots.txt file. If so we ignore
|
---|
280 | // the next add and then the next complete/error.
|
---|
281 | if(line.lastIndexOf("robots.txt;") != -1) {
|
---|
282 | DebugStream.println("***** Requesting robot.txt");
|
---|
283 | ignore_for_robots = true;
|
---|
284 | }
|
---|
285 | // If line contains "=> `" display text as the
|
---|
286 | // currently downloading url. Unique to add download.
|
---|
287 | else if(line.lastIndexOf("=> `") != -1) {
|
---|
288 | if(!ignore_for_robots) {
|
---|
289 | // Add download
|
---|
290 | String new_url = line.substring(line.indexOf("`") + 1, line.lastIndexOf("'"));
|
---|
291 | //addDownload("http:/" + new_url);
|
---|
292 | }
|
---|
293 | }
|
---|
294 | // If line contains "/s) - `" set currently
|
---|
295 | // downloading url to "Download Complete".
|
---|
296 | else if(line.lastIndexOf("/s) - `") != -1) {
|
---|
297 | String current_file_downloading = line.substring(line.indexOf("`") + 1, line.lastIndexOf("'"));
|
---|
298 | if(!ignore_for_robots) {
|
---|
299 | DebugStream.println("Not ignore for robots");
|
---|
300 | // Download complete
|
---|
301 | downloadComplete(current_file_downloading);
|
---|
302 | }
|
---|
303 | else {
|
---|
304 | DebugStream.println("Ignore for robots");
|
---|
305 | ignore_for_robots = false;
|
---|
306 | }
|
---|
307 | }
|
---|
308 | // The already there line begins "File `..." However this
|
---|
309 | // is only true in english, so instead I looked and there
|
---|
310 | // are few (if any at all) other messages than those above
|
---|
311 | // and not overwriting messages that use " `" so we'll
|
---|
312 | // look for that. Note this method is not guarenteed to be
|
---|
313 | // unique like the previous two.
|
---|
314 | else if(line.lastIndexOf(" `") != -1) {
|
---|
315 | // Not Overwriting
|
---|
316 | DebugStream.println("Already there.");
|
---|
317 | String new_url =
|
---|
318 | line.substring(line.indexOf("`") + 1, line.lastIndexOf("'"));
|
---|
319 | //addDownload("http:/" + new_url);
|
---|
320 | downloadWarning();
|
---|
321 | }
|
---|
322 | // Any other important message starts with the time in the form hh:mm:ss
|
---|
323 | else if(line.length() > 7) {
|
---|
324 | if(line.charAt(2) == ':' && line.charAt(5) == ':') {
|
---|
325 | if(!ignore_for_robots) {
|
---|
326 | DebugStream.println("Error.");
|
---|
327 | downloadFailed();
|
---|
328 | }
|
---|
329 | else {
|
---|
330 | ignore_for_robots = false;
|
---|
331 | }
|
---|
332 | }
|
---|
333 | }
|
---|
334 | }
|
---|
335 | else if (max_download == DownloadJob.DEFINED_MAX) {
|
---|
336 | if (line.lastIndexOf("<<Total number of record(s):") != -1) {
|
---|
337 | String total_ID = line.substring(line.indexOf(":") + 1, line.indexOf(">"));
|
---|
338 | progress.setTotalDownload((Integer.valueOf(total_ID)).intValue());
|
---|
339 | progress.resetFileCount();
|
---|
340 | }
|
---|
341 | else if (line.lastIndexOf("<<Done>>") != -1) {
|
---|
342 | progress.increaseFileCount();
|
---|
343 | }
|
---|
344 | else if(line.lastIndexOf("<<Done:") != -1) {
|
---|
345 | String completed_amount = line.substring(line.indexOf(":") + 1, line.indexOf(">"));
|
---|
346 | progress.increaseFileCount((Integer.valueOf(completed_amount)).intValue());
|
---|
347 | }
|
---|
348 |
|
---|
349 | DebugStream.println(line);
|
---|
350 | download_log.appendLine(line);
|
---|
351 | }
|
---|
352 | else {
|
---|
353 | System.out.println("Error!!");
|
---|
354 | System.exit(-1);
|
---|
355 | }
|
---|
356 | }
|
---|
357 | if(state == STOPPED) {
|
---|
358 | isr.close();
|
---|
359 | prcs.destroy(); // This doesn't always work, but it's worth a try
|
---|
360 | }
|
---|
361 |
|
---|
362 | }
|
---|
363 | catch (Exception ioe) {
|
---|
364 | //message(Utility.ERROR, ioe.toString());
|
---|
365 | //JTest
|
---|
366 | DebugStream.printStackTrace(ioe);
|
---|
367 | }
|
---|
368 | // If we've got to here and the state isn't STOPPED then the
|
---|
369 | // job is complete.
|
---|
370 | if(state == DownloadJob.RUNNING) {
|
---|
371 | progress.mirrorComplete();
|
---|
372 | previous_state = state;
|
---|
373 | state = DownloadJob.COMPLETE;
|
---|
374 |
|
---|
375 | }
|
---|
376 | // refresh the workspace tree
|
---|
377 | Gatherer.g_man.refreshWorkspaceTree(WorkspaceTree.DOWNLOADED_FILES_CHANGED);
|
---|
378 |
|
---|
379 | }
|
---|
380 |
|
---|
381 |
|
---|
382 | /** Called by the WGet native code when the current download is
|
---|
383 | * completed. In turn all download listeners are informed.
|
---|
384 | */
|
---|
385 | public void downloadComplete() {
|
---|
386 | progress.downloadComplete();
|
---|
387 | }
|
---|
388 |
|
---|
389 |
|
---|
390 | public void downloadComplete(String current_file_downloading)
|
---|
391 | {
|
---|
392 | progress.downloadComplete();
|
---|
393 | DebugStream.println("Download complete: " + current_file_downloading);
|
---|
394 | }
|
---|
395 |
|
---|
396 |
|
---|
397 | /** Called by the WGet native code when the requested download returns
|
---|
398 | * a status code other than 200.
|
---|
399 | */
|
---|
400 | public void downloadFailed() {
|
---|
401 | // TODO!!
|
---|
402 | //failed_urls.add(current_url); // It is the current url that failed
|
---|
403 | progress.downloadFailed();
|
---|
404 | //DebugStream.println("Download failed: " + current_url);
|
---|
405 | }
|
---|
406 |
|
---|
407 | /**
|
---|
408 | */
|
---|
409 | public void downloadWarning() {
|
---|
410 | progress.downloadWarning();
|
---|
411 | }
|
---|
412 |
|
---|
413 | public AppendLineOnlyFileDocument getLogDocument() {
|
---|
414 | return download_log;
|
---|
415 | }
|
---|
416 |
|
---|
417 | /**
|
---|
418 | * @return Returns the progress bar associated with this job.
|
---|
419 | */
|
---|
420 | public DownloadProgressBar getProgressBar() {
|
---|
421 | return progress;
|
---|
422 | }
|
---|
423 |
|
---|
424 | /** Called to discover if the user wanted this thread to run or if
|
---|
425 | * it is paused.
|
---|
426 | * @return An int representing the current DownloadJob state.
|
---|
427 | */
|
---|
428 | public int getState() {
|
---|
429 | return state;
|
---|
430 | }
|
---|
431 |
|
---|
432 | /** Returns the current state of the stop flag for this job.
|
---|
433 | * @return A boolean representing whether the user has requested to
|
---|
434 | * stop.
|
---|
435 | */
|
---|
436 | public boolean hasSignalledStop() {
|
---|
437 | if(state == DownloadJob.STOPPED || state == DownloadJob.PAUSED ||
|
---|
438 | state == DownloadJob.COMPLETE) {
|
---|
439 | return true;
|
---|
440 | }
|
---|
441 | return false;
|
---|
442 | }
|
---|
443 |
|
---|
444 | public void setState(int state) {
|
---|
445 | previous_state = this.state;
|
---|
446 | this.state = state;
|
---|
447 | }
|
---|
448 |
|
---|
449 | /** A convenience call.
|
---|
450 | * @return A String representing the url of the initial url (root node of the mirrored tree).
|
---|
451 | */
|
---|
452 | public String toString() {
|
---|
453 | return download_url;
|
---|
454 | }
|
---|
455 |
|
---|
456 | /** Called by the WGet native code to signal the current progress of
|
---|
457 | * downloading.
|
---|
458 | * @param current A long representing the number of bytes that have
|
---|
459 | * been downloaded since the last update.
|
---|
460 | * @param expected A long representing the total number of bytes
|
---|
461 | * expected for this download.
|
---|
462 | */
|
---|
463 | public void updateProgress(long current, long expected) {
|
---|
464 | progress.updateProgress(current, expected);
|
---|
465 | }
|
---|
466 | }
|
---|