1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace AOE\Crawler\Backend; |
6
|
|
|
|
7
|
|
|
/*************************************************************** |
8
|
|
|
* Copyright notice |
9
|
|
|
* |
10
|
|
|
* (c) 2020 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
|
|
|
/** |
83
|
|
|
* @var array |
84
|
|
|
*/ |
85
|
|
|
protected $duplicateTrack = []; |
86
|
|
|
|
87
|
|
|
/** |
88
|
|
|
* @var bool |
89
|
|
|
*/ |
90
|
|
|
protected $submitCrawlUrls = false; |
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* @var bool |
94
|
|
|
*/ |
95
|
|
|
protected $downloadCrawlUrls = false; |
96
|
|
|
|
97
|
|
|
/** |
98
|
|
|
* @var int |
99
|
|
|
*/ |
100
|
|
|
protected $scheduledTime = 0; |
101
|
|
|
|
102
|
|
|
/** |
103
|
|
|
* @var int |
104
|
|
|
*/ |
105
|
|
|
protected $reqMinute = 1000; |
106
|
|
|
|
107
|
|
|
/** |
108
|
|
|
* @var array holds the selection of configuration from the configuration selector box |
109
|
|
|
*/ |
110
|
|
|
protected $incomingConfigurationSelection = []; |
111
|
|
|
|
112
|
|
|
/** |
113
|
|
|
* @var CrawlerController |
114
|
|
|
*/ |
115
|
|
|
protected $crawlerController; |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* @var array |
119
|
|
|
*/ |
120
|
|
|
protected $CSVaccu = []; |
121
|
|
|
|
122
|
|
|
/** |
123
|
|
|
* If true the user requested a CSV export of the queue |
124
|
|
|
* |
125
|
|
|
* @var boolean |
126
|
|
|
*/ |
127
|
|
|
protected $CSVExport = false; |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* @var array |
131
|
|
|
*/ |
132
|
|
|
protected $downloadUrls = []; |
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* Holds the configuration from ext_conf_template loaded by getExtensionConfiguration() |
136
|
|
|
* |
137
|
|
|
* @var array |
138
|
|
|
*/ |
139
|
|
|
protected $extensionSettings = []; |
140
|
|
|
|
141
|
|
|
/** |
142
|
|
|
* Indicate that an flash message with an error is present. |
143
|
|
|
* |
144
|
|
|
* @var boolean |
145
|
|
|
*/ |
146
|
|
|
protected $isErrorDetected = false; |
147
|
|
|
|
148
|
|
|
/** |
149
|
|
|
* @var ProcessService |
150
|
|
|
*/ |
151
|
|
|
protected $processManager; |
152
|
|
|
|
153
|
|
|
/** |
154
|
|
|
* @var QueryBuilder |
155
|
|
|
*/ |
156
|
|
|
protected $queryBuilder; |
157
|
|
|
|
158
|
|
|
/** |
159
|
|
|
* @var QueueRepository |
160
|
|
|
*/ |
161
|
|
|
protected $queueRepository; |
162
|
|
|
|
163
|
|
|
/** |
164
|
|
|
* @var StandaloneView |
165
|
|
|
*/ |
166
|
|
|
protected $view; |
167
|
|
|
|
168
|
|
|
/** |
169
|
|
|
* @var IconFactory |
170
|
|
|
*/ |
171
|
|
|
protected $iconFactory; |
172
|
|
|
|
173
|
|
|
public function __construct() |
174
|
|
|
{ |
175
|
|
|
$objectManger = GeneralUtility::makeInstance(ObjectManager::class); |
176
|
|
|
$this->processManager = $objectManger->get(ProcessService::class); |
177
|
|
|
$this->queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable('tx_crawler_queue'); |
178
|
|
|
$this->queueRepository = $objectManger->get(QueueRepository::class); |
179
|
|
|
$this->initializeView(); |
180
|
|
|
$this->extensionSettings = GeneralUtility::makeInstance(ExtensionConfigurationProvider::class)->getExtensionConfiguration(); |
181
|
|
|
$this->iconFactory = GeneralUtility::makeInstance(IconFactory::class); |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* Called by the InfoModuleController |
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:core/Resources/Private/Language/locallang_core.xlf:labels.depth_0'), |
205
|
|
|
1 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_1'), |
206
|
|
|
2 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_2'), |
207
|
|
|
3 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_3'), |
208
|
|
|
4 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_4'), |
209
|
|
|
99 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_infi'), |
210
|
|
|
], |
211
|
|
|
'crawlaction' => [ |
212
|
|
|
'start' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.start'), |
213
|
|
|
'log' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.log'), |
214
|
|
|
'multiprocess' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf: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.xlf:labels.all'), |
221
|
|
|
'pending' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.pending'), |
222
|
|
|
'finished' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.finished'), |
223
|
|
|
], |
224
|
|
|
'itemsPerPage' => [ |
225
|
|
|
'5' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.itemsPerPage.5'), |
226
|
|
|
'10' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.itemsPerPage.10'), |
227
|
|
|
'50' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.itemsPerPage.50'), |
228
|
|
|
'0' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.itemsPerPage.0'), |
229
|
|
|
], |
230
|
|
|
]; |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
public function main(): string |
234
|
|
|
{ |
235
|
|
|
if (empty($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
|
|
|
$quiPart = GeneralUtility::_GP('qid_details') ? '&qid_details=' . (int) GeneralUtility::_GP('qid_details') : ''; |
259
|
|
|
$setId = (int) GeneralUtility::_GP('setID'); |
260
|
|
|
|
261
|
|
|
// Additional menus for the log type: |
262
|
|
|
$theOutput .= $this->getDepthDropDownHtml(); |
263
|
|
|
$theOutput .= $this->showLogAction($setId, $quiPart); |
264
|
|
|
break; |
265
|
|
|
case 'multiprocess': |
266
|
|
|
$theOutput .= $this->processOverviewAction(); |
267
|
|
|
break; |
268
|
|
|
} |
269
|
|
|
|
270
|
|
|
return $theOutput; |
271
|
|
|
} |
272
|
|
|
|
273
|
|
|
/******************************* |
274
|
|
|
* |
275
|
|
|
* Generate URLs for crawling: |
276
|
|
|
* |
277
|
|
|
******************************/ |
278
|
|
|
|
279
|
|
|
/** |
280
|
|
|
* Show a list of URLs to be crawled for each page |
281
|
|
|
*/ |
282
|
|
|
protected function showCrawlerInformationAction(): string |
283
|
|
|
{ |
284
|
|
|
$this->view->setTemplate('ShowCrawlerInformation'); |
285
|
|
|
if (empty($this->id)) { |
286
|
|
|
$this->isErrorDetected = true; |
287
|
|
|
MessageUtility::addErrorMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.noPageSelected')); |
288
|
|
|
} else { |
289
|
|
|
$crawlerParameter = GeneralUtility::_GP('_crawl'); |
290
|
|
|
$downloadParameter = GeneralUtility::_GP('_download'); |
291
|
|
|
|
292
|
|
|
$this->duplicateTrack = []; |
293
|
|
|
$this->submitCrawlUrls = isset($crawlerParameter); |
294
|
|
|
$this->downloadCrawlUrls = isset($downloadParameter); |
295
|
|
|
$this->makeCrawlerProcessableChecks(); |
296
|
|
|
|
297
|
|
|
switch ((string) GeneralUtility::_GP('tstamp')) { |
298
|
|
|
case 'midnight': |
299
|
|
|
$this->scheduledTime = mktime(0, 0, 0); |
300
|
|
|
break; |
301
|
|
|
case '04:00': |
302
|
|
|
$this->scheduledTime = mktime(0, 0, 0) + 4 * 3600; |
303
|
|
|
break; |
304
|
|
|
case 'now': |
305
|
|
|
default: |
306
|
|
|
$this->scheduledTime = time(); |
307
|
|
|
break; |
308
|
|
|
} |
309
|
|
|
|
310
|
|
|
$this->incomingConfigurationSelection = GeneralUtility::_GP('configurationSelection'); |
311
|
|
|
$this->incomingConfigurationSelection = is_array($this->incomingConfigurationSelection) ? $this->incomingConfigurationSelection : []; |
312
|
|
|
|
313
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
314
|
|
|
$this->crawlerController->setAccessMode('gui'); |
315
|
|
|
$this->crawlerController->setID = GeneralUtility::md5int(microtime()); |
316
|
|
|
|
317
|
|
|
$code = ''; |
318
|
|
|
$noConfigurationSelected = empty($this->incomingConfigurationSelection) |
319
|
|
|
|| (count($this->incomingConfigurationSelection) === 1 && empty($this->incomingConfigurationSelection[0])); |
320
|
|
|
if ($noConfigurationSelected) { |
321
|
|
|
MessageUtility::addWarningMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.noConfigSelected')); |
322
|
|
|
} else { |
323
|
|
|
if ($this->submitCrawlUrls) { |
324
|
|
|
$reason = new Reason(); |
325
|
|
|
$reason->setReason(Reason::REASON_GUI_SUBMIT); |
326
|
|
|
$reason->setDetailText('The user ' . $GLOBALS['BE_USER']->user['username'] . ' added pages to the crawler queue manually'); |
327
|
|
|
|
328
|
|
|
EventDispatcher::getInstance()->post( |
329
|
|
|
'invokeQueueChange', |
330
|
|
|
strval($this->findCrawler()->setID), |
331
|
|
|
['reason' => $reason] |
332
|
|
|
); |
333
|
|
|
} |
334
|
|
|
|
335
|
|
|
$code = $this->crawlerController->getPageTreeAndUrls( |
336
|
|
|
$this->id, |
337
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
338
|
|
|
$this->scheduledTime, |
339
|
|
|
$this->reqMinute, |
340
|
|
|
$this->submitCrawlUrls, |
341
|
|
|
$this->downloadCrawlUrls, |
342
|
|
|
[], // Do not filter any processing instructions |
343
|
|
|
$this->incomingConfigurationSelection |
344
|
|
|
); |
345
|
|
|
} |
346
|
|
|
|
347
|
|
|
$this->downloadUrls = $this->crawlerController->downloadUrls; |
348
|
|
|
$this->duplicateTrack = $this->crawlerController->duplicateTrack; |
349
|
|
|
|
350
|
|
|
$this->view->assign('noConfigurationSelected', $noConfigurationSelected); |
351
|
|
|
$this->view->assign('submitCrawlUrls', $this->submitCrawlUrls); |
352
|
|
|
$this->view->assign('amountOfUrls', count(array_keys($this->duplicateTrack))); |
353
|
|
|
$this->view->assign('selectors', $this->generateConfigurationSelectors()); |
354
|
|
|
$this->view->assign('code', $code); |
355
|
|
|
$this->view->assign('displayActions', 0); |
356
|
|
|
|
357
|
|
|
// Download Urls to crawl: |
358
|
|
|
if ($this->downloadCrawlUrls) { |
359
|
|
|
// Creating output header: |
360
|
|
|
header('Content-Type: application/octet-stream'); |
361
|
|
|
header('Content-Disposition: attachment; filename=CrawlerUrls.txt'); |
362
|
|
|
|
363
|
|
|
// Printing the content of the CSV lines: |
364
|
|
|
echo implode(chr(13) . chr(10), $this->downloadUrls); |
365
|
|
|
exit; |
|
|
|
|
366
|
|
|
} |
367
|
|
|
} |
368
|
|
|
return $this->view->render(); |
369
|
|
|
} |
370
|
|
|
|
371
|
|
|
/** |
372
|
|
|
* Generates the configuration selectors for compiling URLs: |
373
|
|
|
*/ |
374
|
|
|
protected function generateConfigurationSelectors(): array |
375
|
|
|
{ |
376
|
|
|
$selectors = []; |
377
|
|
|
$selectors['depth'] = $this->selectorBox( |
378
|
|
|
[ |
379
|
|
|
0 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_0'), |
380
|
|
|
1 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_1'), |
381
|
|
|
2 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_2'), |
382
|
|
|
3 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_3'), |
383
|
|
|
4 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_4'), |
384
|
|
|
99 => $this->getLanguageService()->sL('LLL:EXT:core/Resources/Private/Language/locallang_core.xlf:labels.depth_infi'), |
385
|
|
|
], |
386
|
|
|
'SET[depth]', |
387
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
388
|
|
|
false |
389
|
|
|
); |
390
|
|
|
|
391
|
|
|
// Configurations |
392
|
|
|
$availableConfigurations = $this->crawlerController->getConfigurationsForBranch((int) $this->id, (int) $this->pObj->MOD_SETTINGS['depth'] ?: 0); |
393
|
|
|
$selectors['configurations'] = $this->selectorBox( |
394
|
|
|
empty($availableConfigurations) ? [] : array_combine($availableConfigurations, $availableConfigurations), |
|
|
|
|
395
|
|
|
'configurationSelection', |
396
|
|
|
$this->incomingConfigurationSelection, |
397
|
|
|
true |
398
|
|
|
); |
399
|
|
|
|
400
|
|
|
// Scheduled time: |
401
|
|
|
$selectors['scheduled'] = $this->selectorBox( |
402
|
|
|
[ |
403
|
|
|
'now' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.time.now'), |
404
|
|
|
'midnight' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.time.midnight'), |
405
|
|
|
'04:00' => $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.time.4am'), |
406
|
|
|
], |
407
|
|
|
'tstamp', |
408
|
|
|
GeneralUtility::_POST('tstamp'), |
409
|
|
|
false |
410
|
|
|
); |
411
|
|
|
|
412
|
|
|
return $selectors; |
413
|
|
|
} |
414
|
|
|
|
415
|
|
|
/******************************* |
416
|
|
|
* |
417
|
|
|
* Shows log of indexed URLs |
418
|
|
|
* |
419
|
|
|
******************************/ |
420
|
|
|
|
421
|
|
|
/** |
422
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
423
|
|
|
*/ |
424
|
|
|
protected function showLogAction(int $setId, string $quiPath): string |
425
|
|
|
{ |
426
|
|
|
$this->view->setTemplate('ShowLog'); |
427
|
|
|
if (empty($this->id)) { |
428
|
|
|
$this->isErrorDetected = true; |
429
|
|
|
MessageUtility::addErrorMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.noPageSelected')); |
430
|
|
|
} else { |
431
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
432
|
|
|
$this->crawlerController->setAccessMode('gui'); |
433
|
|
|
$this->crawlerController->setID = GeneralUtility::md5int(microtime()); |
434
|
|
|
|
435
|
|
|
$csvExport = GeneralUtility::_POST('_csv'); |
436
|
|
|
$this->CSVExport = isset($csvExport); |
437
|
|
|
|
438
|
|
|
// Read URL: |
439
|
|
|
if (GeneralUtility::_GP('qid_read')) { |
440
|
|
|
$this->crawlerController->readUrl((int) GeneralUtility::_GP('qid_read'), true); |
441
|
|
|
} |
442
|
|
|
|
443
|
|
|
// Look for set ID sent - if it is, we will display contents of that set: |
444
|
|
|
$showSetId = (int) GeneralUtility::_GP('setID'); |
445
|
|
|
|
446
|
|
|
$queueId = GeneralUtility::_GP('qid_details'); |
447
|
|
|
$this->view->assign('queueId', $queueId); |
448
|
|
|
$this->view->assign('setId', $showSetId); |
449
|
|
|
// Show details: |
450
|
|
|
if ($queueId) { |
451
|
|
|
// Get entry record: |
452
|
|
|
$q_entry = $this->queryBuilder |
453
|
|
|
->from('tx_crawler_queue') |
454
|
|
|
->select('*') |
455
|
|
|
->where( |
456
|
|
|
$this->queryBuilder->expr()->eq('qid', $this->queryBuilder->createNamedParameter($queueId)) |
457
|
|
|
) |
458
|
|
|
->execute() |
459
|
|
|
->fetch(); |
460
|
|
|
|
461
|
|
|
// Explode values |
462
|
|
|
$q_entry['parameters'] = unserialize($q_entry['parameters']); |
463
|
|
|
$q_entry['result_data'] = unserialize($q_entry['result_data']); |
464
|
|
|
$resStatus = $this->getResStatus($q_entry['result_data']); |
465
|
|
|
if (is_array($q_entry['result_data'])) { |
466
|
|
|
$q_entry['result_data']['content'] = unserialize($q_entry['result_data']['content']); |
467
|
|
|
if (! $this->pObj->MOD_SETTINGS['log_resultLog']) { |
468
|
|
|
unset($q_entry['result_data']['content']['log']); |
469
|
|
|
} |
470
|
|
|
} |
471
|
|
|
|
472
|
|
|
$this->view->assign('queueStatus', $resStatus); |
473
|
|
|
$this->view->assign('queueDetails', DebugUtility::viewArray($q_entry)); |
474
|
|
|
} else { |
475
|
|
|
// Show list |
476
|
|
|
// Drawing tree: |
477
|
|
|
$tree = GeneralUtility::makeInstance(PageTreeView::class); |
478
|
|
|
$perms_clause = $GLOBALS['BE_USER']->getPagePermsClause(1); |
479
|
|
|
$tree->init('AND ' . $perms_clause); |
480
|
|
|
|
481
|
|
|
// Set root row: |
482
|
|
|
$pageinfo = BackendUtility::readPageAccess( |
483
|
|
|
$this->id, |
484
|
|
|
$perms_clause |
485
|
|
|
); |
486
|
|
|
$HTML = $this->iconFactory->getIconForRecord('pages', $pageinfo, Icon::SIZE_SMALL)->render(); |
487
|
|
|
$tree->tree[] = [ |
488
|
|
|
'row' => $pageinfo, |
489
|
|
|
'HTML' => $HTML, |
490
|
|
|
]; |
491
|
|
|
|
492
|
|
|
// Get branch beneath: |
493
|
|
|
if ($this->pObj->MOD_SETTINGS['depth']) { |
494
|
|
|
$tree->getTree($this->id, $this->pObj->MOD_SETTINGS['depth']); |
495
|
|
|
} |
496
|
|
|
|
497
|
|
|
// If Flush button is pressed, flush tables instead of selecting entries: |
498
|
|
|
if (GeneralUtility::_POST('_flush')) { |
499
|
|
|
$doFlush = true; |
500
|
|
|
$doFullFlush = false; |
501
|
|
|
} elseif (GeneralUtility::_POST('_flush_all')) { |
502
|
|
|
$doFlush = true; |
503
|
|
|
$doFullFlush = true; |
504
|
|
|
} else { |
505
|
|
|
$doFlush = false; |
506
|
|
|
$doFullFlush = false; |
507
|
|
|
} |
508
|
|
|
$itemsPerPage = (int) $this->pObj->MOD_SETTINGS['itemsPerPage']; |
509
|
|
|
// Traverse page tree: |
510
|
|
|
$code = ''; |
511
|
|
|
$count = 0; |
512
|
|
|
foreach ($tree->tree as $data) { |
513
|
|
|
// Get result: |
514
|
|
|
$logEntriesOfPage = $this->crawlerController->getLogEntriesForPageId( |
515
|
|
|
(int) $data['row']['uid'], |
516
|
|
|
$this->pObj->MOD_SETTINGS['log_display'], |
517
|
|
|
$doFlush, |
518
|
|
|
$doFullFlush, |
519
|
|
|
$itemsPerPage |
520
|
|
|
); |
521
|
|
|
|
522
|
|
|
$code .= $this->drawLog_addRows( |
523
|
|
|
$logEntriesOfPage, |
524
|
|
|
$data['HTML'] . BackendUtility::getRecordTitle('pages', $data['row'], true) |
525
|
|
|
); |
526
|
|
|
if (++$count === 1000) { |
527
|
|
|
break; |
528
|
|
|
} |
529
|
|
|
} |
530
|
|
|
$this->view->assign('code', $code); |
531
|
|
|
} |
532
|
|
|
|
533
|
|
|
if ($this->CSVExport) { |
534
|
|
|
$this->outputCsvFile(); |
535
|
|
|
} |
536
|
|
|
} |
537
|
|
|
$this->view->assign('showResultLog', (bool) $this->pObj->MOD_SETTINGS['log_resultLog']); |
538
|
|
|
$this->view->assign('showFeVars', (bool) $this->pObj->MOD_SETTINGS['log_feVars']); |
539
|
|
|
$this->view->assign('displayActions', 1); |
540
|
|
|
$this->view->assign('displayLogFilterHtml', $this->getDisplayLogFilterHtml($setId)); |
541
|
|
|
$this->view->assign('itemPerPageHtml', $this->getItemsPerPageDropDownHtml()); |
542
|
|
|
$this->view->assign('showResultLogHtml', $this->getShowResultLogCheckBoxHtml($setId, $quiPath)); |
543
|
|
|
$this->view->assign('showFeVarsHtml', $this->getShowFeVarsCheckBoxHtml($setId, $quiPath)); |
544
|
|
|
return $this->view->render(); |
545
|
|
|
} |
546
|
|
|
|
547
|
|
|
/** |
548
|
|
|
* Outputs the CSV file and sets the correct headers |
549
|
|
|
*/ |
550
|
|
|
protected function outputCsvFile(): void |
551
|
|
|
{ |
552
|
|
|
if (! count($this->CSVaccu)) { |
553
|
|
|
MessageUtility::addWarningMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:message.canNotExportEmptyQueueToCsvText')); |
554
|
|
|
return; |
555
|
|
|
} |
556
|
|
|
$csvLines = []; |
557
|
|
|
|
558
|
|
|
// Field names: |
559
|
|
|
reset($this->CSVaccu); |
560
|
|
|
$fieldNames = array_keys(current($this->CSVaccu)); |
561
|
|
|
$csvLines[] = CsvUtility::csvValues($fieldNames); |
562
|
|
|
|
563
|
|
|
// Data: |
564
|
|
|
foreach ($this->CSVaccu as $row) { |
565
|
|
|
$csvLines[] = CsvUtility::csvValues($row); |
566
|
|
|
} |
567
|
|
|
|
568
|
|
|
// Creating output header: |
569
|
|
|
header('Content-Type: application/octet-stream'); |
570
|
|
|
header('Content-Disposition: attachment; filename=CrawlerLog.csv'); |
571
|
|
|
|
572
|
|
|
// Printing the content of the CSV lines: |
573
|
|
|
echo implode(chr(13) . chr(10), $csvLines); |
574
|
|
|
exit; |
|
|
|
|
575
|
|
|
} |
576
|
|
|
|
577
|
|
|
/** |
578
|
|
|
* Create the rows for display of the page tree |
579
|
|
|
* For each page a number of rows are shown displaying GET variable configuration |
580
|
|
|
* |
581
|
|
|
* @param array $logEntriesOfPage Log items of one page |
582
|
|
|
* @param string $titleString Title string |
583
|
|
|
* @return string HTML <tr> content (one or more) |
584
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
585
|
|
|
*/ |
586
|
|
|
protected function drawLog_addRows(array $logEntriesOfPage, string $titleString): string |
587
|
|
|
{ |
588
|
|
|
$colSpan = 9 |
589
|
|
|
+ ($this->pObj->MOD_SETTINGS['log_resultLog'] ? -1 : 0) |
590
|
|
|
+ ($this->pObj->MOD_SETTINGS['log_feVars'] ? 3 : 0); |
591
|
|
|
|
592
|
|
|
if (! empty($logEntriesOfPage)) { |
593
|
|
|
$setId = (int) GeneralUtility::_GP('setID'); |
594
|
|
|
$refreshIcon = $this->iconFactory->getIcon('actions-system-refresh', Icon::SIZE_SMALL); |
595
|
|
|
// Traverse parameter combinations: |
596
|
|
|
$c = 0; |
597
|
|
|
$content = ''; |
598
|
|
|
foreach ($logEntriesOfPage as $vv) { |
599
|
|
|
// Title column: |
600
|
|
|
if (! $c) { |
601
|
|
|
$titleClm = '<td rowspan="' . count($logEntriesOfPage) . '">' . $titleString . '</td>'; |
602
|
|
|
} else { |
603
|
|
|
$titleClm = ''; |
604
|
|
|
} |
605
|
|
|
|
606
|
|
|
// Result: |
607
|
|
|
$resLog = $this->getResultLog($vv); |
608
|
|
|
|
609
|
|
|
$resultData = $vv['result_data'] ? unserialize($vv['result_data']) : []; |
610
|
|
|
$resStatus = $this->getResStatus($resultData); |
611
|
|
|
|
612
|
|
|
// Compile row: |
613
|
|
|
$parameters = unserialize($vv['parameters']); |
614
|
|
|
|
615
|
|
|
// Put data into array: |
616
|
|
|
$rowData = []; |
617
|
|
|
if ($this->pObj->MOD_SETTINGS['log_resultLog']) { |
618
|
|
|
$rowData['result_log'] = $resLog; |
619
|
|
|
} else { |
620
|
|
|
$rowData['scheduled'] = ($vv['scheduled'] > 0) ? BackendUtility::datetime($vv['scheduled']) : ' ' . $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.immediate'); |
621
|
|
|
$rowData['exec_time'] = $vv['exec_time'] ? BackendUtility::datetime($vv['exec_time']) : '-'; |
622
|
|
|
} |
623
|
|
|
$rowData['result_status'] = GeneralUtility::fixed_lgd_cs($resStatus, 50); |
624
|
|
|
$url = htmlspecialchars($parameters['url'] ?? $parameters['alturl']); |
625
|
|
|
$rowData['url'] = '<a href="' . $url . '" target="_newWIndow">' . $url . '</a>'; |
626
|
|
|
$rowData['feUserGroupList'] = $parameters['feUserGroupList'] ?: ''; |
627
|
|
|
$rowData['procInstructions'] = is_array($parameters['procInstructions']) ? implode('; ', $parameters['procInstructions']) : ''; |
628
|
|
|
$rowData['set_id'] = (string) $vv['set_id']; |
629
|
|
|
|
630
|
|
|
if ($this->pObj->MOD_SETTINGS['log_feVars']) { |
631
|
|
|
$resFeVars = $this->getResFeVars($resultData ?: []); |
632
|
|
|
$rowData['tsfe_id'] = $resFeVars['id'] ?: ''; |
633
|
|
|
$rowData['tsfe_gr_list'] = $resFeVars['gr_list'] ?: ''; |
634
|
|
|
$rowData['tsfe_no_cache'] = $resFeVars['no_cache'] ?: ''; |
635
|
|
|
} |
636
|
|
|
|
637
|
|
|
// Put rows together: |
638
|
|
|
$content .= ' |
639
|
|
|
<tr> |
640
|
|
|
' . $titleClm . ' |
641
|
|
|
<td><a href="' . $this->getInfoModuleUrl(['qid_details' => $vv['qid'], 'setID' => $setId]) . '">' . htmlspecialchars((string) $vv['qid']) . '</a></td> |
642
|
|
|
<td><a href="' . $this->getInfoModuleUrl(['qid_read' => $vv['qid'], 'setID' => $setId]) . '">' . $refreshIcon . '</a></td>'; |
643
|
|
|
foreach ($rowData as $fKey => $value) { |
644
|
|
|
if ($fKey === 'url') { |
645
|
|
|
$content .= '<td>' . $value . '</td>'; |
646
|
|
|
} else { |
647
|
|
|
$content .= '<td>' . nl2br(htmlspecialchars(strval($value))) . '</td>'; |
648
|
|
|
} |
649
|
|
|
} |
650
|
|
|
$content .= '</tr>'; |
651
|
|
|
$c++; |
652
|
|
|
|
653
|
|
|
if ($this->CSVExport) { |
654
|
|
|
// Only for CSV (adding qid and scheduled/exec_time if needed): |
655
|
|
|
$rowData['result_log'] = implode('// ', explode(chr(10), $resLog)); |
656
|
|
|
$rowData['qid'] = $vv['qid']; |
657
|
|
|
$rowData['scheduled'] = BackendUtility::datetime($vv['scheduled']); |
658
|
|
|
$rowData['exec_time'] = $vv['exec_time'] ? BackendUtility::datetime($vv['exec_time']) : '-'; |
659
|
|
|
$this->CSVaccu[] = $rowData; |
660
|
|
|
} |
661
|
|
|
} |
662
|
|
|
} else { |
663
|
|
|
// Compile row: |
664
|
|
|
$content = ' |
665
|
|
|
<tr> |
666
|
|
|
<td>' . $titleString . '</td> |
667
|
|
|
<td colspan="' . $colSpan . '"><em>' . $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.noentries') . '</em></td> |
668
|
|
|
</tr>'; |
669
|
|
|
} |
670
|
|
|
|
671
|
|
|
return $content; |
672
|
|
|
} |
673
|
|
|
|
674
|
|
|
/** |
675
|
|
|
* Find Fe vars |
676
|
|
|
*/ |
677
|
|
|
protected function getResFeVars(array $resultData): array |
678
|
|
|
{ |
679
|
|
|
if (empty($resultData)) { |
680
|
|
|
return []; |
681
|
|
|
} |
682
|
|
|
$requestResult = unserialize($resultData['content']); |
683
|
|
|
return $requestResult['vars'] ?? []; |
684
|
|
|
} |
685
|
|
|
|
686
|
|
|
/** |
687
|
|
|
* Extract the log information from the current row and retrive it as formatted string. |
688
|
|
|
* |
689
|
|
|
* @param array $resultRow |
690
|
|
|
* @return string |
691
|
|
|
*/ |
692
|
|
|
protected function getResultLog($resultRow) |
693
|
|
|
{ |
694
|
|
|
$content = ''; |
695
|
|
|
if (is_array($resultRow) && array_key_exists('result_data', $resultRow)) { |
696
|
|
|
$requestContent = unserialize($resultRow['result_data']) ?: ['content' => '']; |
697
|
|
|
$requestResult = unserialize($requestContent['content']); |
698
|
|
|
|
699
|
|
|
if (is_array($requestResult) && array_key_exists('log', $requestResult)) { |
700
|
|
|
$content = implode(chr(10), $requestResult['log']); |
701
|
|
|
} |
702
|
|
|
} |
703
|
|
|
return $content; |
704
|
|
|
} |
705
|
|
|
|
706
|
|
|
protected function getResStatus($requestContent): string |
707
|
|
|
{ |
708
|
|
|
if (empty($requestContent)) { |
709
|
|
|
return '-'; |
710
|
|
|
} |
711
|
|
|
$requestResult = unserialize($requestContent['content']); |
712
|
|
|
if (is_array($requestResult)) { |
713
|
|
|
if (empty($requestResult['errorlog'])) { |
714
|
|
|
return 'OK'; |
715
|
|
|
} |
716
|
|
|
return implode("\n", $requestResult['errorlog']); |
717
|
|
|
} |
718
|
|
|
|
719
|
|
|
if (is_bool($requestResult)) { |
720
|
|
|
return 'Error - no info, sorry!'; |
721
|
|
|
} |
722
|
|
|
|
723
|
|
|
return 'Error: ' . substr(preg_replace('/\s+/', ' ', strip_tags($requestResult)), 0, 10000) . '...'; |
724
|
|
|
} |
725
|
|
|
|
726
|
|
|
/** |
727
|
|
|
* This method is used to show an overview about the active an the finished crawling processes |
728
|
|
|
* |
729
|
|
|
* @return string |
730
|
|
|
*/ |
731
|
|
|
protected function processOverviewAction() |
732
|
|
|
{ |
733
|
|
|
$this->view->setTemplate('ProcessOverview'); |
734
|
|
|
$this->runRefreshHooks(); |
735
|
|
|
$this->makeCrawlerProcessableChecks(); |
736
|
|
|
|
737
|
|
|
try { |
738
|
|
|
$this->handleProcessOverviewActions(); |
739
|
|
|
} catch (\Throwable $e) { |
740
|
|
|
$this->isErrorDetected = true; |
741
|
|
|
MessageUtility::addErrorMessage($e->getMessage()); |
742
|
|
|
} |
743
|
|
|
|
744
|
|
|
$processRepository = new ProcessRepository(); |
745
|
|
|
$queueRepository = new QueueRepository(); |
746
|
|
|
|
747
|
|
|
$mode = $this->pObj->MOD_SETTINGS['processListMode']; |
748
|
|
|
if ($mode === 'simple') { |
749
|
|
|
$allProcesses = $processRepository->findAllActive(); |
750
|
|
|
} else { |
751
|
|
|
$allProcesses = $processRepository->findAll(); |
752
|
|
|
} |
753
|
|
|
$isCrawlerEnabled = ! $this->findCrawler()->getDisabled() && ! $this->isErrorDetected; |
754
|
|
|
$currentActiveProcesses = $processRepository->countActive(); |
755
|
|
|
$maxActiveProcesses = MathUtility::forceIntegerInRange($this->extensionSettings['processLimit'], 1, 99, 1); |
756
|
|
|
$this->view->assignMultiple([ |
757
|
|
|
'pageId' => (int) $this->id, |
758
|
|
|
'refreshLink' => $this->getRefreshLink(), |
759
|
|
|
'addLink' => $this->getAddLink($currentActiveProcesses, $maxActiveProcesses, $isCrawlerEnabled), |
760
|
|
|
'modeLink' => $this->getModeLink($mode), |
761
|
|
|
'enableDisableToggle' => $this->getEnableDisableLink($isCrawlerEnabled), |
762
|
|
|
'processCollection' => $allProcesses, |
763
|
|
|
'cliPath' => $this->processManager->getCrawlerCliPath(), |
764
|
|
|
'isCrawlerEnabled' => $isCrawlerEnabled, |
765
|
|
|
'totalUnprocessedItemCount' => $queueRepository->countAllPendingItems(), |
766
|
|
|
'assignedUnprocessedItemCount' => $queueRepository->countAllAssignedPendingItems(), |
767
|
|
|
'activeProcessCount' => $currentActiveProcesses, |
768
|
|
|
'maxActiveProcessCount' => $maxActiveProcesses, |
769
|
|
|
'mode' => $mode, |
770
|
|
|
'displayActions' => 0, |
771
|
|
|
]); |
772
|
|
|
|
773
|
|
|
return $this->view->render(); |
774
|
|
|
} |
775
|
|
|
|
776
|
|
|
/** |
777
|
|
|
* Returns a tag for the refresh icon |
778
|
|
|
* |
779
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
780
|
|
|
*/ |
781
|
|
|
protected function getRefreshLink(): string |
782
|
|
|
{ |
783
|
|
|
return $this->getLinkButton( |
784
|
|
|
'actions-refresh', |
785
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.refresh'), |
786
|
|
|
$this->getInfoModuleUrl(['SET[\'crawleraction\']' => 'crawleraction', 'id' => $this->id]) |
787
|
|
|
); |
788
|
|
|
} |
789
|
|
|
|
790
|
|
|
/** |
791
|
|
|
* Returns a link for the panel to enable or disable the crawler |
792
|
|
|
* |
793
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
794
|
|
|
*/ |
795
|
|
|
protected function getEnableDisableLink(bool $isCrawlerEnabled): string |
796
|
|
|
{ |
797
|
|
|
if ($isCrawlerEnabled) { |
798
|
|
|
// TODO: Icon Should be bigger + Perhaps better icon |
799
|
|
|
return $this->getLinkButton( |
800
|
|
|
'tx-crawler-stop', |
801
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.disablecrawling'), |
802
|
|
|
$this->getInfoModuleUrl(['action' => 'stopCrawling']) |
803
|
|
|
); |
804
|
|
|
} |
805
|
|
|
// TODO: Icon Should be bigger |
806
|
|
|
return $this->getLinkButton( |
807
|
|
|
'tx-crawler-start', |
808
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.enablecrawling'), |
809
|
|
|
$this->getInfoModuleUrl(['action' => 'resumeCrawling']) |
810
|
|
|
); |
811
|
|
|
} |
812
|
|
|
|
813
|
|
|
/** |
814
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
815
|
|
|
*/ |
816
|
|
|
protected function getModeLink(string $mode): string |
817
|
|
|
{ |
818
|
|
|
if ($mode === 'detail') { |
819
|
|
|
return $this->getLinkButton( |
820
|
|
|
'actions-document-view', |
821
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.show.running'), |
822
|
|
|
$this->getInfoModuleUrl(['SET[\'processListMode\']' => 'simple']) |
823
|
|
|
); |
824
|
|
|
} elseif ($mode === 'simple') { |
825
|
|
|
return $this->getLinkButton( |
826
|
|
|
'actions-document-view', |
827
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.show.all'), |
828
|
|
|
$this->getInfoModuleUrl(['SET[\'processListMode\']' => 'detail']) |
829
|
|
|
); |
830
|
|
|
} |
831
|
|
|
return ''; |
832
|
|
|
} |
833
|
|
|
|
834
|
|
|
/** |
835
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
836
|
|
|
*/ |
837
|
|
|
protected function getAddLink(int $currentActiveProcesses, int $maxActiveProcesses, bool $isCrawlerEnabled): string |
838
|
|
|
{ |
839
|
|
|
if (! $isCrawlerEnabled) { |
840
|
|
|
return ''; |
841
|
|
|
} |
842
|
|
|
if ($currentActiveProcesses >= $maxActiveProcesses) { |
843
|
|
|
return ''; |
844
|
|
|
} |
845
|
|
|
|
846
|
|
|
return $this->getLinkButton( |
847
|
|
|
'actions-add', |
848
|
|
|
$this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.process.add'), |
849
|
|
|
$this->getInfoModuleUrl(['action' => 'addProcess']) |
850
|
|
|
); |
851
|
|
|
} |
852
|
|
|
|
853
|
|
|
/** |
854
|
|
|
* Verify that the crawler is executable. |
855
|
|
|
*/ |
856
|
|
|
protected function makeCrawlerProcessableChecks(): void |
857
|
|
|
{ |
858
|
|
|
if (! $this->isPhpForkAvailable()) { |
859
|
|
|
$this->isErrorDetected = true; |
860
|
|
|
MessageUtility::addErrorMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:message.noPhpForkAvailable')); |
861
|
|
|
} |
862
|
|
|
|
863
|
|
|
$exitCode = 0; |
864
|
|
|
$out = []; |
865
|
|
|
CommandUtility::exec( |
866
|
|
|
PhpBinaryUtility::getPhpBinary() . ' -v', |
867
|
|
|
$out, |
868
|
|
|
$exitCode |
869
|
|
|
); |
870
|
|
|
if ($exitCode > 0) { |
871
|
|
|
$this->isErrorDetected = true; |
872
|
|
|
MessageUtility::addErrorMessage(sprintf($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:message.phpBinaryNotFound'), htmlspecialchars($this->extensionSettings['phpPath']))); |
873
|
|
|
} |
874
|
|
|
} |
875
|
|
|
|
876
|
|
|
/** |
877
|
|
|
* Indicate that the required PHP method "popen" is |
878
|
|
|
* available in the system. |
879
|
|
|
*/ |
880
|
|
|
protected function isPhpForkAvailable(): bool |
881
|
|
|
{ |
882
|
|
|
return function_exists('popen'); |
883
|
|
|
} |
884
|
|
|
|
885
|
|
|
/** |
886
|
|
|
* Method to handle incomming actions of the process overview |
887
|
|
|
* |
888
|
|
|
* @throws \Exception |
889
|
|
|
*/ |
890
|
|
|
protected function handleProcessOverviewActions(): void |
891
|
|
|
{ |
892
|
|
|
$crawler = $this->findCrawler(); |
893
|
|
|
|
894
|
|
|
switch (GeneralUtility::_GP('action')) { |
895
|
|
|
case 'stopCrawling': |
896
|
|
|
//set the cli status to disable (all processes will be terminated) |
897
|
|
|
$crawler->setDisabled(true); |
898
|
|
|
break; |
899
|
|
|
case 'resumeCrawling': |
900
|
|
|
//set the cli status to end (all processes will be terminated) |
901
|
|
|
$crawler->setDisabled(false); |
902
|
|
|
break; |
903
|
|
|
case 'addProcess': |
904
|
|
|
if ($this->processManager->startProcess() === false) { |
|
|
|
|
905
|
|
|
throw new \Exception($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.newprocesserror')); |
906
|
|
|
} |
907
|
|
|
MessageUtility::addNoticeMessage($this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.newprocess')); |
908
|
|
|
break; |
909
|
|
|
} |
910
|
|
|
} |
911
|
|
|
|
912
|
|
|
/** |
913
|
|
|
* Returns the singleton instance of the crawler. |
914
|
|
|
*/ |
915
|
|
|
protected function findCrawler(): CrawlerController |
916
|
|
|
{ |
917
|
|
|
if (! $this->crawlerController instanceof CrawlerController) { |
|
|
|
|
918
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
919
|
|
|
} |
920
|
|
|
return $this->crawlerController; |
921
|
|
|
} |
922
|
|
|
|
923
|
|
|
/***************************** |
924
|
|
|
* |
925
|
|
|
* General Helper Functions |
926
|
|
|
* |
927
|
|
|
*****************************/ |
928
|
|
|
|
929
|
|
|
/** |
930
|
|
|
* Create selector box |
931
|
|
|
* |
932
|
|
|
* @param array $optArray Options key(value) => label pairs |
933
|
|
|
* @param string $name Selector box name |
934
|
|
|
* @param string|array $value Selector box value (array for multiple...) |
935
|
|
|
* @param boolean $multiple If set, will draw multiple box. |
936
|
|
|
* |
937
|
|
|
* @return string HTML select element |
938
|
|
|
*/ |
939
|
|
|
protected function selectorBox($optArray, $name, $value, bool $multiple): string |
940
|
|
|
{ |
941
|
|
|
if (! is_string($value) || ! is_array($value)) { |
|
|
|
|
942
|
|
|
$value = ''; |
943
|
|
|
} |
944
|
|
|
|
945
|
|
|
$options = []; |
946
|
|
|
foreach ($optArray as $key => $val) { |
947
|
|
|
$selected = (! $multiple && ! strcmp($value, (string) $key)) || ($multiple && in_array($key, (array) $value, true)); |
|
|
|
|
948
|
|
|
$options[] = ' |
949
|
|
|
<option value="' . $key . '" ' . ($selected ? ' selected="selected"' : '') . '>' . htmlspecialchars($val) . '</option>'; |
950
|
|
|
} |
951
|
|
|
|
952
|
|
|
return '<select class="form-control" name="' . htmlspecialchars($name . ($multiple ? '[]' : '')) . '"' . ($multiple ? ' multiple' : '') . '>' . implode('', $options) . '</select>'; |
953
|
|
|
} |
954
|
|
|
|
955
|
|
|
/** |
956
|
|
|
* Activate hooks |
957
|
|
|
*/ |
958
|
|
|
protected function runRefreshHooks(): void |
959
|
|
|
{ |
960
|
|
|
$crawlerLib = GeneralUtility::makeInstance(CrawlerController::class); |
961
|
|
|
foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['refresh_hooks'] ?? [] as $objRef) { |
962
|
|
|
/** @var CrawlerHookInterface $hookObj */ |
963
|
|
|
$hookObj = GeneralUtility::makeInstance($objRef); |
964
|
|
|
if (is_object($hookObj)) { |
965
|
|
|
$hookObj->crawler_init($crawlerLib); |
966
|
|
|
} |
967
|
|
|
} |
968
|
|
|
} |
969
|
|
|
|
970
|
|
|
protected function initializeView(): void |
971
|
|
|
{ |
972
|
|
|
$view = GeneralUtility::makeInstance(StandaloneView::class); |
973
|
|
|
$view->setLayoutRootPaths(['EXT:crawler/Resources/Private/Layouts']); |
974
|
|
|
$view->setPartialRootPaths(['EXT:crawler/Resources/Private/Partials']); |
975
|
|
|
$view->setTemplateRootPaths(['EXT:crawler/Resources/Private/Templates/Backend']); |
976
|
|
|
$view->getRequest()->setControllerExtensionName('Crawler'); |
977
|
|
|
$this->view = $view; |
978
|
|
|
} |
979
|
|
|
|
980
|
|
|
protected function getLanguageService(): LanguageService |
981
|
|
|
{ |
982
|
|
|
return $GLOBALS['LANG']; |
983
|
|
|
} |
984
|
|
|
|
985
|
|
|
protected function getLinkButton(string $iconIdentifier, string $title, UriInterface $href): string |
986
|
|
|
{ |
987
|
|
|
$moduleTemplate = GeneralUtility::makeInstance(ModuleTemplate::class); |
988
|
|
|
$buttonBar = $moduleTemplate->getDocHeaderComponent()->getButtonBar(); |
989
|
|
|
return (string) $buttonBar->makeLinkButton() |
990
|
|
|
->setHref((string) $href) |
991
|
|
|
->setIcon($this->iconFactory->getIcon($iconIdentifier, Icon::SIZE_SMALL)) |
992
|
|
|
->setTitle($title) |
993
|
|
|
->setShowLabelText(true); |
994
|
|
|
} |
995
|
|
|
|
996
|
|
|
/** |
997
|
|
|
* Returns the URL to the current module, including $_GET['id']. |
998
|
|
|
* |
999
|
|
|
* @param array $uriParameters optional parameters to add to the URL |
1000
|
|
|
* |
1001
|
|
|
* @throws \TYPO3\CMS\Backend\Routing\Exception\RouteNotFoundException |
1002
|
|
|
*/ |
1003
|
|
|
protected function getInfoModuleUrl(array $uriParameters = []): Uri |
1004
|
|
|
{ |
1005
|
|
|
if (GeneralUtility::_GP('id')) { |
1006
|
|
|
$uriParameters = array_merge($uriParameters, [ |
1007
|
|
|
'id' => GeneralUtility::_GP('id'), |
1008
|
|
|
]); |
1009
|
|
|
} |
1010
|
|
|
/** @var UriBuilder $uriBuilder */ |
1011
|
|
|
$uriBuilder = GeneralUtility::makeInstance(UriBuilder::class); |
1012
|
|
|
return $uriBuilder->buildUriFromRoute('web_info', $uriParameters); |
1013
|
|
|
} |
1014
|
|
|
|
1015
|
|
|
private function getDepthDropDownHtml(): string |
1016
|
|
|
{ |
1017
|
|
|
return BackendUtility::getFuncMenu( |
1018
|
|
|
$this->id, |
1019
|
|
|
'SET[depth]', |
1020
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
1021
|
|
|
$this->pObj->MOD_MENU['depth'] |
1022
|
|
|
); |
1023
|
|
|
} |
1024
|
|
|
|
1025
|
|
|
private function getDisplayLogFilterHtml(int $setId): string |
1026
|
|
|
{ |
1027
|
|
|
return $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.display') . ': ' . BackendUtility::getFuncMenu( |
1028
|
|
|
$this->id, |
1029
|
|
|
'SET[log_display]', |
1030
|
|
|
$this->pObj->MOD_SETTINGS['log_display'], |
1031
|
|
|
$this->pObj->MOD_MENU['log_display'], |
1032
|
|
|
'index.php', |
1033
|
|
|
'&setID=' . $setId |
1034
|
|
|
); |
1035
|
|
|
} |
1036
|
|
|
|
1037
|
|
|
private function getShowResultLogCheckBoxHtml(int $setId, string $quiPart): string |
1038
|
|
|
{ |
1039
|
|
|
return BackendUtility::getFuncCheck( |
1040
|
|
|
$this->id, |
1041
|
|
|
'SET[log_resultLog]', |
1042
|
|
|
$this->pObj->MOD_SETTINGS['log_resultLog'], |
1043
|
|
|
'index.php', |
1044
|
|
|
'&setID=' . $setId . $quiPart |
1045
|
|
|
) . ' ' . $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.showresultlog'); |
1046
|
|
|
} |
1047
|
|
|
|
1048
|
|
|
private function getShowFeVarsCheckBoxHtml(int $setId, string $quiPart): string |
1049
|
|
|
{ |
1050
|
|
|
return BackendUtility::getFuncCheck( |
1051
|
|
|
$this->id, |
1052
|
|
|
'SET[log_feVars]', |
1053
|
|
|
$this->pObj->MOD_SETTINGS['log_feVars'], |
1054
|
|
|
'index.php', |
1055
|
|
|
'&setID=' . $setId . $quiPart |
1056
|
|
|
) . ' ' . $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.showfevars'); |
1057
|
|
|
} |
1058
|
|
|
|
1059
|
|
|
private function getItemsPerPageDropDownHtml(): string |
1060
|
|
|
{ |
1061
|
|
|
return $this->getLanguageService()->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xlf:labels.itemsPerPage') . ': ' . |
1062
|
|
|
BackendUtility::getFuncMenu( |
1063
|
|
|
$this->id, |
1064
|
|
|
'SET[itemsPerPage]', |
1065
|
|
|
$this->pObj->MOD_SETTINGS['itemsPerPage'], |
1066
|
|
|
$this->pObj->MOD_MENU['itemsPerPage'] |
1067
|
|
|
); |
1068
|
|
|
} |
1069
|
|
|
} |
1070
|
|
|
|
For hinted functions/methods where all return statements with the correct type are only reachable via conditions, ?null? gets implicitly returned which may be incompatible with the hinted type. Let?s take a look at an example: