|
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
|
|
|
$crawlerController->getLogEntriesForPageId($pageId, $mode, true, $fullFlush); |
|
80
|
|
|
break; |
|
81
|
|
|
case 'pending': |
|
82
|
|
|
$crawlerController->getLogEntriesForPageId($pageId, $mode, true, $fullFlush); |
|
83
|
|
|
break; |
|
84
|
|
|
default: |
|
85
|
|
|
$this->outputLine("<info>No matching parameters found." . chr(10) . "Try 'typo3cms help crawler:flushqueue' to see your options</info>"); |
|
86
|
|
|
break; |
|
87
|
|
|
|
|
88
|
|
|
} |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* Crawler Command - Submitting URLs to be crawled. |
|
93
|
|
|
* |
|
94
|
|
|
* Works as a CLI interface to some functionality from the Web > Info > Site Crawler module; |
|
95
|
|
|
* It can put entries in the queue from command line options, return the list of URLs and even execute |
|
96
|
|
|
* all entries right away without having to queue them up - this can be useful for immediate re-cache, |
|
97
|
|
|
* re-indexing or static publishing from command line. |
|
98
|
|
|
* |
|
99
|
|
|
* Examples: |
|
100
|
|
|
* |
|
101
|
|
|
* --- Re-cache pages from page 7 and two levels down, executed immediately |
|
102
|
|
|
* $ typo3cms crawler:buildqueue --startpage 7 --depth 2 --conf <configurationKey> --mode exec |
|
103
|
|
|
* |
|
104
|
|
|
* --- Put entries for re-caching pages from page 7 into queue, 4 every minute. |
|
105
|
|
|
* $ typo3cms crawler:buildqueue --startpage 7 --depth 0 --conf <configurationKey> --number 4 --mode queue |
|
106
|
|
|
* |
|
107
|
|
|
* @param int $startpage The page from where the queue building should start. |
|
108
|
|
|
* @param int $depth Tree depth, 0-99', "How many levels under the 'page_id' to include. |
|
109
|
|
|
* @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! |
|
110
|
|
|
* @param int $number Specifies how many items are put in the queue per minute. Only valid for output mode "queue" |
|
111
|
|
|
* @param string $conf A comma separated list of crawler configurations |
|
112
|
|
|
*/ |
|
113
|
|
|
public function buildQueueCommand($startpage = 0, $depth = 0, $mode = '', $number = 0, $conf = '') |
|
114
|
|
|
{ |
|
115
|
|
|
|
|
116
|
|
|
/** @var CrawlerController $crawlerController */ |
|
117
|
|
|
$crawlerController = $this->objectManager->get(CrawlerController::class); |
|
118
|
|
|
|
|
119
|
|
|
// Force user to admin state and set workspace to "Live": |
|
120
|
|
|
// $this->backendUser->user['admin'] = 1; |
|
121
|
|
|
// $this->backendUser->setWorkspace(0); |
|
122
|
|
|
|
|
123
|
|
|
if ($mode === 'exec') { |
|
124
|
|
|
$crawlerController->registerQueueEntriesInternallyOnly = true; |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
if (1===2) { |
|
128
|
|
|
// Crawler is called over TYPO3 BE |
|
129
|
|
|
$pageId = MathUtility::forceIntegerInRange(1, 0); |
|
130
|
|
|
} else { |
|
131
|
|
|
// Crawler is called over cli |
|
132
|
|
|
$pageId = MathUtility::forceIntegerInRange($startpage, 0); |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
$configurationKeys = $this->getConfigurationKeys($conf); |
|
136
|
|
|
|
|
137
|
|
|
if (!is_array($configurationKeys)) { |
|
138
|
|
|
$configurations = $crawlerController->getUrlsForPageId($pageId); |
|
139
|
|
|
if (is_array($configurations)) { |
|
140
|
|
|
$configurationKeys = array_keys($configurations); |
|
141
|
|
|
} else { |
|
142
|
|
|
$configurationKeys = []; |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
|
|
if ($mode === 'queue' || $mode === 'exec') { |
|
147
|
|
|
$reason = new Reason(); |
|
148
|
|
|
$reason->setReason(Reason::REASON_GUI_SUBMIT); |
|
149
|
|
|
$reason->setDetailText('The cli script of the crawler added to the queue'); |
|
150
|
|
|
EventDispatcher::getInstance()->post( |
|
151
|
|
|
'invokeQueueChange', |
|
152
|
|
|
$crawlerController->setID, |
|
153
|
|
|
['reason' => $reason] |
|
154
|
|
|
); |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
if ($crawlerController->extensionSettings['cleanUpOldQueueEntries']) { |
|
158
|
|
|
$crawlerController->cleanUpOldQueueEntries(); |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
$crawlerController->setID = (int) GeneralUtility::md5int(microtime()); |
|
162
|
|
|
$crawlerController->getPageTreeAndUrls( |
|
163
|
|
|
$pageId, |
|
164
|
|
|
MathUtility::forceIntegerInRange($depth, 0, 99), |
|
165
|
|
|
$crawlerController->getCurrentTime(), |
|
166
|
|
|
MathUtility::forceIntegerInRange($number ? intval($number) : 30, 1, 1000), |
|
167
|
|
|
$mode === 'queue' || $mode === 'exec', |
|
168
|
|
|
$mode === 'url', |
|
169
|
|
|
[], |
|
170
|
|
|
$configurationKeys |
|
171
|
|
|
); |
|
172
|
|
|
|
|
173
|
|
|
if ($mode === 'url') { |
|
174
|
|
|
$this->outputLine("<info>" . implode(chr(10), $crawlerController->downloadUrls) . chr(10) . "</info>"); |
|
175
|
|
|
} elseif ($mode === 'exec') { |
|
176
|
|
|
$this->outputLine("<info>Executing " . count($crawlerController->urlList) . " requests right away:</info>"); |
|
177
|
|
|
$this->outputLine("<info>" . implode(chr(10), $crawlerController->urlList) . "</info>" .chr(10)); |
|
178
|
|
|
$this->outputLine("<info>Processing</info>" . chr(10)); |
|
179
|
|
|
|
|
180
|
|
|
foreach ($crawlerController->queueEntries as $queueRec) { |
|
181
|
|
|
$p = unserialize($queueRec['parameters']); |
|
182
|
|
|
$this->outputLine("<info>" . $p['url'] . " (" . implode(',', $p['procInstructions']) . ") => " . "</info>" . chr(10)); |
|
183
|
|
|
$result = $crawlerController->readUrlFromArray($queueRec); |
|
184
|
|
|
|
|
185
|
|
|
$requestResult = unserialize($result['content']); |
|
186
|
|
|
if (is_array($requestResult)) { |
|
187
|
|
|
$resLog = is_array($requestResult['log']) ? chr(10) . chr(9) . chr(9) . implode(chr(10) . chr(9) . chr(9), $requestResult['log']) : ''; |
|
188
|
|
|
$this->outputLine("<info>OK: " . $resLog . "</info>" . chr(10)); |
|
189
|
|
|
} else { |
|
190
|
|
|
$this->outputLine("<errror>Error checking Crawler Result: " . substr(preg_replace('/\s+/', ' ', strip_tags($result['content'])), 0, 30000) . '...' . chr(10) . "</errror>" . chr(10)); |
|
191
|
|
|
} |
|
192
|
|
|
} |
|
193
|
|
|
} elseif ($mode === 'queue') { |
|
194
|
|
|
$this->outputLine("<info>Putting " . count($crawlerController->urlList) . " entries in queue:</info>" . chr(10)); |
|
195
|
|
|
$this->outputLine("<info>" . implode(chr(10), $crawlerController->urlList) . "</info>" . chr(10)); |
|
196
|
|
|
} else { |
|
197
|
|
|
$this->outputLine("<info>" . count($crawlerController->urlList) . " entries found for processing. (Use --mode to decide action):</info>" . chr(10)); |
|
198
|
|
|
$this->outputLine("<info>" . implode(chr(10), $crawlerController->urlList) . "</info>" . chr(10)); |
|
199
|
|
|
} |
|
200
|
|
|
} |
|
201
|
|
|
|
|
202
|
|
|
/** |
|
203
|
|
|
* Crawler Command - Crawling the URLs from the queue |
|
204
|
|
|
* |
|
205
|
|
|
* Examples: |
|
206
|
|
|
* |
|
207
|
|
|
* --- Will trigger the crawler which starts to process the queue entires |
|
208
|
|
|
* $ typo3cms crawler:crawlqueue |
|
209
|
|
|
* |
|
210
|
|
|
* @param int $amount How many pages should be crawled during that run. |
|
211
|
|
|
* @param int $sleeptime Amount of milliseconds which the system should use to relax between crawls. |
|
212
|
|
|
* @param int $sleepafter Amount of seconds which the system should use to relax after all crawls are done. |
|
213
|
|
|
* |
|
214
|
|
|
* @return int |
|
215
|
|
|
*/ |
|
216
|
|
|
public function crawlQueueCommand($amount = 0, $sleeptime = 0, $sleepafter = 0) |
|
217
|
|
|
{ |
|
218
|
|
|
$result = self::CLI_STATUS_NOTHING_PROCCESSED; |
|
219
|
|
|
|
|
220
|
|
|
/** @var CrawlerController $crawlerController */ |
|
221
|
|
|
$crawlerController = $this->objectManager->get(CrawlerController::class); |
|
222
|
|
|
$queueRepository = $this->objectManager->get(QueueRepository::class); |
|
223
|
|
|
|
|
224
|
|
|
$settings = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['crawler']); |
|
225
|
|
|
$settings = is_array($settings) ? $settings : []; |
|
226
|
|
|
$crawlerController->setExtensionSettings($settings); |
|
227
|
|
|
|
|
228
|
|
|
if (!$crawlerController->getDisabled() && $crawlerController->CLI_checkAndAcquireNewProcess($crawlerController->CLI_buildProcessId())) { |
|
229
|
|
|
$countInARun = $amount ? intval($amount) : $crawlerController->extensionSettings['countInARun']; |
|
230
|
|
|
$sleepAfterFinish = $sleeptime ? intval($sleeptime) : $crawlerController->extensionSettings['sleepAfterFinish']; |
|
231
|
|
|
$sleepTime = $sleepafter ? intval($sleepafter) : $crawlerController->extensionSettings['sleepTime']; |
|
232
|
|
|
|
|
233
|
|
|
try { |
|
234
|
|
|
// Run process: |
|
235
|
|
|
$result = $crawlerController->CLI_run($countInARun, $sleepTime, $sleepAfterFinish); |
|
236
|
|
|
} catch (\Exception $e) { |
|
237
|
|
|
$this->outputLine("<warning>". get_class($e) . ': ' . $e->getMessage() . "</warning>"); |
|
238
|
|
|
$result = self::CLI_STATUS_ABORTED; |
|
239
|
|
|
} |
|
240
|
|
|
|
|
241
|
|
|
// Cleanup |
|
242
|
|
|
$GLOBALS['TYPO3_DB']->exec_DELETEquery('tx_crawler_process', 'assigned_items_count = 0'); |
|
243
|
|
|
$crawlerController->CLI_releaseProcesses($crawlerController->CLI_buildProcessId()); |
|
244
|
|
|
|
|
245
|
|
|
$this->outputLine("<info>Unprocessed Items remaining:" . $queueRepository->countUnprocessedItems() . " (" . $crawlerController->CLI_buildProcessId() . ")</info>"); |
|
246
|
|
|
$result |= ($queueRepository->countUnprocessedItems() > 0 ? self::CLI_STATUS_REMAIN : self::CLI_STATUS_NOTHING_PROCCESSED); |
|
247
|
|
|
} else { |
|
248
|
|
|
$result |= self::CLI_STATUS_ABORTED; |
|
249
|
|
|
} |
|
250
|
|
|
|
|
251
|
|
|
return $result; |
|
252
|
|
|
|
|
253
|
|
|
} |
|
254
|
|
|
|
|
255
|
|
|
/** |
|
256
|
|
|
* Obtains configuration keys from the CLI arguments |
|
257
|
|
|
* |
|
258
|
|
|
* @param $conf string |
|
259
|
|
|
* @return array |
|
260
|
|
|
*/ |
|
261
|
|
|
private function getConfigurationKeys($conf) |
|
262
|
|
|
{ |
|
263
|
|
|
$parameter = trim($conf); |
|
264
|
|
|
return ($parameter != '' ? GeneralUtility::trimExplode(',', $parameter) : []); |
|
265
|
|
|
} |
|
266
|
|
|
} |
AOEpeople /
crawler
