1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace AOE\Crawler\Backend; |
6
|
|
|
|
7
|
|
|
/*************************************************************** |
8
|
|
|
* Copyright notice |
9
|
|
|
* |
10
|
|
|
* (c) 2018 AOE GmbH <[email protected]> |
11
|
|
|
* |
12
|
|
|
* All rights reserved |
13
|
|
|
* |
14
|
|
|
* This script is part of the TYPO3 project. The TYPO3 project is |
15
|
|
|
* free software; you can redistribute it and/or modify |
16
|
|
|
* it under the terms of the GNU General Public License as published by |
17
|
|
|
* the Free Software Foundation; either version 3 of the License, or |
18
|
|
|
* (at your option) any later version. |
19
|
|
|
* |
20
|
|
|
* The GNU General Public License can be found at |
21
|
|
|
* http://www.gnu.org/copyleft/gpl.html. |
22
|
|
|
* |
23
|
|
|
* This script is distributed in the hope that it will be useful, |
24
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
25
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
26
|
|
|
* GNU General Public License for more details. |
27
|
|
|
* |
28
|
|
|
* This copyright notice MUST APPEAR in all copies of the script! |
29
|
|
|
***************************************************************/ |
30
|
|
|
|
31
|
|
|
use AOE\Crawler\Configuration\ExtensionConfigurationProvider; |
32
|
|
|
use AOE\Crawler\Controller\CrawlerController; |
33
|
|
|
use AOE\Crawler\Domain\Model\Reason; |
34
|
|
|
use AOE\Crawler\Domain\Repository\ProcessRepository; |
35
|
|
|
use AOE\Crawler\Domain\Repository\QueueRepository; |
36
|
|
|
use AOE\Crawler\Event\EventDispatcher; |
37
|
|
|
use AOE\Crawler\Hooks\CrawlerHookInterface; |
38
|
|
|
use AOE\Crawler\Service\ProcessService; |
39
|
|
|
use AOE\Crawler\Utility\MessageUtility; |
40
|
|
|
use AOE\Crawler\Utility\PhpBinaryUtility; |
41
|
|
|
use Psr\Http\Message\UriInterface; |
42
|
|
|
use TYPO3\CMS\Backend\Routing\UriBuilder; |
43
|
|
|
use TYPO3\CMS\Backend\Template\ModuleTemplate; |
44
|
|
|
use TYPO3\CMS\Backend\Tree\View\PageTreeView; |
45
|
|
|
use TYPO3\CMS\Backend\Utility\BackendUtility; |
46
|
|
|
use TYPO3\CMS\Core\Database\ConnectionPool; |
47
|
|
|
use TYPO3\CMS\Core\Database\Query\QueryBuilder; |
48
|
|
|
use TYPO3\CMS\Core\Http\Uri; |
49
|
|
|
use TYPO3\CMS\Core\Imaging\Icon; |
50
|
|
|
use TYPO3\CMS\Core\Imaging\IconFactory; |
51
|
|
|
use TYPO3\CMS\Core\Localization\LanguageService; |
52
|
|
|
use TYPO3\CMS\Core\Utility\CommandUtility; |
53
|
|
|
use TYPO3\CMS\Core\Utility\CsvUtility; |
54
|
|
|
use TYPO3\CMS\Core\Utility\DebugUtility; |
55
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
56
|
|
|
use TYPO3\CMS\Core\Utility\MathUtility; |
57
|
|
|
use TYPO3\CMS\Extbase\Object\ObjectManager; |
58
|
|
|
use TYPO3\CMS\Fluid\View\StandaloneView; |
59
|
|
|
use TYPO3\CMS\Info\Controller\InfoModuleController; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* Function for Info module, containing three main actions: |
63
|
|
|
* - List of all queued items |
64
|
|
|
* - Log functionality |
65
|
|
|
* - Process overview |
66
|
|
|
*/ |
67
|
|
|
class BackendModule |
68
|
|
|
{ |
69
|
|
|
/** |
70
|
|
|
* @var InfoModuleController Contains a reference to the parent calling object |
71
|
|
|
*/ |
72
|
|
|
protected $pObj; |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* The current page ID |
76
|
|
|
* @var int |
77
|
|
|
*/ |
78
|
|
|
protected $id; |
79
|
|
|
|
80
|
|
|
// Internal, dynamic: |
81
|
|
|
/** |
82
|
|
|
* @var array |
83
|
|
|
*/ |
84
|
|
|
protected $duplicateTrack = []; |
85
|
|
|
|
86
|
|
|
/** |
87
|
|
|
* @var bool |
88
|
|
|
*/ |
89
|
|
|
protected $submitCrawlUrls = false; |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* @var bool |
93
|
|
|
*/ |
94
|
|
|
protected $downloadCrawlUrls = false; |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* @var int |
98
|
|
|
*/ |
99
|
|
|
protected $scheduledTime = 0; |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* @var int |
103
|
|
|
*/ |
104
|
|
|
protected $reqMinute = 1000; |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* @var array holds the selection of configuration from the configuration selector box |
108
|
|
|
*/ |
109
|
|
|
protected $incomingConfigurationSelection = []; |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* @var CrawlerController |
113
|
|
|
*/ |
114
|
|
|
protected $crawlerController; |
115
|
|
|
|
116
|
|
|
/** |
117
|
|
|
* @var array |
118
|
|
|
*/ |
119
|
|
|
protected $CSVaccu = []; |
120
|
|
|
|
121
|
|
|
/** |
122
|
|
|
* If true the user requested a CSV export of the queue |
123
|
|
|
* |
124
|
|
|
* @var boolean |
125
|
|
|
*/ |
126
|
|
|
protected $CSVExport = false; |
127
|
|
|
|
128
|
|
|
/** |
129
|
|
|
* @var array |
130
|
|
|
*/ |
131
|
|
|
protected $downloadUrls = []; |
132
|
|
|
|
133
|
|
|
/** |
134
|
|
|
* Holds the configuration from ext_conf_template loaded by getExtensionConfiguration() |
135
|
|
|
* |
136
|
|
|
* @var array |
137
|
|
|
*/ |
138
|
|
|
protected $extensionSettings = []; |
139
|
|
|
|
140
|
|
|
/** |
141
|
|
|
* Indicate that an flash message with an error is present. |
142
|
|
|
* |
143
|
|
|
* @var boolean |
144
|
|
|
*/ |
145
|
|
|
protected $isErrorDetected = false; |
146
|
|
|
|
147
|
|
|
/** |
148
|
|
|
* @var ProcessService |
149
|
|
|
*/ |
150
|
|
|
protected $processManager; |
151
|
|
|
|
152
|
|
|
/** |
153
|
|
|
* @var QueryBuilder |
154
|
|
|
*/ |
155
|
|
|
protected $queryBuilder; |
156
|
|
|
|
157
|
|
|
/** |
158
|
|
|
* @var QueueRepository |
159
|
|
|
*/ |
160
|
|
|
protected $queueRepository; |
161
|
|
|
|
162
|
|
|
/** |
163
|
|
|
* @var StandaloneView |
164
|
|
|
*/ |
165
|
|
|
protected $view; |
166
|
|
|
|
167
|
|
|
/** |
168
|
|
|
* @var IconFactory |
169
|
|
|
*/ |
170
|
|
|
protected $iconFactory; |
171
|
|
|
|
172
|
|
|
public function __construct() |
173
|
|
|
{ |
174
|
|
|
$objectManger = GeneralUtility::makeInstance(ObjectManager::class); |
175
|
|
|
$this->processManager = $objectManger->get(ProcessService::class); |
176
|
|
|
$this->queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable('tx_crawler_queue'); |
177
|
|
|
$this->queueRepository = $objectManger->get(QueueRepository::class); |
178
|
|
|
$this->initializeView(); |
179
|
|
|
$this->extensionSettings = GeneralUtility::makeInstance(ExtensionConfigurationProvider::class)->getExtensionConfiguration(); |
180
|
|
|
$this->iconFactory = GeneralUtility::makeInstance(IconFactory::class); |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
/** |
184
|
|
|
* Called by the InfoModuleController |
185
|
|
|
* @param InfoModuleController $pObj |
186
|
|
|
*/ |
187
|
|
|
public function init(InfoModuleController $pObj): void |
188
|
|
|
{ |
189
|
|
|
$this->pObj = $pObj; |
190
|
|
|
$this->id = (int)GeneralUtility::_GP('id'); |
191
|
|
|
// Setting MOD_MENU items as we need them for logging: |
192
|
|
|
$this->pObj->MOD_MENU = array_merge($this->pObj->MOD_MENU, $this->modMenu()); |
193
|
|
|
} |
194
|
|
|
|
195
|
|
|
/** |
196
|
|
|
* Additions to the function menu array |
197
|
|
|
* |
198
|
|
|
* @return array Menu array |
199
|
|
|
*/ |
200
|
|
|
public function modMenu(): array |
201
|
|
|
{ |
202
|
|
|
return [ |
203
|
|
|
'depth' => [ |
204
|
|
|
0 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_0'), |
205
|
|
|
1 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_1'), |
206
|
|
|
2 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_2'), |
207
|
|
|
3 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_3'), |
208
|
|
|
4 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_4'), |
209
|
|
|
99 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_infi'), |
210
|
|
|
], |
211
|
|
|
'crawlaction' => [ |
212
|
|
|
'start' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.start'), |
213
|
|
|
'log' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.log'), |
214
|
|
|
'multiprocess' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.multiprocess'), |
215
|
|
|
], |
216
|
|
|
'log_resultLog' => '', |
217
|
|
|
'log_feVars' => '', |
218
|
|
|
'processListMode' => '', |
219
|
|
|
'log_display' => [ |
220
|
|
|
'all' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.all'), |
221
|
|
|
'pending' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.pending'), |
222
|
|
|
'finished' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.finished'), |
223
|
|
|
], |
224
|
|
|
'itemsPerPage' => [ |
225
|
|
|
'5' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage.5'), |
226
|
|
|
'10' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage.10'), |
227
|
|
|
'50' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage.50'), |
228
|
|
|
'0' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage.0'), |
229
|
|
|
], |
230
|
|
|
]; |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
public function main(): string |
234
|
|
|
{ |
235
|
|
|
if (empty($this->pObj->MOD_SETTINGS['processListMode'])) { |
236
|
|
|
$this->pObj->MOD_SETTINGS['processListMode'] = 'simple'; |
237
|
|
|
} |
238
|
|
|
$this->view->assign('currentPageId', $this->id); |
239
|
|
|
|
240
|
|
|
$selectedAction = (string)$this->pObj->MOD_SETTINGS['crawlaction'] ?? 'start'; |
241
|
|
|
|
242
|
|
|
// Type function menu: |
243
|
|
|
$actionDropdown = BackendUtility::getFuncMenu( |
244
|
|
|
$this->id, |
245
|
|
|
'SET[crawlaction]', |
246
|
|
|
$selectedAction, |
247
|
|
|
$this->pObj->MOD_MENU['crawlaction'] |
248
|
|
|
); |
249
|
|
|
|
250
|
|
|
$theOutput = '<h2>' . htmlspecialchars($this->getLanguageService()->getLL('title')) . '</h2>' . $actionDropdown; |
251
|
|
|
|
252
|
|
|
// Branch based on type: |
253
|
|
|
switch ($selectedAction) { |
254
|
|
|
case 'start': |
255
|
|
|
$theOutput .= $this->showCrawlerInformationAction(); |
256
|
|
|
break; |
257
|
|
|
case 'log': |
258
|
|
|
|
259
|
|
|
// Additional menus for the log type: |
260
|
|
|
$theOutput .= BackendUtility::getFuncMenu( |
261
|
|
|
$this->id, |
262
|
|
|
'SET[depth]', |
263
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
264
|
|
|
$this->pObj->MOD_MENU['depth'] |
265
|
|
|
); |
266
|
|
|
|
267
|
|
|
$quiPart = GeneralUtility::_GP('qid_details') ? '&qid_details=' . (int)GeneralUtility::_GP('qid_details') : ''; |
268
|
|
|
$setId = (int)GeneralUtility::_GP('setID'); |
269
|
|
|
|
270
|
|
|
$theOutput .= '<br><br>' . |
271
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.display') . ': ' . BackendUtility::getFuncMenu( |
272
|
|
|
$this->id, |
273
|
|
|
'SET[log_display]', |
274
|
|
|
$this->pObj->MOD_SETTINGS['log_display'], |
275
|
|
|
$this->pObj->MOD_MENU['log_display'], |
276
|
|
|
'index.php', |
277
|
|
|
'&setID=' . $setId |
278
|
|
|
) . ' ' . |
279
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.showresultlog') . ': ' . BackendUtility::getFuncCheck( |
280
|
|
|
$this->id, |
281
|
|
|
'SET[log_resultLog]', |
282
|
|
|
$this->pObj->MOD_SETTINGS['log_resultLog'], |
283
|
|
|
'index.php', |
284
|
|
|
'&setID=' . $setId . $quiPart |
285
|
|
|
) . ' ' . |
286
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.showfevars') . ': ' . BackendUtility::getFuncCheck( |
287
|
|
|
$this->id, |
288
|
|
|
'SET[log_feVars]', |
289
|
|
|
$this->pObj->MOD_SETTINGS['log_feVars'], |
290
|
|
|
'index.php', |
291
|
|
|
'&setID=' . $setId . $quiPart |
292
|
|
|
) . ' ' . |
293
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage') . ': ' . |
294
|
|
|
BackendUtility::getFuncMenu( |
295
|
|
|
$this->id, |
296
|
|
|
'SET[itemsPerPage]', |
297
|
|
|
$this->pObj->MOD_SETTINGS['itemsPerPage'], |
298
|
|
|
$this->pObj->MOD_MENU['itemsPerPage'] |
299
|
|
|
); |
300
|
|
|
$theOutput .= $this->showLogAction(); |
301
|
|
|
break; |
302
|
|
|
case 'multiprocess': |
303
|
|
|
$theOutput .= $this->processOverviewAction(); |
304
|
|
|
break; |
305
|
|
|
} |
306
|
|
|
|
307
|
|
|
return $theOutput; |
308
|
|
|
} |
309
|
|
|
|
310
|
|
|
/******************************* |
311
|
|
|
* |
312
|
|
|
* Generate URLs for crawling: |
313
|
|
|
* |
314
|
|
|
******************************/ |
315
|
|
|
|
316
|
|
|
/** |
317
|
|
|
* Show a list of URLs to be crawled for each page |
318
|
|
|
* @return string |
319
|
|
|
*/ |
320
|
|
|
protected function showCrawlerInformationAction(): string |
321
|
|
|
{ |
322
|
|
|
$this->view->setTemplate('ShowCrawlerInformation'); |
323
|
|
|
if (empty($this->id)) { |
324
|
|
|
$this->isErrorDetected = true; |
325
|
|
|
MessageUtility::addErrorMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noPageSelected')); |
326
|
|
|
} else { |
327
|
|
|
$crawlerParameter = GeneralUtility::_GP('_crawl'); |
328
|
|
|
$downloadParameter = GeneralUtility::_GP('_download'); |
329
|
|
|
|
330
|
|
|
$this->duplicateTrack = []; |
331
|
|
|
$this->submitCrawlUrls = isset($crawlerParameter); |
332
|
|
|
$this->downloadCrawlUrls = isset($downloadParameter); |
333
|
|
|
$this->makeCrawlerProcessableChecks(); |
334
|
|
|
|
335
|
|
|
switch ((string) GeneralUtility::_GP('tstamp')) { |
336
|
|
|
case 'midnight': |
337
|
|
|
$this->scheduledTime = mktime(0, 0, 0); |
338
|
|
|
break; |
339
|
|
|
case '04:00': |
340
|
|
|
$this->scheduledTime = mktime(0, 0, 0) + 4 * 3600; |
341
|
|
|
break; |
342
|
|
|
case 'now': |
343
|
|
|
default: |
344
|
|
|
$this->scheduledTime = time(); |
345
|
|
|
break; |
346
|
|
|
} |
347
|
|
|
|
348
|
|
|
$this->incomingConfigurationSelection = GeneralUtility::_GP('configurationSelection'); |
349
|
|
|
$this->incomingConfigurationSelection = is_array($this->incomingConfigurationSelection) ? $this->incomingConfigurationSelection : []; |
350
|
|
|
|
351
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
352
|
|
|
$this->crawlerController->setAccessMode('gui'); |
353
|
|
|
$this->crawlerController->setID = GeneralUtility::md5int(microtime()); |
354
|
|
|
|
355
|
|
|
$code = ''; |
356
|
|
|
$noConfigurationSelected = empty($this->incomingConfigurationSelection) |
357
|
|
|
|| (count($this->incomingConfigurationSelection) == 1 && empty($this->incomingConfigurationSelection[0])); |
358
|
|
|
if ($noConfigurationSelected) { |
359
|
|
|
MessageUtility::addWarningMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noConfigSelected')); |
360
|
|
|
} else { |
361
|
|
|
if ($this->submitCrawlUrls) { |
362
|
|
|
$reason = new Reason(); |
363
|
|
|
$reason->setReason(Reason::REASON_GUI_SUBMIT); |
364
|
|
|
$reason->setDetailText('The user ' . $GLOBALS['BE_USER']->user['username'] . ' added pages to the crawler queue manually'); |
365
|
|
|
|
366
|
|
|
EventDispatcher::getInstance()->post( |
367
|
|
|
'invokeQueueChange', |
368
|
|
|
strval($this->findCrawler()->setID), |
369
|
|
|
['reason' => $reason] |
370
|
|
|
); |
371
|
|
|
} |
372
|
|
|
|
373
|
|
|
$code = $this->crawlerController->getPageTreeAndUrls( |
374
|
|
|
$this->id, |
375
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
376
|
|
|
$this->scheduledTime, |
377
|
|
|
$this->reqMinute, |
378
|
|
|
$this->submitCrawlUrls, |
379
|
|
|
$this->downloadCrawlUrls, |
380
|
|
|
[], // Do not filter any processing instructions |
381
|
|
|
$this->incomingConfigurationSelection |
382
|
|
|
); |
383
|
|
|
} |
384
|
|
|
|
385
|
|
|
$this->downloadUrls = $this->crawlerController->downloadUrls; |
386
|
|
|
$this->duplicateTrack = $this->crawlerController->duplicateTrack; |
387
|
|
|
|
388
|
|
|
$this->view->assign('noConfigurationSelected', $noConfigurationSelected); |
389
|
|
|
$this->view->assign('submitCrawlUrls', $this->submitCrawlUrls); |
390
|
|
|
$this->view->assign('amountOfUrls', count(array_keys($this->duplicateTrack))); |
391
|
|
|
$this->view->assign('selectors', $this->generateConfigurationSelectors()); |
392
|
|
|
$this->view->assign('code', $code); |
393
|
|
|
|
394
|
|
|
// Download Urls to crawl: |
395
|
|
|
if ($this->downloadCrawlUrls) { |
396
|
|
|
// Creating output header: |
397
|
|
|
header('Content-Type: application/octet-stream'); |
398
|
|
|
header('Content-Disposition: attachment; filename=CrawlerUrls.txt'); |
399
|
|
|
|
400
|
|
|
// Printing the content of the CSV lines: |
401
|
|
|
echo implode(chr(13) . chr(10), $this->downloadUrls); |
402
|
|
|
exit; |
|
|
|
|
403
|
|
|
} |
404
|
|
|
} |
405
|
|
|
return $this->view->render(); |
406
|
|
|
} |
407
|
|
|
|
408
|
|
|
/** |
409
|
|
|
* Generates the configuration selectors for compiling URLs: |
410
|
|
|
*/ |
411
|
|
|
protected function generateConfigurationSelectors(): array |
412
|
|
|
{ |
413
|
|
|
$selectors = []; |
414
|
|
|
$selectors['depth'] = $this->selectorBox( |
415
|
|
|
[ |
416
|
|
|
0 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_0'), |
417
|
|
|
1 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_1'), |
418
|
|
|
2 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_2'), |
419
|
|
|
3 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_3'), |
420
|
|
|
4 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_4'), |
421
|
|
|
99 => $this->getLanguageService()->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_infi'), |
422
|
|
|
], |
423
|
|
|
'SET[depth]', |
424
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
425
|
|
|
false |
426
|
|
|
); |
427
|
|
|
|
428
|
|
|
// Configurations |
429
|
|
|
$availableConfigurations = $this->crawlerController->getConfigurationsForBranch((int)$this->id, $this->pObj->MOD_SETTINGS['depth'] ?: 0); |
430
|
|
|
$selectors['configurations'] = $this->selectorBox( |
431
|
|
|
empty($availableConfigurations) ? [] : array_combine($availableConfigurations, $availableConfigurations), |
|
|
|
|
432
|
|
|
'configurationSelection', |
433
|
|
|
$this->incomingConfigurationSelection, |
434
|
|
|
true |
435
|
|
|
); |
436
|
|
|
|
437
|
|
|
// Scheduled time: |
438
|
|
|
$selectors['scheduled'] = $this->selectorBox( |
439
|
|
|
[ |
440
|
|
|
'now' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.time.now'), |
441
|
|
|
'midnight' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.time.midnight'), |
442
|
|
|
'04:00' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.time.4am'), |
443
|
|
|
], |
444
|
|
|
'tstamp', |
445
|
|
|
GeneralUtility::_POST('tstamp'), |
446
|
|
|
false |
447
|
|
|
); |
448
|
|
|
|
449
|
|
|
return $selectors; |
450
|
|
|
} |
451
|
|
|
|
452
|
|
|
/******************************* |
453
|
|
|
* |
454
|
|
|
* Shows log of indexed URLs |
455
|
|
|
* |
456
|
|
|
******************************/ |
457
|
|
|
|
458
|
|
|
protected function showLogAction(): string |
459
|
|
|
{ |
460
|
|
|
$this->view->setTemplate('ShowLog'); |
461
|
|
|
if (empty($this->id)) { |
462
|
|
|
$this->isErrorDetected = true; |
463
|
|
|
MessageUtility::addErrorMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noPageSelected')); |
464
|
|
|
} else { |
465
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
466
|
|
|
$this->crawlerController->setAccessMode('gui'); |
467
|
|
|
$this->crawlerController->setID = GeneralUtility::md5int(microtime()); |
468
|
|
|
|
469
|
|
|
$csvExport = GeneralUtility::_POST('_csv'); |
470
|
|
|
$this->CSVExport = isset($csvExport); |
471
|
|
|
|
472
|
|
|
// Read URL: |
473
|
|
|
if (GeneralUtility::_GP('qid_read')) { |
474
|
|
|
$this->crawlerController->readUrl((int)GeneralUtility::_GP('qid_read'), true); |
475
|
|
|
} |
476
|
|
|
|
477
|
|
|
// Look for set ID sent - if it is, we will display contents of that set: |
478
|
|
|
$showSetId = (int)GeneralUtility::_GP('setID'); |
479
|
|
|
|
480
|
|
|
$queueId = GeneralUtility::_GP('qid_details'); |
481
|
|
|
$this->view->assign('queueId', $queueId); |
482
|
|
|
$this->view->assign('setId', $showSetId); |
483
|
|
|
// Show details: |
484
|
|
|
if ($queueId) { |
485
|
|
|
// Get entry record: |
486
|
|
|
$q_entry = $this->queryBuilder |
487
|
|
|
->from('tx_crawler_queue') |
488
|
|
|
->select('*') |
489
|
|
|
->where( |
490
|
|
|
$this->queryBuilder->expr()->eq('qid', $this->queryBuilder->createNamedParameter($queueId)) |
491
|
|
|
) |
492
|
|
|
->execute() |
493
|
|
|
->fetch(); |
494
|
|
|
|
495
|
|
|
// Explode values |
496
|
|
|
$q_entry['parameters'] = unserialize($q_entry['parameters']); |
497
|
|
|
$q_entry['result_data'] = unserialize($q_entry['result_data']); |
498
|
|
|
$resStatus = $this->getResStatus($q_entry['result_data']); |
499
|
|
|
if (is_array($q_entry['result_data'])) { |
500
|
|
|
$q_entry['result_data']['content'] = unserialize($q_entry['result_data']['content']); |
501
|
|
|
if (!$this->pObj->MOD_SETTINGS['log_resultLog']) { |
502
|
|
|
unset($q_entry['result_data']['content']['log']); |
503
|
|
|
} |
504
|
|
|
} |
505
|
|
|
|
506
|
|
|
$this->view->assign('queueStatus', $resStatus); |
507
|
|
|
$this->view->assign('queueDetails', DebugUtility::viewArray($q_entry)); |
508
|
|
|
} else { |
509
|
|
|
// Show list |
510
|
|
|
// Drawing tree: |
511
|
|
|
$tree = GeneralUtility::makeInstance(PageTreeView::class); |
512
|
|
|
$perms_clause = $GLOBALS['BE_USER']->getPagePermsClause(1); |
513
|
|
|
$tree->init('AND ' . $perms_clause); |
514
|
|
|
|
515
|
|
|
// Set root row: |
516
|
|
|
$pageinfo = BackendUtility::readPageAccess( |
517
|
|
|
$this->id, |
518
|
|
|
$perms_clause |
519
|
|
|
); |
520
|
|
|
$HTML = $this->iconFactory->getIconForRecord('pages', $pageinfo, Icon::SIZE_SMALL)->render(); |
521
|
|
|
$tree->tree[] = [ |
522
|
|
|
'row' => $pageinfo, |
523
|
|
|
'HTML' => $HTML, |
524
|
|
|
]; |
525
|
|
|
|
526
|
|
|
// Get branch beneath: |
527
|
|
|
if ($this->pObj->MOD_SETTINGS['depth']) { |
528
|
|
|
$tree->getTree($this->id, $this->pObj->MOD_SETTINGS['depth']); |
529
|
|
|
} |
530
|
|
|
|
531
|
|
|
// If Flush button is pressed, flush tables instead of selecting entries: |
532
|
|
|
if (GeneralUtility::_POST('_flush')) { |
533
|
|
|
$doFlush = true; |
534
|
|
|
$doFullFlush = false; |
535
|
|
|
} elseif (GeneralUtility::_POST('_flush_all')) { |
536
|
|
|
$doFlush = true; |
537
|
|
|
$doFullFlush = true; |
538
|
|
|
} else { |
539
|
|
|
$doFlush = false; |
540
|
|
|
$doFullFlush = false; |
541
|
|
|
} |
542
|
|
|
$itemsPerPage = (int)$this->pObj->MOD_SETTINGS['itemsPerPage']; |
543
|
|
|
// Traverse page tree: |
544
|
|
|
$code = ''; |
545
|
|
|
$count = 0; |
546
|
|
|
foreach ($tree->tree as $data) { |
547
|
|
|
// Get result: |
548
|
|
|
$logEntriesOfPage = $this->crawlerController->getLogEntriesForPageId( |
549
|
|
|
(int)$data['row']['uid'], |
550
|
|
|
$this->pObj->MOD_SETTINGS['log_display'], |
551
|
|
|
$doFlush, |
552
|
|
|
$doFullFlush, |
553
|
|
|
$itemsPerPage |
554
|
|
|
); |
555
|
|
|
|
556
|
|
|
$code .= $this->drawLog_addRows( |
557
|
|
|
$logEntriesOfPage, |
558
|
|
|
$data['HTML'] . BackendUtility::getRecordTitle('pages', $data['row'], true) |
559
|
|
|
); |
560
|
|
|
if (++$count === 1000) { |
561
|
|
|
break; |
562
|
|
|
} |
563
|
|
|
} |
564
|
|
|
$this->view->assign('code', $code); |
565
|
|
|
} |
566
|
|
|
|
567
|
|
|
if ($this->CSVExport) { |
568
|
|
|
$this->outputCsvFile(); |
569
|
|
|
} |
570
|
|
|
} |
571
|
|
|
$this->view->assign('showResultLog', (bool)$this->pObj->MOD_SETTINGS['log_resultLog']); |
572
|
|
|
$this->view->assign('showFeVars', (bool)$this->pObj->MOD_SETTINGS['log_feVars']); |
573
|
|
|
return $this->view->render(); |
574
|
|
|
} |
575
|
|
|
|
576
|
|
|
/** |
577
|
|
|
* Outputs the CSV file and sets the correct headers |
578
|
|
|
*/ |
579
|
|
|
protected function outputCsvFile(): void |
580
|
|
|
{ |
581
|
|
|
if (!count($this->CSVaccu)) { |
582
|
|
|
MessageUtility::addWarningMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:message.canNotExportEmptyQueueToCsvText')); |
583
|
|
|
return; |
584
|
|
|
} |
585
|
|
|
$csvLines = []; |
586
|
|
|
|
587
|
|
|
// Field names: |
588
|
|
|
reset($this->CSVaccu); |
589
|
|
|
$fieldNames = array_keys(current($this->CSVaccu)); |
590
|
|
|
$csvLines[] = CsvUtility::csvValues($fieldNames); |
591
|
|
|
|
592
|
|
|
// Data: |
593
|
|
|
foreach ($this->CSVaccu as $row) { |
594
|
|
|
$csvLines[] = CsvUtility::csvValues($row); |
595
|
|
|
} |
596
|
|
|
|
597
|
|
|
// Creating output header: |
598
|
|
|
header('Content-Type: application/octet-stream'); |
599
|
|
|
header('Content-Disposition: attachment; filename=CrawlerLog.csv'); |
600
|
|
|
|
601
|
|
|
// Printing the content of the CSV lines: |
602
|
|
|
echo implode(chr(13) . chr(10), $csvLines); |
603
|
|
|
exit; |
|
|
|
|
604
|
|
|
} |
605
|
|
|
|
606
|
|
|
/** |
607
|
|
|
* Create the rows for display of the page tree |
608
|
|
|
* For each page a number of rows are shown displaying GET variable configuration |
609
|
|
|
* |
610
|
|
|
* @param array $logEntriesOfPage Log items of one page |
611
|
|
|
* @param string $titleString Title string |
612
|
|
|
* @return string HTML <tr> content (one or more) |
613
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
614
|
|
|
*/ |
615
|
|
|
protected function drawLog_addRows(array $logEntriesOfPage, string $titleString): string |
616
|
|
|
{ |
617
|
|
|
$colSpan = 9 |
618
|
|
|
+ ($this->pObj->MOD_SETTINGS['log_resultLog'] ? -1 : 0) |
619
|
|
|
+ ($this->pObj->MOD_SETTINGS['log_feVars'] ? 3 : 0); |
620
|
|
|
|
621
|
|
|
if (!empty($logEntriesOfPage)) { |
622
|
|
|
$setId = (int)GeneralUtility::_GP('setID'); |
623
|
|
|
$refreshIcon = $this->iconFactory->getIcon('actions-system-refresh', Icon::SIZE_SMALL); |
624
|
|
|
// Traverse parameter combinations: |
625
|
|
|
$c = 0; |
626
|
|
|
$content = ''; |
627
|
|
|
foreach ($logEntriesOfPage as $vv) { |
628
|
|
|
// Title column: |
629
|
|
|
if (!$c) { |
630
|
|
|
$titleClm = '<td rowspan="' . count($logEntriesOfPage) . '">' . $titleString . '</td>'; |
631
|
|
|
} else { |
632
|
|
|
$titleClm = ''; |
633
|
|
|
} |
634
|
|
|
|
635
|
|
|
// Result: |
636
|
|
|
$resLog = $this->getResultLog($vv); |
637
|
|
|
|
638
|
|
|
$resultData = $vv['result_data'] ? unserialize($vv['result_data']) : []; |
639
|
|
|
$resStatus = $this->getResStatus($resultData); |
640
|
|
|
|
641
|
|
|
// Compile row: |
642
|
|
|
$parameters = unserialize($vv['parameters']); |
643
|
|
|
|
644
|
|
|
// Put data into array: |
645
|
|
|
$rowData = []; |
646
|
|
|
if ($this->pObj->MOD_SETTINGS['log_resultLog']) { |
647
|
|
|
$rowData['result_log'] = $resLog; |
648
|
|
|
} else { |
649
|
|
|
$rowData['scheduled'] = ($vv['scheduled'] > 0) ? BackendUtility::datetime($vv['scheduled']) : ' ' . $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.immediate'); |
650
|
|
|
$rowData['exec_time'] = $vv['exec_time'] ? BackendUtility::datetime($vv['exec_time']) : '-'; |
651
|
|
|
} |
652
|
|
|
$rowData['result_status'] = GeneralUtility::fixed_lgd_cs($resStatus, 50); |
653
|
|
|
$rowData['url'] = '<a href="' . htmlspecialchars($parameters['url']) . '" target="_newWIndow">' . htmlspecialchars($parameters['url']) . '</a>'; |
654
|
|
|
$rowData['feUserGroupList'] = $parameters['feUserGroupList'] ?: ''; |
655
|
|
|
$rowData['procInstructions'] = is_array($parameters['procInstructions']) ? implode('; ', $parameters['procInstructions']) : ''; |
656
|
|
|
$rowData['set_id'] = (string)$vv['set_id']; |
657
|
|
|
|
658
|
|
|
if ($this->pObj->MOD_SETTINGS['log_feVars']) { |
659
|
|
|
$resFeVars = $this->getResFeVars($resultData ?: []); |
660
|
|
|
$rowData['tsfe_id'] = $resFeVars['id']; |
661
|
|
|
$rowData['tsfe_gr_list'] = $resFeVars['gr_list']; |
662
|
|
|
$rowData['tsfe_no_cache'] = $resFeVars['no_cache']; |
663
|
|
|
} |
664
|
|
|
|
665
|
|
|
// Put rows together: |
666
|
|
|
$content .= ' |
667
|
|
|
<tr> |
668
|
|
|
' . $titleClm . ' |
669
|
|
|
<td><a href="' . $this->getInfoModuleUrl(['qid_details' => $vv['qid'], 'setID' => $setId]) . '">' . htmlspecialchars((string)$vv['qid']) . '</a></td> |
670
|
|
|
<td><a href="' . $this->getInfoModuleUrl(['qid_read' => $vv['qid'], 'setID' => $setId]) . '">' . $refreshIcon . '</a></td>'; |
671
|
|
|
foreach ($rowData as $fKey => $value) { |
672
|
|
|
if ($fKey === 'url') { |
673
|
|
|
$content .= '<td>' . $value . '</td>'; |
674
|
|
|
} else { |
675
|
|
|
$content .= '<td>' . nl2br(htmlspecialchars($value)) . '</td>'; |
676
|
|
|
} |
677
|
|
|
} |
678
|
|
|
$content .= '</tr>'; |
679
|
|
|
$c++; |
680
|
|
|
|
681
|
|
|
if ($this->CSVExport) { |
682
|
|
|
// Only for CSV (adding qid and scheduled/exec_time if needed): |
683
|
|
|
$rowData['result_log'] = implode('// ', explode(chr(10), $resLog)); |
684
|
|
|
$rowData['qid'] = $vv['qid']; |
685
|
|
|
$rowData['scheduled'] = BackendUtility::datetime($vv['scheduled']); |
686
|
|
|
$rowData['exec_time'] = $vv['exec_time'] ? BackendUtility::datetime($vv['exec_time']) : '-'; |
687
|
|
|
$this->CSVaccu[] = $rowData; |
688
|
|
|
} |
689
|
|
|
} |
690
|
|
|
} else { |
691
|
|
|
// Compile row: |
692
|
|
|
$content = ' |
693
|
|
|
<tr> |
694
|
|
|
<td>' . $titleString . '</td> |
695
|
|
|
<td colspan="' . $colSpan . '"><em>' . $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noentries') . '</em></td> |
696
|
|
|
</tr>'; |
697
|
|
|
} |
698
|
|
|
|
699
|
|
|
return $content; |
700
|
|
|
} |
701
|
|
|
|
702
|
|
|
/** |
703
|
|
|
* Find Fe vars |
704
|
|
|
* |
705
|
|
|
* @param array $resultData |
706
|
|
|
* @return array |
707
|
|
|
*/ |
708
|
|
|
protected function getResFeVars(array $resultData): array |
709
|
|
|
{ |
710
|
|
|
if (empty($resultData)) { |
711
|
|
|
return []; |
712
|
|
|
} |
713
|
|
|
$requestResult = unserialize($resultData['content']); |
714
|
|
|
return $requestResult['vars']; |
715
|
|
|
} |
716
|
|
|
|
717
|
|
|
/** |
718
|
|
|
* Extract the log information from the current row and retrive it as formatted string. |
719
|
|
|
* |
720
|
|
|
* @param array $resultRow |
721
|
|
|
* @return string |
722
|
|
|
*/ |
723
|
|
|
protected function getResultLog($resultRow) |
724
|
|
|
{ |
725
|
|
|
$content = ''; |
726
|
|
|
if (is_array($resultRow) && array_key_exists('result_data', $resultRow)) { |
727
|
|
|
$requestContent = unserialize($resultRow['result_data']) ?: ['content' => '']; |
728
|
|
|
$requestResult = unserialize($requestContent['content']); |
729
|
|
|
|
730
|
|
|
if (is_array($requestResult) && array_key_exists('log', $requestResult)) { |
731
|
|
|
$content = implode(chr(10), $requestResult['log']); |
732
|
|
|
} |
733
|
|
|
} |
734
|
|
|
return $content; |
735
|
|
|
} |
736
|
|
|
|
737
|
|
|
protected function getResStatus(array $requestContent): string |
738
|
|
|
{ |
739
|
|
|
if (empty($requestContent)) { |
740
|
|
|
return '-'; |
741
|
|
|
} |
742
|
|
|
$requestResult = unserialize($requestContent['content']); |
743
|
|
|
if (is_array($requestResult)) { |
744
|
|
|
if (empty($requestResult['errorlog'])) { |
745
|
|
|
return 'OK'; |
746
|
|
|
} else { |
747
|
|
|
return implode("\n", $requestResult['errorlog']); |
748
|
|
|
} |
749
|
|
|
} |
750
|
|
|
return 'Error: ' . substr(preg_replace('/\s+/', ' ', strip_tags($requestResult)), 0, 10000) . '...'; |
751
|
|
|
} |
752
|
|
|
|
753
|
|
|
/** |
754
|
|
|
* This method is used to show an overview about the active an the finished crawling processes |
755
|
|
|
* |
756
|
|
|
* @return string |
757
|
|
|
*/ |
758
|
|
|
protected function processOverviewAction() |
759
|
|
|
{ |
760
|
|
|
$this->view->setTemplate('ProcessOverview'); |
761
|
|
|
$this->runRefreshHooks(); |
762
|
|
|
$this->makeCrawlerProcessableChecks(); |
763
|
|
|
|
764
|
|
|
try { |
765
|
|
|
$this->handleProcessOverviewActions(); |
766
|
|
|
} catch (\Throwable $e) { |
767
|
|
|
$this->isErrorDetected = true; |
768
|
|
|
MessageUtility::addErrorMessage($e->getMessage()); |
769
|
|
|
} |
770
|
|
|
|
771
|
|
|
$processRepository = new ProcessRepository(); |
772
|
|
|
$queueRepository = new QueueRepository(); |
773
|
|
|
|
774
|
|
|
$mode = $this->pObj->MOD_SETTINGS['processListMode']; |
775
|
|
|
if ($mode == 'simple') { |
776
|
|
|
$allProcesses = $processRepository->findAllActive(); |
777
|
|
|
} else { |
778
|
|
|
$allProcesses = $processRepository->findAll(); |
779
|
|
|
} |
780
|
|
|
$isCrawlerEnabled = !$this->findCrawler()->getDisabled() && !$this->isErrorDetected; |
781
|
|
|
$currentActiveProcesses = $processRepository->countActive(); |
782
|
|
|
$maxActiveProcesses = MathUtility::forceIntegerInRange($this->extensionSettings['processLimit'], 1, 99, 1); |
783
|
|
|
$this->view->assignMultiple([ |
784
|
|
|
'pageId' => (int)$this->id, |
785
|
|
|
'refreshLink' => $this->getRefreshLink(), |
786
|
|
|
'addLink' => $this->getAddLink($currentActiveProcesses, $maxActiveProcesses, $isCrawlerEnabled), |
787
|
|
|
'modeLink' => $this->getModeLink($mode), |
788
|
|
|
'enableDisableToggle' => $this->getEnableDisableLink($isCrawlerEnabled), |
789
|
|
|
'processCollection' => $allProcesses, |
790
|
|
|
'cliPath' => $this->processManager->getCrawlerCliPath(), |
791
|
|
|
'isCrawlerEnabled' => $isCrawlerEnabled, |
792
|
|
|
'totalUnprocessedItemCount' => $queueRepository->countAllPendingItems(), |
793
|
|
|
'assignedUnprocessedItemCount' => $queueRepository->countAllAssignedPendingItems(), |
794
|
|
|
'activeProcessCount' => $currentActiveProcesses, |
795
|
|
|
'maxActiveProcessCount' => $maxActiveProcesses, |
796
|
|
|
'mode' => $mode, |
797
|
|
|
]); |
798
|
|
|
|
799
|
|
|
return $this->view->render(); |
800
|
|
|
} |
801
|
|
|
|
802
|
|
|
/** |
803
|
|
|
* Returns a tag for the refresh icon |
804
|
|
|
* |
805
|
|
|
* @return string |
806
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
807
|
|
|
*/ |
808
|
|
|
protected function getRefreshLink(): string |
809
|
|
|
{ |
810
|
|
|
return $this->getLinkButton( |
811
|
|
|
'actions-refresh', |
812
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.refresh'), |
813
|
|
|
$this->getInfoModuleUrl(['SET[\'crawleraction\']' => 'crawleraction', 'id' => $this->id]) |
814
|
|
|
); |
815
|
|
|
} |
816
|
|
|
|
817
|
|
|
/** |
818
|
|
|
* Returns a link for the panel to enable or disable the crawler |
819
|
|
|
* |
820
|
|
|
* @param bool $isCrawlerEnabled |
821
|
|
|
* @return string |
822
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
823
|
|
|
*/ |
824
|
|
|
protected function getEnableDisableLink(bool $isCrawlerEnabled): string |
825
|
|
|
{ |
826
|
|
|
if ($isCrawlerEnabled) { |
827
|
|
|
// TODO: Icon Should be bigger + Perhaps better icon |
828
|
|
|
return $this->getLinkButton( |
829
|
|
|
'tx-crawler-stop', |
830
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.disablecrawling'), |
831
|
|
|
$this->getInfoModuleUrl(['action' => 'stopCrawling']) |
832
|
|
|
); |
833
|
|
|
} else { |
834
|
|
|
// TODO: Icon Should be bigger |
835
|
|
|
return $this->getLinkButton( |
836
|
|
|
'tx-crawler-start', |
837
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.enablecrawling'), |
838
|
|
|
$this->getInfoModuleUrl(['action' => 'resumeCrawling']) |
839
|
|
|
); |
840
|
|
|
} |
841
|
|
|
} |
842
|
|
|
|
843
|
|
|
/** |
844
|
|
|
* @param string $mode |
845
|
|
|
* @return string |
846
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
847
|
|
|
*/ |
848
|
|
|
protected function getModeLink(string $mode): string |
849
|
|
|
{ |
850
|
|
|
if ($mode === 'detail') { |
851
|
|
|
return $this->getLinkButton( |
852
|
|
|
'actions-document-view', |
853
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.show.running'), |
854
|
|
|
$this->getInfoModuleUrl(['SET[\'processListMode\']' => 'simple']) |
855
|
|
|
); |
856
|
|
|
} elseif ($mode === 'simple') { |
857
|
|
|
return $this->getLinkButton( |
858
|
|
|
'actions-document-view', |
859
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.show.all'), |
860
|
|
|
$this->getInfoModuleUrl(['SET[\'processListMode\']' => 'detail']) |
861
|
|
|
); |
862
|
|
|
} |
863
|
|
|
return ''; |
864
|
|
|
} |
865
|
|
|
|
866
|
|
|
/** |
867
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
868
|
|
|
*/ |
869
|
|
|
protected function getAddLink(int $currentActiveProcesses, int $maxActiveProcesses, bool $isCrawlerEnabled): string |
870
|
|
|
{ |
871
|
|
|
if (!$isCrawlerEnabled) { |
872
|
|
|
return ''; |
873
|
|
|
} |
874
|
|
|
if ($currentActiveProcesses >= $maxActiveProcesses) { |
875
|
|
|
return ''; |
876
|
|
|
} |
877
|
|
|
|
878
|
|
|
return $this->getLinkButton( |
879
|
|
|
'actions-add', |
880
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.process.add'), |
881
|
|
|
$this->getInfoModuleUrl(['action' => 'addProcess']) |
882
|
|
|
); |
883
|
|
|
} |
884
|
|
|
|
885
|
|
|
/** |
886
|
|
|
* Verify that the crawler is executable. |
887
|
|
|
*/ |
888
|
|
|
protected function makeCrawlerProcessableChecks(): void |
889
|
|
|
{ |
890
|
|
|
if (!$this->isPhpForkAvailable()) { |
891
|
|
|
$this->isErrorDetected = true; |
892
|
|
|
MessageUtility::addErrorMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:message.noPhpForkAvailable')); |
893
|
|
|
} |
894
|
|
|
|
895
|
|
|
$exitCode = 0; |
896
|
|
|
$out = []; |
897
|
|
|
CommandUtility::exec( |
898
|
|
|
PhpBinaryUtility::getPhpBinary() . ' -v', |
899
|
|
|
$out, |
900
|
|
|
$exitCode |
901
|
|
|
); |
902
|
|
|
if ($exitCode > 0) { |
903
|
|
|
$this->isErrorDetected = true; |
904
|
|
|
MessageUtility::addErrorMessage(sprintf($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:message.phpBinaryNotFound'), htmlspecialchars($this->extensionSettings['phpPath']))); |
905
|
|
|
} |
906
|
|
|
} |
907
|
|
|
|
908
|
|
|
/** |
909
|
|
|
* Indicate that the required PHP method "popen" is |
910
|
|
|
* available in the system. |
911
|
|
|
*/ |
912
|
|
|
protected function isPhpForkAvailable(): bool |
913
|
|
|
{ |
914
|
|
|
return function_exists('popen'); |
915
|
|
|
} |
916
|
|
|
|
917
|
|
|
/** |
918
|
|
|
* Method to handle incomming actions of the process overview |
919
|
|
|
* |
920
|
|
|
* @throws \Exception |
921
|
|
|
*/ |
922
|
|
|
protected function handleProcessOverviewActions(): void |
923
|
|
|
{ |
924
|
|
|
$crawler = $this->findCrawler(); |
925
|
|
|
|
926
|
|
|
switch (GeneralUtility::_GP('action')) { |
927
|
|
|
case 'stopCrawling': |
928
|
|
|
//set the cli status to disable (all processes will be terminated) |
929
|
|
|
$crawler->setDisabled(true); |
930
|
|
|
break; |
931
|
|
|
case 'resumeCrawling': |
932
|
|
|
//set the cli status to end (all processes will be terminated) |
933
|
|
|
$crawler->setDisabled(false); |
934
|
|
|
break; |
935
|
|
|
case 'addProcess': |
936
|
|
|
if ($this->processManager->startProcess() === false) { |
|
|
|
|
937
|
|
|
throw new \Exception($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.newprocesserror')); |
938
|
|
|
} |
939
|
|
|
MessageUtility::addNoticeMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.newprocess')); |
940
|
|
|
break; |
941
|
|
|
} |
942
|
|
|
} |
943
|
|
|
|
944
|
|
|
/** |
945
|
|
|
* Returns the singleton instance of the crawler. |
946
|
|
|
*/ |
947
|
|
|
protected function findCrawler(): CrawlerController |
948
|
|
|
{ |
949
|
|
|
if (!$this->crawlerController instanceof CrawlerController) { |
|
|
|
|
950
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
951
|
|
|
} |
952
|
|
|
return $this->crawlerController; |
953
|
|
|
} |
954
|
|
|
|
955
|
|
|
/***************************** |
956
|
|
|
* |
957
|
|
|
* General Helper Functions |
958
|
|
|
* |
959
|
|
|
*****************************/ |
960
|
|
|
|
961
|
|
|
/** |
962
|
|
|
* Create selector box |
963
|
|
|
* |
964
|
|
|
* @param array $optArray Options key(value) => label pairs |
965
|
|
|
* @param string $name Selector box name |
966
|
|
|
* @param string|array $value Selector box value (array for multiple...) |
967
|
|
|
* @param boolean $multiple If set, will draw multiple box. |
968
|
|
|
* |
969
|
|
|
* @return string HTML select element |
970
|
|
|
*/ |
971
|
|
|
protected function selectorBox($optArray, $name, $value, bool $multiple): string |
972
|
|
|
{ |
973
|
|
|
if (!is_string($value) || !is_array($value)) { |
|
|
|
|
974
|
|
|
$value = ''; |
975
|
|
|
} |
976
|
|
|
|
977
|
|
|
$options = []; |
978
|
|
|
foreach ($optArray as $key => $val) { |
979
|
|
|
$selected = (!$multiple && !strcmp($value, (string)$key)) || ($multiple && in_array($key, (array)$value)); |
|
|
|
|
980
|
|
|
$options[] = ' |
981
|
|
|
<option value="' . $key . '" ' . ($selected ? ' selected="selected"' : '') . '>' . htmlspecialchars($val) . '</option>'; |
982
|
|
|
} |
983
|
|
|
|
984
|
|
|
return '<select class="form-control" name="' . htmlspecialchars($name . ($multiple ? '[]' : '')) . '"' . ($multiple ? ' multiple' : '') . '>' . implode('', $options) . '</select>'; |
985
|
|
|
} |
986
|
|
|
|
987
|
|
|
/** |
988
|
|
|
* Activate hooks |
989
|
|
|
*/ |
990
|
|
|
protected function runRefreshHooks(): void |
991
|
|
|
{ |
992
|
|
|
$crawlerLib = GeneralUtility::makeInstance(CrawlerController::class); |
993
|
|
|
foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['refresh_hooks'] ?? [] as $objRef) { |
994
|
|
|
/** @var CrawlerHookInterface $hookObj */ |
995
|
|
|
$hookObj = GeneralUtility::makeInstance($objRef); |
996
|
|
|
if (is_object($hookObj)) { |
997
|
|
|
$hookObj->crawler_init($crawlerLib); |
998
|
|
|
} |
999
|
|
|
} |
1000
|
|
|
} |
1001
|
|
|
|
1002
|
|
|
protected function initializeView(): void |
1003
|
|
|
{ |
1004
|
|
|
$view = GeneralUtility::makeInstance(StandaloneView::class); |
1005
|
|
|
$view->setLayoutRootPaths(['EXT:crawler/Resources/Private/Layouts']); |
1006
|
|
|
$view->setPartialRootPaths(['EXT:crawler/Resources/Private/Partials']); |
1007
|
|
|
$view->setTemplateRootPaths(['EXT:crawler/Resources/Private/Templates/Backend']); |
1008
|
|
|
$view->getRequest()->setControllerExtensionName('Crawler'); |
1009
|
|
|
$this->view = $view; |
1010
|
|
|
} |
1011
|
|
|
|
1012
|
|
|
protected function getLanguageService(): LanguageService |
1013
|
|
|
{ |
1014
|
|
|
return $GLOBALS['LANG']; |
1015
|
|
|
} |
1016
|
|
|
|
1017
|
|
|
/** |
1018
|
|
|
* @param string $iconIdentifier |
1019
|
|
|
* @param string $title |
1020
|
|
|
* @param UriInterface $href |
1021
|
|
|
* @return string |
1022
|
|
|
*/ |
1023
|
|
|
protected function getLinkButton(string $iconIdentifier, string $title, UriInterface $href): string |
1024
|
|
|
{ |
1025
|
|
|
$moduleTemplate = GeneralUtility::makeInstance(ModuleTemplate::class); |
1026
|
|
|
$buttonBar = $moduleTemplate->getDocHeaderComponent()->getButtonBar(); |
1027
|
|
|
return (string)$buttonBar->makeLinkButton() |
1028
|
|
|
->setHref((string)$href) |
1029
|
|
|
->setIcon($this->iconFactory->getIcon($iconIdentifier, Icon::SIZE_SMALL)) |
1030
|
|
|
->setTitle($title) |
1031
|
|
|
->setShowLabelText(true); |
1032
|
|
|
} |
1033
|
|
|
|
1034
|
|
|
/** |
1035
|
|
|
* Returns the URL to the current module, including $_GET['id']. |
1036
|
|
|
* |
1037
|
|
|
* @param array $uriParameters optional parameters to add to the URL |
1038
|
|
|
* @return Uri |
1039
|
|
|
* |
1040
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
1041
|
|
|
*/ |
1042
|
|
|
protected function getInfoModuleUrl(array $uriParameters = []): Uri |
1043
|
|
|
{ |
1044
|
|
|
if (GeneralUtility::_GP('id')) { |
1045
|
|
|
$uriParameters = array_merge($uriParameters, [ |
1046
|
|
|
'id' => GeneralUtility::_GP('id'), |
1047
|
|
|
]); |
1048
|
|
|
} |
1049
|
|
|
/** @var UriBuilder $uriBuilder */ |
1050
|
|
|
$uriBuilder = GeneralUtility::makeInstance(UriBuilder::class); |
1051
|
|
|
return $uriBuilder->buildUriFromRoute('web_info', $uriParameters); |
1052
|
|
|
} |
1053
|
|
|
} |
1054
|
|
|
|
In general, usage of exit should be done with care and only when running in a scripting context like a CLI script.