1
|
|
|
<?php |
2
|
|
|
namespace AOE\Crawler\Command; |
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\Controller\CrawlerController; |
29
|
|
|
use AOE\Crawler\Domain\Model\Reason; |
30
|
|
|
use AOE\Crawler\Domain\Repository\QueueRepository; |
31
|
|
|
use AOE\Crawler\Event\EventDispatcher; |
32
|
|
|
use Helhum\Typo3Console\Mvc\Controller\CommandController; |
33
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
34
|
|
|
use TYPO3\CMS\Core\Utility\MathUtility; |
35
|
|
|
use TYPO3\CMS\Extbase\Object\ObjectManager; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* Class CrawlerCommandController |
39
|
|
|
*/ |
40
|
|
|
class CrawlerCommandController extends CommandController |
41
|
|
|
{ |
42
|
|
|
const CLI_STATUS_NOTHING_PROCCESSED = 0; |
43
|
|
|
const CLI_STATUS_REMAIN = 1; //queue not empty |
44
|
|
|
const CLI_STATUS_PROCESSED = 2; //(some) queue items where processed |
45
|
|
|
const CLI_STATUS_ABORTED = 4; //instance didn't finish |
46
|
|
|
const CLI_STATUS_POLLABLE_PROCESSED = 8; |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* Crawler Command - Cleaning up the queue. |
50
|
|
|
* |
51
|
|
|
* Works as a CLI interface to some functionality from the Web > Info > Site Crawler module; |
52
|
|
|
* It will remove queue entries and perform a cleanup. |
53
|
|
|
* |
54
|
|
|
* Examples: |
55
|
|
|
* |
56
|
|
|
* --- Remove all finished queue-entries in the sub-branch of page 5 |
57
|
|
|
* $ typo3cms crawler:flushqueue --mode finished --page 5 |
58
|
|
|
* |
59
|
|
|
* --- Remove all pending queue-entries for all pages |
60
|
|
|
* $ typo3cms crawler:flushqueue --mode pending |
61
|
|
|
* |
62
|
|
|
* @param string $mode Output mode: "finished", "all", "pending"', "Specifies the type queue entries which is flushed in the process." |
63
|
|
|
* @param int $page Page to start clearing the queue recursively, 0 is default and clears all. |
64
|
|
|
* |
65
|
|
|
*/ |
66
|
|
|
public function flushQueueCommand($mode = 'finished', $page = 0) |
67
|
|
|
{ |
68
|
|
|
/** @var CrawlerController $crawlerController */ |
69
|
|
|
$crawlerController = $this->objectManager->get(CrawlerController::class); |
70
|
|
|
|
71
|
|
|
$pageId = MathUtility::forceIntegerInRange($page, 0); |
72
|
|
|
$fullFlush = ($pageId == 0); |
73
|
|
|
|
74
|
|
|
switch ($mode) { |
75
|
|
|
case 'all': |
76
|
|
|
$crawlerController->getLogEntriesForPageId($pageId, '', true, $fullFlush); |
77
|
|
|
break; |
78
|
|
|
case 'finished': |
79
|
|
|
case 'pending': |
80
|
|
|
$crawlerController->getLogEntriesForPageId($pageId, $mode, true, $fullFlush); |
81
|
|
|
break; |
82
|
|
|
default: |
83
|
|
|
$this->outputLine('<info>No matching parameters found.' . PHP_EOL . 'Try "typo3cms help crawler:flushqueue" to see your options</info>'); |
84
|
|
|
break; |
85
|
|
|
|
86
|
|
|
} |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* Crawler Command - Submitting URLs to be crawled. |
91
|
|
|
* |
92
|
|
|
* Works as a CLI interface to some functionality from the Web > Info > Site Crawler module; |
93
|
|
|
* It can put entries in the queue from command line options, return the list of URLs and even execute |
94
|
|
|
* all entries right away without having to queue them up - this can be useful for immediate re-cache, |
95
|
|
|
* re-indexing or static publishing from command line. |
96
|
|
|
* |
97
|
|
|
* Examples: |
98
|
|
|
* |
99
|
|
|
* --- Re-cache pages from page 7 and two levels down, executed immediately |
100
|
|
|
* $ typo3cms crawler:buildqueue --startpage 7 --depth 2 --conf <configurationKey> --mode exec |
101
|
|
|
* |
102
|
|
|
* --- Put entries for re-caching pages from page 7 into queue, 4 every minute. |
103
|
|
|
* $ typo3cms crawler:buildqueue --startpage 7 --depth 0 --conf <configurationKey> --number 4 --mode queue |
104
|
|
|
* |
105
|
|
|
* @param int $startpage The page from where the queue building should start. |
106
|
|
|
* @param int $depth Tree depth, 0-99', "How many levels under the 'page_id' to include. |
107
|
|
|
* @param string $mode Output mode: "url", "exec", "queue"', "Specifies output modes url : Will list URLs which wget could use as input. queue: Will put entries in queue table. exec: Will execute all entries right away! |
108
|
|
|
* @param int $number Specifies how many items are put in the queue per minute. Only valid for output mode "queue" |
109
|
|
|
* @param string $conf A comma separated list of crawler configurations |
110
|
|
|
*/ |
111
|
|
|
public function buildQueueCommand($startpage = 0, $depth = 0, $mode = '', $number = 0, $conf = '') |
112
|
|
|
{ |
113
|
|
|
|
114
|
|
|
/** @var CrawlerController $crawlerController */ |
115
|
|
|
$crawlerController = $this->objectManager->get(CrawlerController::class); |
116
|
|
|
|
117
|
|
|
if ($mode === 'exec') { |
118
|
|
|
$crawlerController->registerQueueEntriesInternallyOnly = true; |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
if (defined('TYPO3_MODE') && 'BE' === TYPO3_MODE) { |
122
|
|
|
// Crawler is called over TYPO3 BE |
123
|
|
|
$pageId = 1; |
124
|
|
|
} else { |
125
|
|
|
// Crawler is called over cli |
126
|
|
|
$pageId = MathUtility::forceIntegerInRange($startpage, 0); |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
$configurationKeys = $this->getConfigurationKeys($conf); |
130
|
|
|
|
131
|
|
|
if (!is_array($configurationKeys)) { |
132
|
|
|
$configurations = $crawlerController->getUrlsForPageId($pageId); |
133
|
|
|
if (is_array($configurations)) { |
134
|
|
|
$configurationKeys = array_keys($configurations); |
135
|
|
|
} else { |
136
|
|
|
$configurationKeys = []; |
137
|
|
|
} |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
if ($mode === 'queue' || $mode === 'exec') { |
141
|
|
|
$reason = new Reason(); |
142
|
|
|
$reason->setReason(Reason::REASON_GUI_SUBMIT); |
143
|
|
|
$reason->setDetailText('The cli script of the crawler added to the queue'); |
144
|
|
|
EventDispatcher::getInstance()->post( |
145
|
|
|
'invokeQueueChange', |
146
|
|
|
$crawlerController->setID, |
147
|
|
|
['reason' => $reason] |
148
|
|
|
); |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
if ($crawlerController->extensionSettings['cleanUpOldQueueEntries']) { |
152
|
|
|
$crawlerController->cleanUpOldQueueEntries(); |
153
|
|
|
} |
154
|
|
|
|
155
|
|
|
$crawlerController->setID = (int) GeneralUtility::md5int(microtime()); |
156
|
|
|
$crawlerController->getPageTreeAndUrls( |
157
|
|
|
$pageId, |
158
|
|
|
MathUtility::forceIntegerInRange($depth, 0, 99), |
159
|
|
|
$crawlerController->getCurrentTime(), |
160
|
|
|
MathUtility::forceIntegerInRange($number ? intval($number) : 30, 1, 1000), |
161
|
|
|
$mode === 'queue' || $mode === 'exec', |
162
|
|
|
$mode === 'url', |
163
|
|
|
[], |
164
|
|
|
$configurationKeys |
165
|
|
|
); |
166
|
|
|
|
167
|
|
|
if ($mode === 'url') { |
168
|
|
|
$this->outputLine('<info>' . implode(PHP_EOL, $crawlerController->downloadUrls) . PHP_EOL . '</info>'); |
169
|
|
|
} elseif ($mode === 'exec') { |
170
|
|
|
$this->outputLine('<info>Executing ' . count($crawlerController->urlList) . ' requests right away:</info>'); |
171
|
|
|
$this->outputLine('<info>' . implode(PHP_EOL, $crawlerController->urlList) . '</info>' . PHP_EOL); |
172
|
|
|
$this->outputLine('<info>Processing</info>' . PHP_EOL); |
173
|
|
|
|
174
|
|
|
foreach ($crawlerController->queueEntries as $queueRec) { |
175
|
|
|
$p = unserialize($queueRec['parameters']); |
176
|
|
|
$this->outputLine('<info>' . $p['url'] . ' (' . implode(',', $p['procInstructions']) . ') => ' . '</info>' . PHP_EOL); |
177
|
|
|
$result = $crawlerController->readUrlFromArray($queueRec); |
178
|
|
|
|
179
|
|
|
$requestResult = unserialize($result['content']); |
180
|
|
|
if (is_array($requestResult)) { |
181
|
|
|
$resLog = is_array($requestResult['log']) ? PHP_EOL . chr(9) . chr(9) . implode(PHP_EOL . chr(9) . chr(9), $requestResult['log']) : ''; |
182
|
|
|
$this->outputLine('<info>OK: ' . $resLog . '</info>' . PHP_EOL); |
183
|
|
|
} else { |
184
|
|
|
$this->outputLine('<errror>Error checking Crawler Result: ' . substr(preg_replace('/\s+/', ' ', strip_tags($result['content'])), 0, 30000) . '...' . PHP_EOL . '</errror>' . PHP_EOL); |
185
|
|
|
} |
186
|
|
|
} |
187
|
|
|
} elseif ($mode === 'queue') { |
188
|
|
|
$this->outputLine('<info>Putting ' . count($crawlerController->urlList) . ' entries in queue:</info>' . PHP_EOL); |
189
|
|
|
$this->outputLine('<info>' . implode(PHP_EOL, $crawlerController->urlList) . '</info>' . PHP_EOL); |
190
|
|
|
} else { |
191
|
|
|
$this->outputLine('<info>' . count($crawlerController->urlList) . ' entries found for processing. (Use --mode to decide action):</info>' . PHP_EOL); |
192
|
|
|
$this->outputLine('<info>' . implode(PHP_EOL, $crawlerController->urlList) . '</info>' . PHP_EOL); |
193
|
|
|
} |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
/** |
197
|
|
|
* Crawler Command - Crawling the URLs from the queue |
198
|
|
|
* |
199
|
|
|
* Examples: |
200
|
|
|
* |
201
|
|
|
* --- Will trigger the crawler which starts to process the queue entires |
202
|
|
|
* $ typo3cms crawler:crawlqueue |
203
|
|
|
* |
204
|
|
|
* @param int $amount How many pages should be crawled during that run. |
205
|
|
|
* @param int $sleeptime Amount of milliseconds which the system should use to relax between crawls. |
206
|
|
|
* @param int $sleepafter Amount of seconds which the system should use to relax after all crawls are done. |
207
|
|
|
* |
208
|
|
|
* @return int |
209
|
|
|
*/ |
210
|
|
|
public function crawlQueueCommand($amount = 0, $sleeptime = 0, $sleepafter = 0) |
211
|
|
|
{ |
212
|
|
|
$result = self::CLI_STATUS_NOTHING_PROCCESSED; |
213
|
|
|
|
214
|
|
|
/** @var CrawlerController $crawlerController */ |
215
|
|
|
$crawlerController = $this->objectManager->get(CrawlerController::class); |
216
|
|
|
/** @var QueueRepository $queueRepository */ |
217
|
|
|
$queueRepository = $this->objectManager->get(QueueRepository::class); |
218
|
|
|
|
219
|
|
|
$settings = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['crawler']); |
220
|
|
|
$settings = is_array($settings) ? $settings : []; |
221
|
|
|
$crawlerController->setExtensionSettings($settings); |
222
|
|
|
|
223
|
|
|
if (!$crawlerController->getDisabled() && $crawlerController->CLI_checkAndAcquireNewProcess($crawlerController->CLI_buildProcessId())) { |
224
|
|
|
$countInARun = $amount ? intval($amount) : $crawlerController->extensionSettings['countInARun']; |
225
|
|
|
$sleepAfterFinish = $sleeptime ? intval($sleeptime) : $crawlerController->extensionSettings['sleepAfterFinish']; |
226
|
|
|
$sleepTime = $sleepafter ? intval($sleepafter) : $crawlerController->extensionSettings['sleepTime']; |
227
|
|
|
|
228
|
|
|
try { |
229
|
|
|
// Run process: |
230
|
|
|
$result = $crawlerController->CLI_run($countInARun, $sleepTime, $sleepAfterFinish); |
231
|
|
|
} catch (\Exception $e) { |
232
|
|
|
$this->outputLine('<warning>'. get_class($e) . ': ' . $e->getMessage() . '</warning>'); |
233
|
|
|
$result = self::CLI_STATUS_ABORTED; |
234
|
|
|
} |
235
|
|
|
|
236
|
|
|
// Cleanup |
237
|
|
|
$GLOBALS['TYPO3_DB']->exec_DELETEquery('tx_crawler_process', 'assigned_items_count = 0'); |
238
|
|
|
$crawlerController->CLI_releaseProcesses($crawlerController->CLI_buildProcessId()); |
239
|
|
|
|
240
|
|
|
$this->outputLine('<info>Unprocessed Items remaining:' . $queueRepository->countUnprocessedItems() . ' (' . $crawlerController->CLI_buildProcessId() . ')</info>'); |
241
|
|
|
$result |= ($queueRepository->countUnprocessedItems() > 0 ? self::CLI_STATUS_REMAIN : self::CLI_STATUS_NOTHING_PROCCESSED); |
242
|
|
|
} else { |
243
|
|
|
$result |= self::CLI_STATUS_ABORTED; |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
return $result; |
247
|
|
|
|
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* Obtains configuration keys from the CLI arguments |
252
|
|
|
* |
253
|
|
|
* @param $conf string |
254
|
|
|
* @return array |
255
|
|
|
*/ |
256
|
|
|
private function getConfigurationKeys($conf) |
257
|
|
|
{ |
258
|
|
|
$parameter = trim($conf); |
259
|
|
|
return ($parameter != '' ? GeneralUtility::trimExplode(',', $parameter) : []); |
260
|
|
|
} |
261
|
|
|
} |
262
|
|
|
|