1
|
|
|
<?php |
2
|
|
|
namespace AOE\Crawler\Backend; |
3
|
|
|
|
4
|
|
|
/*************************************************************** |
5
|
|
|
* Copyright notice |
6
|
|
|
* |
7
|
|
|
* (c) 2018 AOE GmbH <[email protected]> |
8
|
|
|
* |
9
|
|
|
* All rights reserved |
10
|
|
|
* |
11
|
|
|
* This script is part of the TYPO3 project. The TYPO3 project is |
12
|
|
|
* free software; you can redistribute it and/or modify |
13
|
|
|
* it under the terms of the GNU General Public License as published by |
14
|
|
|
* the Free Software Foundation; either version 3 of the License, or |
15
|
|
|
* (at your option) any later version. |
16
|
|
|
* |
17
|
|
|
* The GNU General Public License can be found at |
18
|
|
|
* http://www.gnu.org/copyleft/gpl.html. |
19
|
|
|
* |
20
|
|
|
* This script is distributed in the hope that it will be useful, |
21
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
22
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
23
|
|
|
* GNU General Public License for more details. |
24
|
|
|
* |
25
|
|
|
* This copyright notice MUST APPEAR in all copies of the script! |
26
|
|
|
***************************************************************/ |
27
|
|
|
|
28
|
|
|
use AOE\Crawler\Backend\View\PaginationView; |
29
|
|
|
use AOE\Crawler\Backend\View\ProcessListView; |
30
|
|
|
use AOE\Crawler\Controller\CrawlerController; |
31
|
|
|
use AOE\Crawler\Domain\Model\Reason; |
32
|
|
|
use AOE\Crawler\Domain\Repository\ProcessRepository; |
33
|
|
|
use AOE\Crawler\Domain\Repository\QueueRepository; |
34
|
|
|
use AOE\Crawler\Event\EventDispatcher; |
35
|
|
|
use AOE\Crawler\Service\ProcessService; |
36
|
|
|
use AOE\Crawler\Utility\IconUtility; |
37
|
|
|
use TYPO3\CMS\Backend\Module\AbstractFunctionModule; |
38
|
|
|
use TYPO3\CMS\Backend\Tree\View\PageTreeView; |
39
|
|
|
use TYPO3\CMS\Backend\Utility\BackendUtility; |
40
|
|
|
use TYPO3\CMS\Core\Authentication\BackendUserAuthentication; |
41
|
|
|
use TYPO3\CMS\Core\Imaging\Icon; |
42
|
|
|
use TYPO3\CMS\Core\Messaging\FlashMessage; |
43
|
|
|
use TYPO3\CMS\Core\Messaging\FlashMessageService; |
44
|
|
|
use TYPO3\CMS\Core\Utility\DebugUtility; |
45
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
46
|
|
|
use TYPO3\CMS\Core\Utility\MathUtility; |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* Class BackendModule |
50
|
|
|
* |
51
|
|
|
* @package AOE\Crawler\Backend |
52
|
|
|
*/ |
53
|
|
|
class BackendModule extends AbstractFunctionModule |
54
|
|
|
{ |
55
|
|
|
// Internal, dynamic: |
56
|
|
|
public $duplicateTrack = []; |
57
|
|
|
public $submitCrawlUrls = false; |
58
|
|
|
public $downloadCrawlUrls = false; |
59
|
|
|
|
60
|
|
|
public $scheduledTime = 0; |
61
|
|
|
public $reqMinute = 0; |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* @var array holds the selection of configuration from the configuration selector box |
65
|
|
|
*/ |
66
|
|
|
public $incomingConfigurationSelection = []; |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* @var CrawlerController |
70
|
|
|
*/ |
71
|
|
|
public $crawlerController; |
72
|
|
|
|
73
|
|
|
public $CSVaccu = []; |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* If true the user requested a CSV export of the queue |
77
|
|
|
* |
78
|
|
|
* @var boolean |
79
|
|
|
*/ |
80
|
|
|
public $CSVExport = false; |
81
|
|
|
|
82
|
|
|
public $downloadUrls = []; |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* Holds the configuration from ext_conf_template loaded by loadExtensionSettings() |
86
|
|
|
* |
87
|
|
|
* @var array |
88
|
|
|
*/ |
89
|
|
|
protected $extensionSettings = []; |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* Indicate that an flash message with an error is present. |
93
|
|
|
* |
94
|
|
|
* @var boolean |
95
|
|
|
*/ |
96
|
|
|
protected $isErrorDetected = false; |
97
|
|
|
|
98
|
|
|
/** |
99
|
|
|
* @var ProcessService |
100
|
|
|
*/ |
101
|
|
|
protected $processManager; |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* the constructor |
105
|
|
|
*/ |
106
|
|
|
public function __construct() |
107
|
|
|
{ |
108
|
|
|
$this->processManager = new ProcessService(); |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* Additions to the function menu array |
113
|
|
|
* |
114
|
|
|
* @return array Menu array |
115
|
|
|
*/ |
116
|
|
|
public function modMenu() |
117
|
|
|
{ |
118
|
|
|
global $LANG; |
119
|
|
|
|
120
|
|
|
return [ |
121
|
|
|
'depth' => [ |
122
|
|
|
0 => $LANG->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_0'), |
123
|
|
|
1 => $LANG->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_1'), |
124
|
|
|
2 => $LANG->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_2'), |
125
|
|
|
3 => $LANG->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_3'), |
126
|
|
|
4 => $LANG->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_4'), |
127
|
|
|
99 => $LANG->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_infi'), |
128
|
|
|
], |
129
|
|
|
'crawlaction' => [ |
130
|
|
|
'start' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.start'), |
131
|
|
|
'log' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.log'), |
132
|
|
|
'multiprocess' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.multiprocess') |
133
|
|
|
], |
134
|
|
|
'log_resultLog' => '', |
135
|
|
|
'log_feVars' => '', |
136
|
|
|
'processListMode' => '', |
137
|
|
|
'log_display' => [ |
138
|
|
|
'all' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.all'), |
139
|
|
|
'pending' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.pending'), |
140
|
|
|
'finished' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.finished') |
141
|
|
|
], |
142
|
|
|
'itemsPerPage' => [ |
143
|
|
|
'5' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage.5'), |
144
|
|
|
'10' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage.10'), |
145
|
|
|
'50' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage.50'), |
146
|
|
|
'0' => $LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage.0') |
147
|
|
|
] |
148
|
|
|
]; |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
/** |
152
|
|
|
* Load extension settings |
153
|
|
|
* |
154
|
|
|
* @return void |
155
|
|
|
*/ |
156
|
|
|
protected function loadExtensionSettings() |
157
|
|
|
{ |
158
|
|
|
$this->extensionSettings = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['crawler']); |
159
|
|
|
} |
160
|
|
|
|
161
|
|
|
/** |
162
|
|
|
* Main function |
163
|
|
|
* |
164
|
|
|
* @return string HTML output |
165
|
|
|
*/ |
166
|
|
|
public function main() |
167
|
|
|
{ |
168
|
|
|
global $LANG, $BACK_PATH; |
169
|
|
|
|
170
|
|
|
$this->incLocalLang(); |
171
|
|
|
|
172
|
|
|
$this->loadExtensionSettings(); |
173
|
|
|
if (empty($this->pObj->MOD_SETTINGS['processListMode'])) { |
174
|
|
|
$this->pObj->MOD_SETTINGS['processListMode'] = 'simple'; |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
// Set CSS styles specific for this document: |
178
|
|
|
$this->pObj->content = str_replace('/*###POSTCSSMARKER###*/', ' |
179
|
|
|
TABLE.c-list TR TD { white-space: nowrap; vertical-align: top; } |
180
|
|
|
', $this->pObj->content); |
181
|
|
|
|
182
|
|
|
$this->pObj->content .= '<style type="text/css"><!-- |
183
|
|
|
table.url-table, |
184
|
|
|
table.param-expanded, |
185
|
|
|
table.crawlerlog { |
186
|
|
|
border-bottom: 1px solid grey; |
187
|
|
|
border-spacing: 0; |
188
|
|
|
border-collapse: collapse; |
189
|
|
|
} |
190
|
|
|
table.crawlerlog td, |
191
|
|
|
table.url-table td { |
192
|
|
|
border: 1px solid lightgrey; |
193
|
|
|
border-bottom: 1px solid grey; |
194
|
|
|
white-space: nowrap; vertical-align: top; |
195
|
|
|
} |
196
|
|
|
--></style> |
197
|
|
|
<link rel="stylesheet" type="text/css" href="' . $BACK_PATH . '../typo3conf/ext/crawler/template/res.css" /> |
198
|
|
|
'; |
199
|
|
|
|
200
|
|
|
// Type function menu: |
201
|
|
|
$h_func = BackendUtility::getFuncMenu( |
202
|
|
|
$this->pObj->id, |
203
|
|
|
'SET[crawlaction]', |
204
|
|
|
$this->pObj->MOD_SETTINGS['crawlaction'], |
205
|
|
|
$this->pObj->MOD_MENU['crawlaction'], |
206
|
|
|
'index.php' |
207
|
|
|
); |
208
|
|
|
|
209
|
|
|
// Additional menus for the log type: |
210
|
|
|
if ($this->pObj->MOD_SETTINGS['crawlaction'] === 'log') { |
211
|
|
|
$h_func .= BackendUtility::getFuncMenu( |
212
|
|
|
$this->pObj->id, |
213
|
|
|
'SET[depth]', |
214
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
215
|
|
|
$this->pObj->MOD_MENU['depth'], |
216
|
|
|
'index.php' |
217
|
|
|
); |
218
|
|
|
|
219
|
|
|
$quiPart = GeneralUtility::_GP('qid_details') ? '&qid_details=' . intval(GeneralUtility::_GP('qid_details')) : ''; |
220
|
|
|
|
221
|
|
|
$setId = intval(GeneralUtility::_GP('setID')); |
222
|
|
|
|
223
|
|
|
$h_func .= '<hr/>' . |
224
|
|
|
$GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.display') . ': ' . BackendUtility::getFuncMenu($this->pObj->id, 'SET[log_display]', $this->pObj->MOD_SETTINGS['log_display'], $this->pObj->MOD_MENU['log_display'], 'index.php', '&setID=' . $setId) . ' - ' . |
225
|
|
|
$GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.showresultlog') . ': ' . BackendUtility::getFuncCheck($this->pObj->id, 'SET[log_resultLog]', $this->pObj->MOD_SETTINGS['log_resultLog'], 'index.php', '&setID=' . $setId . $quiPart) . ' - ' . |
226
|
|
|
$GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.showfevars') . ': ' . BackendUtility::getFuncCheck($this->pObj->id, 'SET[log_feVars]', $this->pObj->MOD_SETTINGS['log_feVars'], 'index.php', '&setID=' . $setId . $quiPart) . ' - ' . |
227
|
|
|
$GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.itemsPerPage') . ': ' . |
228
|
|
|
BackendUtility::getFuncMenu( |
229
|
|
|
$this->pObj->id, |
230
|
|
|
'SET[itemsPerPage]', |
231
|
|
|
$this->pObj->MOD_SETTINGS['itemsPerPage'], |
232
|
|
|
$this->pObj->MOD_MENU['itemsPerPage'], |
233
|
|
|
'index.php' |
234
|
|
|
); |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
$theOutput = $this->pObj->doc->section($LANG->getLL('title'), $h_func, false, true); |
238
|
|
|
|
239
|
|
|
// Branch based on type: |
240
|
|
|
switch ((string)$this->pObj->MOD_SETTINGS['crawlaction']) { |
241
|
|
|
case 'start': |
242
|
|
|
if (empty($this->pObj->id)) { |
243
|
|
|
$this->addErrorMessage($GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noPageSelected')); |
244
|
|
|
} else { |
245
|
|
|
$theOutput .= $this->pObj->doc->section('', $this->drawURLs(), false, true); |
246
|
|
|
} |
247
|
|
|
break; |
248
|
|
|
case 'log': |
249
|
|
|
if (empty($this->pObj->id)) { |
250
|
|
|
$this->addErrorMessage($GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noPageSelected')); |
251
|
|
|
} else { |
252
|
|
|
$theOutput .= $this->pObj->doc->section('', $this->drawLog(), false, true); |
253
|
|
|
} |
254
|
|
|
break; |
255
|
|
|
case 'cli': |
256
|
|
|
// TODO: drawCLIstatus is not defined, why did we refactor it away? |
257
|
|
|
$theOutput .= $this->pObj->doc->section('', $this->drawCLIstatus(), false, true); |
|
|
|
|
258
|
|
|
break; |
259
|
|
|
case 'multiprocess': |
260
|
|
|
$theOutput .= $this->pObj->doc->section('', $this->drawProcessOverviewAction(), false, true); |
261
|
|
|
break; |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
return $theOutput; |
265
|
|
|
} |
266
|
|
|
|
267
|
|
|
/******************************* |
268
|
|
|
* |
269
|
|
|
* Generate URLs for crawling: |
270
|
|
|
* |
271
|
|
|
******************************/ |
272
|
|
|
|
273
|
|
|
/** |
274
|
|
|
* Produces a table with overview of the URLs to be crawled for each page |
275
|
|
|
* |
276
|
|
|
* @return string HTML output |
277
|
|
|
*/ |
278
|
|
|
public function drawURLs() |
279
|
|
|
{ |
280
|
|
|
global $BACK_PATH, $BE_USER; |
281
|
|
|
|
282
|
|
|
$crawlerParameter = GeneralUtility::_GP('_crawl'); |
283
|
|
|
$downloadParameter = GeneralUtility::_GP('_download'); |
284
|
|
|
|
285
|
|
|
// Init: |
286
|
|
|
$this->duplicateTrack = []; |
287
|
|
|
$this->submitCrawlUrls = isset($crawlerParameter); |
288
|
|
|
$this->downloadCrawlUrls = isset($downloadParameter); |
289
|
|
|
$this->makeCrawlerProcessableChecks(); |
290
|
|
|
|
291
|
|
|
switch ((string) GeneralUtility::_GP('tstamp')) { |
292
|
|
|
case 'midnight': |
293
|
|
|
$this->scheduledTime = mktime(0, 0, 0); |
294
|
|
|
break; |
295
|
|
|
case '04:00': |
296
|
|
|
$this->scheduledTime = mktime(0, 0, 0) + 4 * 3600; |
297
|
|
|
break; |
298
|
|
|
case 'now': |
299
|
|
|
default: |
300
|
|
|
$this->scheduledTime = time(); |
301
|
|
|
break; |
302
|
|
|
} |
303
|
|
|
// $this->reqMinute = \TYPO3\CMS\Core\Utility\GeneralUtility::intInRange(\TYPO3\CMS\Core\Utility\GeneralUtility::_GP('perminute'),1,10000); |
304
|
|
|
// TODO: check relevance |
305
|
|
|
$this->reqMinute = 1000; |
306
|
|
|
|
307
|
|
|
$this->incomingConfigurationSelection = GeneralUtility::_GP('configurationSelection'); |
|
|
|
|
308
|
|
|
$this->incomingConfigurationSelection = is_array($this->incomingConfigurationSelection) ? $this->incomingConfigurationSelection : []; |
309
|
|
|
|
310
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
311
|
|
|
$this->crawlerController->setAccessMode('gui'); |
312
|
|
|
$this->crawlerController->setID = GeneralUtility::md5int(microtime()); |
313
|
|
|
|
314
|
|
|
if (empty($this->incomingConfigurationSelection) |
315
|
|
|
|| (count($this->incomingConfigurationSelection) == 1 && empty($this->incomingConfigurationSelection[0])) |
316
|
|
|
) { |
317
|
|
|
$this->addWarningMessage($GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noConfigSelected')); |
318
|
|
|
$code = ' |
319
|
|
|
<tr> |
320
|
|
|
<td colspan="7"><b>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noConfigSelected') . '</b></td> |
321
|
|
|
</tr>'; |
322
|
|
|
} else { |
323
|
|
|
if ($this->submitCrawlUrls) { |
324
|
|
|
$reason = new Reason(); |
325
|
|
|
$reason->setReason(Reason::REASON_GUI_SUBMIT); |
326
|
|
|
|
327
|
|
|
if ($BE_USER instanceof BackendUserAuthentication) { |
328
|
|
|
$username = $BE_USER->user['username']; |
329
|
|
|
$reason->setDetailText('The user ' . $username . ' added pages to the crawler queue manually '); |
330
|
|
|
} |
331
|
|
|
|
332
|
|
|
EventDispatcher::getInstance()->post( |
333
|
|
|
'invokeQueueChange', |
334
|
|
|
$this->findCrawler()->setID, |
335
|
|
|
['reason' => $reason] |
336
|
|
|
); |
337
|
|
|
} |
338
|
|
|
|
339
|
|
|
$code = $this->crawlerController->getPageTreeAndUrls( |
340
|
|
|
$this->pObj->id, |
341
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
342
|
|
|
$this->scheduledTime, |
343
|
|
|
$this->reqMinute, |
344
|
|
|
$this->submitCrawlUrls, |
345
|
|
|
$this->downloadCrawlUrls, |
346
|
|
|
[], // Do not filter any processing instructions |
347
|
|
|
$this->incomingConfigurationSelection |
348
|
|
|
); |
349
|
|
|
} |
350
|
|
|
|
351
|
|
|
$this->downloadUrls = $this->crawlerController->downloadUrls; |
352
|
|
|
$this->duplicateTrack = $this->crawlerController->duplicateTrack; |
353
|
|
|
|
354
|
|
|
$output = ''; |
355
|
|
|
if ($code) { |
356
|
|
|
$output .= '<h3>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.configuration') . ':</h3>'; |
357
|
|
|
$output .= '<input type="hidden" name="id" value="' . intval($this->pObj->id) . '" />'; |
358
|
|
|
|
359
|
|
|
if (!$this->submitCrawlUrls) { |
360
|
|
|
$output .= $this->drawURLs_cfgSelectors() . '<br />'; |
361
|
|
|
$output .= '<input type="submit" name="_update" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.triggerUpdate') . '" /> '; |
362
|
|
|
$output .= '<input type="submit" name="_crawl" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.triggerCrawl') . '" /> '; |
363
|
|
|
$output .= '<input type="submit" name="_download" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.triggerDownload') . '" /><br /><br />'; |
364
|
|
|
$output .= $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.count') . ': ' . count(array_keys($this->duplicateTrack)) . '<br />'; |
365
|
|
|
$output .= $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.curtime') . ': ' . date('H:i:s', time()) . '<br />'; |
366
|
|
|
$output .= '<br /> |
367
|
|
|
<table class="lrPadding c-list url-table">' . |
368
|
|
|
$this->drawURLs_printTableHeader() . |
369
|
|
|
$code . |
370
|
|
|
'</table>'; |
371
|
|
|
} else { |
372
|
|
|
$output .= count(array_keys($this->duplicateTrack)) . ' ' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.submitted') . '. <br /><br />'; |
373
|
|
|
$output .= '<input type="submit" name="_" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.continue') . '" />'; |
374
|
|
|
$output .= '<input type="submit" onclick="this.form.elements[\'SET[crawlaction]\'].value=\'log\';" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.continueinlog') . '" />'; |
375
|
|
|
} |
376
|
|
|
} |
377
|
|
|
|
378
|
|
|
// Download Urls to crawl: |
379
|
|
|
if ($this->downloadCrawlUrls) { |
380
|
|
|
|
381
|
|
|
// Creating output header: |
382
|
|
|
$mimeType = 'application/octet-stream'; |
383
|
|
|
Header('Content-Type: ' . $mimeType); |
384
|
|
|
Header('Content-Disposition: attachment; filename=CrawlerUrls.txt'); |
385
|
|
|
|
386
|
|
|
// Printing the content of the CSV lines: |
387
|
|
|
echo implode(chr(13) . chr(10), $this->downloadUrls); |
388
|
|
|
|
389
|
|
|
exit; |
|
|
|
|
390
|
|
|
} |
391
|
|
|
|
392
|
|
|
return $output; |
393
|
|
|
} |
394
|
|
|
|
395
|
|
|
/** |
396
|
|
|
* Draws the configuration selectors for compiling URLs: |
397
|
|
|
* |
398
|
|
|
* @return string HTML table |
399
|
|
|
*/ |
400
|
|
|
public function drawURLs_cfgSelectors() |
401
|
|
|
{ |
402
|
|
|
$cell = []; |
403
|
|
|
|
404
|
|
|
// depth |
405
|
|
|
$cell[] = $this->selectorBox( |
406
|
|
|
[ |
407
|
|
|
0 => $GLOBALS['LANG']->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_0'), |
408
|
|
|
1 => $GLOBALS['LANG']->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_1'), |
409
|
|
|
2 => $GLOBALS['LANG']->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_2'), |
410
|
|
|
3 => $GLOBALS['LANG']->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_3'), |
411
|
|
|
4 => $GLOBALS['LANG']->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_4'), |
412
|
|
|
99 => $GLOBALS['LANG']->sL('LLL:EXT:lang/Resources/Private/Language/locallang_core.xlf:labels.depth_infi'), |
413
|
|
|
], |
414
|
|
|
'SET[depth]', |
415
|
|
|
$this->pObj->MOD_SETTINGS['depth'], |
416
|
|
|
false |
417
|
|
|
); |
418
|
|
|
$availableConfigurations = $this->crawlerController->getConfigurationsForBranch($this->pObj->id, $this->pObj->MOD_SETTINGS['depth'] ? $this->pObj->MOD_SETTINGS['depth'] : 0); |
419
|
|
|
|
420
|
|
|
// Configurations |
421
|
|
|
$cell[] = $this->selectorBox( |
422
|
|
|
empty($availableConfigurations) ? [] : array_combine($availableConfigurations, $availableConfigurations), |
423
|
|
|
'configurationSelection', |
424
|
|
|
$this->incomingConfigurationSelection, |
|
|
|
|
425
|
|
|
true |
426
|
|
|
); |
427
|
|
|
|
428
|
|
|
// Scheduled time: |
429
|
|
|
$cell[] = $this->selectorBox( |
430
|
|
|
[ |
431
|
|
|
'now' => $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.time.now'), |
432
|
|
|
'midnight' => $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.time.midnight'), |
433
|
|
|
'04:00' => $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.time.4am'), |
434
|
|
|
], |
435
|
|
|
'tstamp', |
436
|
|
|
GeneralUtility::_POST('tstamp'), |
|
|
|
|
437
|
|
|
false |
438
|
|
|
); |
439
|
|
|
|
440
|
|
|
$output = ' |
441
|
|
|
<table class="lrPadding c-list"> |
442
|
|
|
<tr class="bgColor5 tableheader"> |
443
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.depth') . ':</td> |
444
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.configurations') . ':</td> |
445
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.scheduled') . ':</td> |
446
|
|
|
</tr> |
447
|
|
|
<tr class="bgColor4"> |
448
|
|
|
<td valign="top">' . implode('</td> |
449
|
|
|
<td valign="top">', $cell) . '</td> |
450
|
|
|
</tr> |
451
|
|
|
</table>'; |
452
|
|
|
|
453
|
|
|
return $output; |
454
|
|
|
} |
455
|
|
|
|
456
|
|
|
/** |
457
|
|
|
* Create Table header row for URL display |
458
|
|
|
* |
459
|
|
|
* @return string Table header |
460
|
|
|
*/ |
461
|
|
|
public function drawURLs_printTableHeader() |
462
|
|
|
{ |
463
|
|
|
$content = ' |
464
|
|
|
<tr class="bgColor5 tableheader"> |
465
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.pagetitle') . ':</td> |
466
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.key') . ':</td> |
467
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.parametercfg') . ':</td> |
468
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.values') . ':</td> |
469
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.urls') . ':</td> |
470
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.options') . ':</td> |
471
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.parameters') . ':</td> |
472
|
|
|
</tr>'; |
473
|
|
|
|
474
|
|
|
return $content; |
475
|
|
|
} |
476
|
|
|
|
477
|
|
|
/******************************* |
478
|
|
|
* |
479
|
|
|
* Shows log of indexed URLs |
480
|
|
|
* |
481
|
|
|
******************************/ |
482
|
|
|
|
483
|
|
|
/** |
484
|
|
|
* Shows the log of indexed URLs |
485
|
|
|
* |
486
|
|
|
* @return string HTML output |
487
|
|
|
*/ |
488
|
|
|
public function drawLog() |
489
|
|
|
{ |
490
|
|
|
global $BACK_PATH; |
491
|
|
|
$output = ''; |
492
|
|
|
|
493
|
|
|
// Init: |
494
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
495
|
|
|
$this->crawlerController->setAccessMode('gui'); |
496
|
|
|
$this->crawlerController->setID = GeneralUtility::md5int(microtime()); |
497
|
|
|
|
498
|
|
|
$csvExport = GeneralUtility::_POST('_csv'); |
499
|
|
|
$this->CSVExport = isset($csvExport); |
500
|
|
|
|
501
|
|
|
// Read URL: |
502
|
|
|
if (GeneralUtility::_GP('qid_read')) { |
503
|
|
|
$this->crawlerController->readUrl(intval(GeneralUtility::_GP('qid_read')), true); |
504
|
|
|
} |
505
|
|
|
|
506
|
|
|
// Look for set ID sent - if it is, we will display contents of that set: |
507
|
|
|
$showSetId = intval(GeneralUtility::_GP('setID')); |
508
|
|
|
|
509
|
|
|
// Show details: |
510
|
|
|
if (GeneralUtility::_GP('qid_details')) { |
511
|
|
|
|
512
|
|
|
// Get entry record: |
513
|
|
|
list($q_entry) = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('*', 'tx_crawler_queue', 'qid=' . intval(GeneralUtility::_GP('qid_details'))); |
514
|
|
|
|
515
|
|
|
// Explode values: |
516
|
|
|
$resStatus = $this->getResStatus($q_entry); |
517
|
|
|
$q_entry['parameters'] = unserialize($q_entry['parameters']); |
518
|
|
|
$q_entry['result_data'] = unserialize($q_entry['result_data']); |
519
|
|
|
if (is_array($q_entry['result_data'])) { |
520
|
|
|
$q_entry['result_data']['content'] = unserialize($q_entry['result_data']['content']); |
521
|
|
|
|
522
|
|
|
if (!$this->pObj->MOD_SETTINGS['log_resultLog']) { |
523
|
|
|
unset($q_entry['result_data']['content']['log']); |
524
|
|
|
} |
525
|
|
|
} |
526
|
|
|
|
527
|
|
|
// Print rudimentary details: |
528
|
|
|
$output .= ' |
529
|
|
|
<br /><br /> |
530
|
|
|
<input type="submit" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.back') . '" name="_back" /> |
531
|
|
|
<input type="hidden" value="' . $this->pObj->id . '" name="id" /> |
532
|
|
|
<input type="hidden" value="' . $showSetId . '" name="setID" /> |
533
|
|
|
<br /> |
534
|
|
|
Current server time: ' . date('H:i:s', time()) . '<br />' . |
535
|
|
|
'Status: ' . $resStatus . '<br />' . |
536
|
|
|
DebugUtility::viewArray($q_entry); |
537
|
|
|
} else { // Show list: |
538
|
|
|
|
539
|
|
|
// If either id or set id, show list: |
540
|
|
|
if ($this->pObj->id || $showSetId) { |
541
|
|
|
if ($this->pObj->id) { |
542
|
|
|
// Drawing tree: |
543
|
|
|
$tree = GeneralUtility::makeInstance(PageTreeView::class); |
544
|
|
|
$perms_clause = $GLOBALS['BE_USER']->getPagePermsClause(1); |
545
|
|
|
$tree->init('AND ' . $perms_clause); |
546
|
|
|
|
547
|
|
|
// Set root row: |
548
|
|
|
$HTML = IconUtility::getIconForRecord('pages', $this->pObj->pageinfo); |
|
|
|
|
549
|
|
|
$tree->tree[] = [ |
550
|
|
|
'row' => $this->pObj->pageinfo, |
551
|
|
|
'HTML' => $HTML |
552
|
|
|
]; |
553
|
|
|
|
554
|
|
|
// Get branch beneath: |
555
|
|
|
if ($this->pObj->MOD_SETTINGS['depth']) { |
556
|
|
|
$tree->getTree($this->pObj->id, $this->pObj->MOD_SETTINGS['depth'], ''); |
557
|
|
|
} |
558
|
|
|
|
559
|
|
|
// Traverse page tree: |
560
|
|
|
$code = ''; |
561
|
|
|
$count = 0; |
562
|
|
|
foreach ($tree->tree as $data) { |
563
|
|
|
$code .= $this->drawLog_addRows( |
564
|
|
|
$data['row'], |
565
|
|
|
$data['HTML'] . BackendUtility::getRecordTitle('pages', $data['row'], true), |
566
|
|
|
intval($this->pObj->MOD_SETTINGS['itemsPerPage']) |
567
|
|
|
); |
568
|
|
|
if (++$count == 1000) { |
569
|
|
|
break; |
570
|
|
|
} |
571
|
|
|
} |
572
|
|
|
} else { |
573
|
|
|
$code = ''; |
574
|
|
|
$code .= $this->drawLog_addRows( |
575
|
|
|
$showSetId, |
|
|
|
|
576
|
|
|
'Set ID: ' . $showSetId |
577
|
|
|
); |
578
|
|
|
} |
579
|
|
|
|
580
|
|
|
if ($code) { |
581
|
|
|
$output .= ' |
582
|
|
|
<br /><br /> |
583
|
|
|
<input type="submit" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.reloadlist') . '" name="_reload" /> |
584
|
|
|
<input type="submit" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.downloadcsv') . '" name="_csv" /> |
585
|
|
|
<input type="submit" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.flushvisiblequeue') . '" name="_flush" onclick="return confirm(\'' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.confirmyouresure') . '\');" /> |
586
|
|
|
<input type="submit" value="' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.flushfullqueue') . '" name="_flush_all" onclick="return confirm(\'' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.confirmyouresure') . '\');" /> |
587
|
|
|
<input type="hidden" value="' . $this->pObj->id . '" name="id" /> |
588
|
|
|
<input type="hidden" value="' . $showSetId . '" name="setID" /> |
589
|
|
|
<br /> |
590
|
|
|
' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.curtime') . ': ' . date('H:i:s', time()) . ' |
591
|
|
|
<br /><br /> |
592
|
|
|
|
593
|
|
|
|
594
|
|
|
<table class="lrPadding c-list crawlerlog">' . |
595
|
|
|
$this->drawLog_printTableHeader() . |
596
|
|
|
$code . |
597
|
|
|
'</table>'; |
598
|
|
|
} |
599
|
|
|
} else { // Otherwise show available sets: |
600
|
|
|
$setList = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows( |
601
|
|
|
'set_id, count(*) as count_value, scheduled', |
602
|
|
|
'tx_crawler_queue', |
603
|
|
|
'', |
604
|
|
|
'set_id, scheduled', |
605
|
|
|
'scheduled DESC' |
606
|
|
|
); |
607
|
|
|
|
608
|
|
|
$code = ' |
609
|
|
|
<tr class="bgColor5 tableheader"> |
610
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.setid') . ':</td> |
611
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.count') . 't:</td> |
612
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.time') . ':</td> |
613
|
|
|
</tr> |
614
|
|
|
'; |
615
|
|
|
|
616
|
|
|
$cc = 0; |
617
|
|
|
foreach ($setList as $set) { |
618
|
|
|
$code .= ' |
619
|
|
|
<tr class="bgColor' . ($cc % 2 ? '-20' : '-10') . '"> |
620
|
|
|
<td><a href="' . htmlspecialchars('index.php?setID=' . $set['set_id']) . '">' . $set['set_id'] . '</a></td> |
621
|
|
|
<td>' . $set['count_value'] . '</td> |
622
|
|
|
<td>' . BackendUtility::dateTimeAge($set['scheduled']) . '</td> |
623
|
|
|
</tr> |
624
|
|
|
'; |
625
|
|
|
|
626
|
|
|
$cc++; |
627
|
|
|
} |
628
|
|
|
|
629
|
|
|
$output .= ' |
630
|
|
|
<br /><br /> |
631
|
|
|
<table class="lrPadding c-list">' . |
632
|
|
|
$code . |
633
|
|
|
'</table>'; |
634
|
|
|
} |
635
|
|
|
} |
636
|
|
|
|
637
|
|
|
if ($this->CSVExport) { |
638
|
|
|
$this->outputCsvFile(); |
639
|
|
|
} |
640
|
|
|
|
641
|
|
|
// Return output |
642
|
|
|
return $output; |
643
|
|
|
} |
644
|
|
|
|
645
|
|
|
/** |
646
|
|
|
* Outputs the CSV file and sets the correct headers |
647
|
|
|
*/ |
648
|
|
|
protected function outputCsvFile() |
649
|
|
|
{ |
650
|
|
|
if (!count($this->CSVaccu)) { |
651
|
|
|
$this->addWarningMessage($GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:message.canNotExportEmptyQueueToCsvText')); |
652
|
|
|
return; |
653
|
|
|
} |
654
|
|
|
|
655
|
|
|
$csvLines = []; |
656
|
|
|
|
657
|
|
|
// Field names: |
658
|
|
|
reset($this->CSVaccu); |
659
|
|
|
$fieldNames = array_keys(current($this->CSVaccu)); |
660
|
|
|
$csvLines[] = GeneralUtility::csvValues($fieldNames); |
661
|
|
|
|
662
|
|
|
// Data: |
663
|
|
|
foreach ($this->CSVaccu as $row) { |
664
|
|
|
$csvLines[] = GeneralUtility::csvValues($row); |
665
|
|
|
} |
666
|
|
|
|
667
|
|
|
// Creating output header: |
668
|
|
|
$mimeType = 'application/octet-stream'; |
669
|
|
|
Header('Content-Type: ' . $mimeType); |
670
|
|
|
Header('Content-Disposition: attachment; filename=CrawlerLog.csv'); |
671
|
|
|
|
672
|
|
|
// Printing the content of the CSV lines: |
673
|
|
|
echo implode(chr(13) . chr(10), $csvLines); |
674
|
|
|
|
675
|
|
|
// Exits: |
676
|
|
|
exit; |
|
|
|
|
677
|
|
|
} |
678
|
|
|
|
679
|
|
|
/** |
680
|
|
|
* Create the rows for display of the page tree |
681
|
|
|
* For each page a number of rows are shown displaying GET variable configuration |
682
|
|
|
* |
683
|
|
|
* @param array $pageRow_setId Page row or set-id |
684
|
|
|
* @param string $titleString Title string |
685
|
|
|
* @param int $itemsPerPage Items per Page setting |
686
|
|
|
* |
687
|
|
|
* @return string HTML <tr> content (one or more) |
688
|
|
|
*/ |
689
|
|
|
public function drawLog_addRows($pageRow_setId, $titleString, $itemsPerPage = 10) |
690
|
|
|
{ |
691
|
|
|
|
692
|
|
|
// If Flush button is pressed, flush tables instead of selecting entries: |
693
|
|
|
|
694
|
|
|
if (GeneralUtility::_POST('_flush')) { |
695
|
|
|
$doFlush = true; |
696
|
|
|
$doFullFlush = false; |
697
|
|
|
} elseif (GeneralUtility::_POST('_flush_all')) { |
698
|
|
|
$doFlush = true; |
699
|
|
|
$doFullFlush = true; |
700
|
|
|
} else { |
701
|
|
|
$doFlush = false; |
702
|
|
|
$doFullFlush = false; |
703
|
|
|
} |
704
|
|
|
|
705
|
|
|
// Get result: |
706
|
|
|
if (is_array($pageRow_setId)) { |
707
|
|
|
$res = $this->crawlerController->getLogEntriesForPageId($pageRow_setId['uid'], $this->pObj->MOD_SETTINGS['log_display'], $doFlush, $doFullFlush, intval($itemsPerPage)); |
708
|
|
|
} else { |
709
|
|
|
$res = $this->crawlerController->getLogEntriesForSetId($pageRow_setId, $this->pObj->MOD_SETTINGS['log_display'], $doFlush, $doFullFlush, intval($itemsPerPage)); |
710
|
|
|
} |
711
|
|
|
|
712
|
|
|
// Init var: |
713
|
|
|
$colSpan = 9 |
714
|
|
|
+ ($this->pObj->MOD_SETTINGS['log_resultLog'] ? -1 : 0) |
715
|
|
|
+ ($this->pObj->MOD_SETTINGS['log_feVars'] ? 3 : 0); |
716
|
|
|
|
717
|
|
|
if (count($res)) { |
718
|
|
|
// Traverse parameter combinations: |
719
|
|
|
$c = 0; |
720
|
|
|
$content = ''; |
721
|
|
|
foreach ($res as $kk => $vv) { |
|
|
|
|
722
|
|
|
|
723
|
|
|
// Title column: |
724
|
|
|
if (!$c) { |
725
|
|
|
$titleClm = '<td rowspan="' . count($res) . '">' . $titleString . '</td>'; |
726
|
|
|
} else { |
727
|
|
|
$titleClm = ''; |
728
|
|
|
} |
729
|
|
|
|
730
|
|
|
// Result: |
731
|
|
|
$resLog = $this->getResultLog($vv); |
732
|
|
|
|
733
|
|
|
$resStatus = $this->getResStatus($vv); |
734
|
|
|
$resFeVars = $this->getResFeVars($vv); |
735
|
|
|
|
736
|
|
|
// Compile row: |
737
|
|
|
$parameters = unserialize($vv['parameters']); |
738
|
|
|
|
739
|
|
|
// Put data into array: |
740
|
|
|
$rowData = []; |
741
|
|
|
if ($this->pObj->MOD_SETTINGS['log_resultLog']) { |
742
|
|
|
$rowData['result_log'] = $resLog; |
743
|
|
|
} else { |
744
|
|
|
$rowData['scheduled'] = ($vv['scheduled'] > 0) ? BackendUtility::datetime($vv['scheduled']) : ' ' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.immediate'); |
745
|
|
|
$rowData['exec_time'] = $vv['exec_time'] ? BackendUtility::datetime($vv['exec_time']) : '-'; |
746
|
|
|
} |
747
|
|
|
$rowData['result_status'] = GeneralUtility::fixed_lgd_cs($resStatus, 50); |
748
|
|
|
$rowData['url'] = '<a href="' . htmlspecialchars($parameters['url']) . '" target="_newWIndow">' . htmlspecialchars($parameters['url']) . '</a>'; |
749
|
|
|
$rowData['feUserGroupList'] = $parameters['feUserGroupList']; |
750
|
|
|
$rowData['procInstructions'] = is_array($parameters['procInstructions']) ? implode('; ', $parameters['procInstructions']) : ''; |
751
|
|
|
$rowData['set_id'] = $vv['set_id']; |
752
|
|
|
|
753
|
|
|
if ($this->pObj->MOD_SETTINGS['log_feVars']) { |
754
|
|
|
$rowData['tsfe_id'] = $resFeVars['id']; |
755
|
|
|
$rowData['tsfe_gr_list'] = $resFeVars['gr_list']; |
756
|
|
|
$rowData['tsfe_no_cache'] = $resFeVars['no_cache']; |
757
|
|
|
} |
758
|
|
|
|
759
|
|
|
$setId = intval(GeneralUtility::_GP('setID')); |
760
|
|
|
$refreshIcon = IconUtility::getIcon('actions-system-refresh', Icon::SIZE_SMALL); |
761
|
|
|
|
762
|
|
|
// Put rows together: |
763
|
|
|
$content .= ' |
764
|
|
|
<tr class="bgColor' . ($c % 2 ? '-20' : '-10') . '"> |
765
|
|
|
' . $titleClm . ' |
766
|
|
|
<td><a href="' . $this->getModuleUrl(['qid_details' => $vv['qid'], 'setID' => $setId]) . '">' . htmlspecialchars($vv['qid']) . '</a></td> |
767
|
|
|
<td><a href="' . $this->getModuleUrl(['qid_read' => $vv['qid'], 'setID' => $setId]) . '">' . $refreshIcon . '</a></td>'; |
768
|
|
|
foreach ($rowData as $fKey => $value) { |
769
|
|
|
if (GeneralUtility::inList('url', $fKey)) { |
770
|
|
|
$content .= ' |
771
|
|
|
<td>' . $value . '</td>'; |
772
|
|
|
} else { |
773
|
|
|
$content .= ' |
774
|
|
|
<td>' . nl2br(htmlspecialchars($value)) . '</td>'; |
775
|
|
|
} |
776
|
|
|
} |
777
|
|
|
$content .= ' |
778
|
|
|
</tr>'; |
779
|
|
|
$c++; |
780
|
|
|
|
781
|
|
|
if ($this->CSVExport) { |
782
|
|
|
// Only for CSV (adding qid and scheduled/exec_time if needed): |
783
|
|
|
$rowData['result_log'] = implode('// ', explode(chr(10), $resLog)); |
784
|
|
|
$rowData['qid'] = $vv['qid']; |
785
|
|
|
$rowData['scheduled'] = BackendUtility::datetime($vv['scheduled']); |
786
|
|
|
$rowData['exec_time'] = $vv['exec_time'] ? BackendUtility::datetime($vv['exec_time']) : '-'; |
787
|
|
|
$this->CSVaccu[] = $rowData; |
788
|
|
|
} |
789
|
|
|
} |
790
|
|
|
} else { |
791
|
|
|
|
792
|
|
|
// Compile row: |
793
|
|
|
$content = ' |
794
|
|
|
<tr class="bgColor-20"> |
795
|
|
|
<td>' . $titleString . '</td> |
796
|
|
|
<td colspan="' . $colSpan . '"><em>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.noentries') . '</em></td> |
797
|
|
|
</tr>'; |
798
|
|
|
} |
799
|
|
|
|
800
|
|
|
return $content; |
801
|
|
|
} |
802
|
|
|
|
803
|
|
|
/** |
804
|
|
|
* Find Fe vars |
805
|
|
|
* |
806
|
|
|
* @param array $row |
807
|
|
|
* @return array |
808
|
|
|
*/ |
809
|
|
|
public function getResFeVars($row) |
810
|
|
|
{ |
811
|
|
|
$feVars = []; |
812
|
|
|
|
813
|
|
|
if ($row['result_data']) { |
814
|
|
|
$resultData = unserialize($row['result_data']); |
815
|
|
|
$requestResult = unserialize($resultData['content']); |
816
|
|
|
$feVars = $requestResult['vars']; |
817
|
|
|
} |
818
|
|
|
|
819
|
|
|
return $feVars; |
820
|
|
|
} |
821
|
|
|
|
822
|
|
|
/** |
823
|
|
|
* Create Table header row (log) |
824
|
|
|
* |
825
|
|
|
* @return string Table header |
826
|
|
|
*/ |
827
|
|
|
public function drawLog_printTableHeader() |
828
|
|
|
{ |
829
|
|
|
$content = ' |
830
|
|
|
<tr class="bgColor5 tableheader"> |
831
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.pagetitle') . ':</td> |
832
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.qid') . ':</td> |
833
|
|
|
<td> </td>' . |
834
|
|
|
($this->pObj->MOD_SETTINGS['log_resultLog'] ? ' |
835
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.resultlog') . ':</td>' : ' |
836
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.scheduledtime') . ':</td> |
837
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.runtime') . ':</td>') . ' |
838
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.status') . ':</td> |
839
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.url') . ':</td> |
840
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.groups') . ':</td> |
841
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.procinstr') . ':</td> |
842
|
|
|
<td>' . $GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.setid') . ':</td>' . |
843
|
|
|
($this->pObj->MOD_SETTINGS['log_feVars'] ? ' |
844
|
|
|
<td>' . htmlspecialchars('TSFE->id') . '</td> |
845
|
|
|
<td>' . htmlspecialchars('TSFE->gr_list') . '</td> |
846
|
|
|
<td>' . htmlspecialchars('TSFE->no_cache') . '</td>' : '') . ' |
847
|
|
|
</tr>'; |
848
|
|
|
|
849
|
|
|
return $content; |
850
|
|
|
} |
851
|
|
|
|
852
|
|
|
/** |
853
|
|
|
* Extract the log information from the current row and retrive it as formatted string. |
854
|
|
|
* |
855
|
|
|
* @param array $resultRow |
856
|
|
|
* |
857
|
|
|
* @return string |
858
|
|
|
*/ |
859
|
|
|
protected function getResultLog($resultRow) |
860
|
|
|
{ |
861
|
|
|
$content = ''; |
862
|
|
|
|
863
|
|
|
if (is_array($resultRow) && array_key_exists('result_data', $resultRow)) { |
864
|
|
|
$requestContent = unserialize($resultRow['result_data']); |
865
|
|
|
$requestResult = unserialize($requestContent['content']); |
866
|
|
|
|
867
|
|
|
if (is_array($requestResult) && array_key_exists('log', $requestResult)) { |
868
|
|
|
$content = implode(chr(10), $requestResult['log']); |
869
|
|
|
} |
870
|
|
|
} |
871
|
|
|
|
872
|
|
|
return $content; |
873
|
|
|
} |
874
|
|
|
|
875
|
|
|
public function getResStatus($vv) |
876
|
|
|
{ |
877
|
|
|
if ($vv['result_data']) { |
878
|
|
|
$requestContent = unserialize($vv['result_data']); |
879
|
|
|
$requestResult = unserialize($requestContent['content']); |
880
|
|
|
if (is_array($requestResult)) { |
881
|
|
|
if (empty($requestResult['errorlog'])) { |
882
|
|
|
$resStatus = 'OK'; |
883
|
|
|
} else { |
884
|
|
|
$resStatus = implode("\n", $requestResult['errorlog']); |
885
|
|
|
} |
886
|
|
|
} else { |
887
|
|
|
$resStatus = 'Error: ' . substr(preg_replace('/\s+/', ' ', strip_tags($requestContent['content'])), 0, 10000) . '...'; |
888
|
|
|
} |
889
|
|
|
} else { |
890
|
|
|
$resStatus = '-'; |
891
|
|
|
} |
892
|
|
|
return $resStatus; |
893
|
|
|
} |
894
|
|
|
|
895
|
|
|
/***************************** |
896
|
|
|
* |
897
|
|
|
* CLI status display |
898
|
|
|
* |
899
|
|
|
*****************************/ |
900
|
|
|
|
901
|
|
|
/** |
902
|
|
|
* This method is used to show an overview about the active an the finished crawling processes |
903
|
|
|
* |
904
|
|
|
* @param void |
905
|
|
|
* @return string |
906
|
|
|
*/ |
907
|
|
|
protected function drawProcessOverviewAction() |
908
|
|
|
{ |
909
|
|
|
$this->runRefreshHooks(); |
910
|
|
|
|
911
|
|
|
global $BACK_PATH; |
912
|
|
|
$this->makeCrawlerProcessableChecks(); |
913
|
|
|
|
914
|
|
|
$crawler = $this->findCrawler(); |
915
|
|
|
try { |
916
|
|
|
$this->handleProcessOverviewActions(); |
917
|
|
|
} catch (\Exception $e) { |
918
|
|
|
$this->addErrorMessage($e->getMessage()); |
919
|
|
|
} |
920
|
|
|
|
921
|
|
|
$offset = intval(GeneralUtility::_GP('offset')); |
922
|
|
|
$perpage = 20; |
923
|
|
|
|
924
|
|
|
$processRepository = new ProcessRepository(); |
925
|
|
|
$queueRepository = new QueueRepository(); |
926
|
|
|
|
927
|
|
|
$mode = $this->pObj->MOD_SETTINGS['processListMode']; |
928
|
|
|
$where = ''; |
929
|
|
|
if ($mode == 'simple') { |
930
|
|
|
$where = 'active = 1'; |
931
|
|
|
} |
932
|
|
|
|
933
|
|
|
$allProcesses = $processRepository->findAll('ttl', 'DESC', $perpage, $offset, $where); |
934
|
|
|
$allCount = $processRepository->countAll($where); |
935
|
|
|
|
936
|
|
|
$listView = new ProcessListView(); |
937
|
|
|
$listView->setPageId($this->pObj->id); |
938
|
|
|
$listView->setIconPath($BACK_PATH . '../typo3conf/ext/crawler/template/process/res/img/'); |
939
|
|
|
$listView->setProcessCollection($allProcesses); |
940
|
|
|
$listView->setCliPath($this->processManager->getCrawlerCliPath()); |
941
|
|
|
$listView->setIsCrawlerEnabled(!$crawler->getDisabled() && !$this->isErrorDetected); |
942
|
|
|
$listView->setTotalUnprocessedItemCount($queueRepository->countAllPendingItems()); |
943
|
|
|
$listView->setAssignedUnprocessedItemCount($queueRepository->countAllAssignedPendingItems()); |
944
|
|
|
$listView->setActiveProcessCount($processRepository->countActive()); |
945
|
|
|
$listView->setMaxActiveProcessCount(MathUtility::forceIntegerInRange($this->extensionSettings['processLimit'], 1, 99, 1)); |
946
|
|
|
$listView->setMode($mode); |
947
|
|
|
|
948
|
|
|
$paginationView = new PaginationView(); |
949
|
|
|
$paginationView->setCurrentOffset($offset); |
950
|
|
|
$paginationView->setPerPage($perpage); |
951
|
|
|
$paginationView->setTotalItemCount($allCount); |
952
|
|
|
|
953
|
|
|
$output = $listView->render(); |
954
|
|
|
|
955
|
|
|
if ($paginationView->getTotalPagesCount() > 1) { |
956
|
|
|
$output .= ' <br />' . $paginationView->render(); |
957
|
|
|
} |
958
|
|
|
|
959
|
|
|
return $output; |
960
|
|
|
} |
961
|
|
|
|
962
|
|
|
/** |
963
|
|
|
* Verify that the crawler is exectuable. |
964
|
|
|
* |
965
|
|
|
* @return void |
966
|
|
|
*/ |
967
|
|
|
protected function makeCrawlerProcessableChecks() |
968
|
|
|
{ |
969
|
|
|
global $LANG; |
970
|
|
|
|
971
|
|
|
if ($this->isCrawlerUserAvailable() === false) { |
972
|
|
|
$this->addErrorMessage($LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:message.noBeUserAvailable')); |
973
|
|
|
} elseif ($this->isCrawlerUserNotAdmin() === false) { |
974
|
|
|
$this->addErrorMessage($LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:message.beUserIsAdmin')); |
975
|
|
|
} |
976
|
|
|
|
977
|
|
|
if ($this->isPhpForkAvailable() === false) { |
978
|
|
|
$this->addErrorMessage($LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:message.noPhpForkAvailable')); |
979
|
|
|
} |
980
|
|
|
|
981
|
|
|
$exitCode = 0; |
982
|
|
|
$out = []; |
983
|
|
|
exec(escapeshellcmd($this->extensionSettings['phpPath'] . ' -v'), $out, $exitCode); |
984
|
|
|
if ($exitCode > 0) { |
985
|
|
|
$this->addErrorMessage(sprintf($LANG->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:message.phpBinaryNotFound'), htmlspecialchars($this->extensionSettings['phpPath']))); |
986
|
|
|
} |
987
|
|
|
} |
988
|
|
|
|
989
|
|
|
/** |
990
|
|
|
* Indicate that the required PHP method "popen" is |
991
|
|
|
* available in the system. |
992
|
|
|
* |
993
|
|
|
* @return boolean |
994
|
|
|
*/ |
995
|
|
|
protected function isPhpForkAvailable() |
996
|
|
|
{ |
997
|
|
|
return function_exists('popen'); |
998
|
|
|
} |
999
|
|
|
|
1000
|
|
|
/** |
1001
|
|
|
* Indicate that the required be_user "_cli_crawler" is |
1002
|
|
|
* global available in the system. |
1003
|
|
|
* |
1004
|
|
|
* @return boolean |
1005
|
|
|
*/ |
1006
|
|
|
protected function isCrawlerUserAvailable() |
1007
|
|
|
{ |
1008
|
|
|
$isAvailable = false; |
1009
|
|
|
$userArray = BackendUtility::getRecordsByField('be_users', 'username', '_cli_crawler'); |
1010
|
|
|
|
1011
|
|
|
if (is_array($userArray)) { |
1012
|
|
|
$isAvailable = true; |
1013
|
|
|
} |
1014
|
|
|
|
1015
|
|
|
return $isAvailable; |
1016
|
|
|
} |
1017
|
|
|
|
1018
|
|
|
/** |
1019
|
|
|
* Indicate that the required be_user "_cli_crawler" is |
1020
|
|
|
* has no admin rights. |
1021
|
|
|
* |
1022
|
|
|
* @return boolean |
1023
|
|
|
*/ |
1024
|
|
|
protected function isCrawlerUserNotAdmin() |
1025
|
|
|
{ |
1026
|
|
|
$isAvailable = false; |
1027
|
|
|
$userArray = BackendUtility::getRecordsByField('be_users', 'username', '_cli_crawler'); |
1028
|
|
|
|
1029
|
|
|
if (is_array($userArray) && $userArray[0]['admin'] == 0) { |
1030
|
|
|
$isAvailable = true; |
1031
|
|
|
} |
1032
|
|
|
|
1033
|
|
|
return $isAvailable; |
1034
|
|
|
} |
1035
|
|
|
|
1036
|
|
|
/** |
1037
|
|
|
* Method to handle incomming actions of the process overview |
1038
|
|
|
* |
1039
|
|
|
* @throws \Exception |
1040
|
|
|
* |
1041
|
|
|
* @return void |
1042
|
|
|
*/ |
1043
|
|
|
protected function handleProcessOverviewActions() |
1044
|
|
|
{ |
1045
|
|
|
$crawler = $this->findCrawler(); |
1046
|
|
|
|
1047
|
|
|
switch (GeneralUtility::_GP('action')) { |
1048
|
|
|
case 'stopCrawling': |
1049
|
|
|
//set the cli status to disable (all processes will be terminated) |
1050
|
|
|
$crawler->setDisabled(true); |
1051
|
|
|
break; |
1052
|
|
|
case 'resumeCrawling': |
1053
|
|
|
//set the cli status to end (all processes will be terminated) |
1054
|
|
|
$crawler->setDisabled(false); |
1055
|
|
|
break; |
1056
|
|
|
case 'addProcess': |
1057
|
|
|
$handle = $this->processManager->startProcess(); |
1058
|
|
|
if ($handle === false) { |
1059
|
|
|
throw new \Exception($GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.newprocesserror')); |
1060
|
|
|
} |
1061
|
|
|
$this->addNoticeMessage($GLOBALS['LANG']->sL('LLL:EXT:crawler/Resources/Private/Language/locallang.xml:labels.newprocess')); |
1062
|
|
|
break; |
1063
|
|
|
} |
1064
|
|
|
} |
1065
|
|
|
|
1066
|
|
|
/** |
1067
|
|
|
* Returns the singleton instance of the crawler. |
1068
|
|
|
* |
1069
|
|
|
* @param void |
1070
|
|
|
* @return CrawlerController crawler object |
1071
|
|
|
*/ |
1072
|
|
|
protected function findCrawler() |
1073
|
|
|
{ |
1074
|
|
|
if (!$this->crawlerController instanceof CrawlerController) { |
1075
|
|
|
$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class); |
1076
|
|
|
} |
1077
|
|
|
return $this->crawlerController; |
1078
|
|
|
} |
1079
|
|
|
|
1080
|
|
|
/***************************** |
1081
|
|
|
* |
1082
|
|
|
* General Helper Functions |
1083
|
|
|
* |
1084
|
|
|
*****************************/ |
1085
|
|
|
|
1086
|
|
|
/** |
1087
|
|
|
* This method is used to add a message to the internal queue |
1088
|
|
|
* |
1089
|
|
|
* @param string the message itself |
1090
|
|
|
* @param integer message level (-1 = success (default), 0 = info, 1 = notice, 2 = warning, 3 = error) |
1091
|
|
|
* |
1092
|
|
|
* @return void |
1093
|
|
|
*/ |
1094
|
|
|
private function addMessage($message, $severity = FlashMessage::OK) |
1095
|
|
|
{ |
1096
|
|
|
$message = GeneralUtility::makeInstance( |
1097
|
|
|
FlashMessage::class, |
1098
|
|
|
$message, |
1099
|
|
|
'', |
1100
|
|
|
$severity |
1101
|
|
|
); |
1102
|
|
|
|
1103
|
|
|
// TODO: |
1104
|
|
|
/** @var FlashMessageService $flashMessageService */ |
1105
|
|
|
$flashMessageService = GeneralUtility::makeInstance(FlashMessageService::class); |
1106
|
|
|
$flashMessageService->getMessageQueueByIdentifier()->addMessage($message); |
1107
|
|
|
} |
1108
|
|
|
|
1109
|
|
|
/** |
1110
|
|
|
* Add notice message to the user interface. |
1111
|
|
|
* |
1112
|
|
|
* @param string The message |
1113
|
|
|
* |
1114
|
|
|
* @return void |
1115
|
|
|
*/ |
1116
|
|
|
protected function addNoticeMessage($message) |
1117
|
|
|
{ |
1118
|
|
|
$this->addMessage($message, FlashMessage::NOTICE); |
1119
|
|
|
} |
1120
|
|
|
|
1121
|
|
|
/** |
1122
|
|
|
* Add error message to the user interface. |
1123
|
|
|
* |
1124
|
|
|
* @param string The message |
1125
|
|
|
* |
1126
|
|
|
* @return void |
1127
|
|
|
*/ |
1128
|
|
|
protected function addErrorMessage($message) |
1129
|
|
|
{ |
1130
|
|
|
$this->isErrorDetected = true; |
1131
|
|
|
$this->addMessage($message, FlashMessage::ERROR); |
1132
|
|
|
} |
1133
|
|
|
|
1134
|
|
|
/** |
1135
|
|
|
* Add error message to the user interface. |
1136
|
|
|
* |
1137
|
|
|
* NOTE: |
1138
|
|
|
* This method is basesd on TYPO3 4.3 or higher! |
1139
|
|
|
* |
1140
|
|
|
* @param string The message |
1141
|
|
|
* |
1142
|
|
|
* @return void |
1143
|
|
|
*/ |
1144
|
|
|
protected function addWarningMessage($message) |
1145
|
|
|
{ |
1146
|
|
|
$this->addMessage($message, FlashMessage::WARNING); |
1147
|
|
|
} |
1148
|
|
|
|
1149
|
|
|
/** |
1150
|
|
|
* Create selector box |
1151
|
|
|
* |
1152
|
|
|
* @param array $optArray Options key(value) => label pairs |
1153
|
|
|
* @param string $name Selector box name |
1154
|
|
|
* @param string $value Selector box value (array for multiple...) |
1155
|
|
|
* @param boolean $multiple If set, will draw multiple box. |
1156
|
|
|
* |
1157
|
|
|
* @return string HTML select element |
1158
|
|
|
*/ |
1159
|
|
|
public function selectorBox($optArray, $name, $value, $multiple) |
1160
|
|
|
{ |
1161
|
|
|
$options = []; |
1162
|
|
|
foreach ($optArray as $key => $val) { |
1163
|
|
|
$options[] = ' |
1164
|
|
|
<option value="' . htmlspecialchars($key) . '"' . ((!$multiple && !strcmp($value, $key)) || ($multiple && in_array($key, (array)$value)) ? ' selected="selected"' : '') . '>' . htmlspecialchars($val) . '</option>'; |
1165
|
|
|
} |
1166
|
|
|
|
1167
|
|
|
$output = '<select name="' . htmlspecialchars($name . ($multiple ? '[]' : '')) . '"' . ($multiple ? ' multiple="multiple" size="' . count($options) . '"' : '') . '>' . implode('', $options) . '</select>'; |
1168
|
|
|
|
1169
|
|
|
return $output; |
1170
|
|
|
} |
1171
|
|
|
|
1172
|
|
|
/** |
1173
|
|
|
* Activate hooks |
1174
|
|
|
* |
1175
|
|
|
* @return void |
1176
|
|
|
*/ |
1177
|
|
|
public function runRefreshHooks() |
1178
|
|
|
{ |
1179
|
|
|
$crawlerLib = GeneralUtility::makeInstance(CrawlerController::class); |
1180
|
|
|
if (is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['refresh_hooks'])) { |
1181
|
|
|
foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['refresh_hooks'] as $objRef) { |
1182
|
|
|
$hookObj = &GeneralUtility::getUserObj($objRef); |
1183
|
|
|
if (is_object($hookObj)) { |
1184
|
|
|
$hookObj->crawler_init($crawlerLib); |
1185
|
|
|
} |
1186
|
|
|
} |
1187
|
|
|
} |
1188
|
|
|
} |
1189
|
|
|
|
1190
|
|
|
/** |
1191
|
|
|
* Returns the URL to the current module, including $_GET['id']. |
1192
|
|
|
* |
1193
|
|
|
* @param array $urlParameters optional parameters to add to the URL |
1194
|
|
|
* @return string |
1195
|
|
|
*/ |
1196
|
|
|
protected function getModuleUrl(array $urlParameters = []) |
1197
|
|
|
{ |
1198
|
|
|
if ($this->pObj->id) { |
1199
|
|
|
$urlParameters = array_merge($urlParameters, [ |
1200
|
|
|
'id' => $this->pObj->id |
1201
|
|
|
]); |
1202
|
|
|
} |
1203
|
|
|
return BackendUtility::getModuleUrl(GeneralUtility::_GP('M'), $urlParameters); |
|
|
|
|
1204
|
|
|
} |
1205
|
|
|
} |
1206
|
|
|
|
This check looks for calls to methods that do not seem to exist on a given type. It looks for the method on the type itself as well as in inherited classes or implemented interfaces.
This is most likely a typographical error or the method has been renamed.