|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types=1); |
|
3
|
|
|
namespace AOE\Crawler\Middleware; |
|
4
|
|
|
|
|
5
|
|
|
/* |
|
6
|
|
|
* This file is part of the TYPO3 Crawler Extension |
|
7
|
|
|
* |
|
8
|
|
|
* It is free software; you can redistribute it and/or modify it under |
|
9
|
|
|
* the terms of the GNU General Public License, either version 2 |
|
10
|
|
|
* of the License, or any later version. |
|
11
|
|
|
* |
|
12
|
|
|
* For the full copyright and license information, please read the |
|
13
|
|
|
* LICENSE.txt file that was distributed with this source code. |
|
14
|
|
|
* |
|
15
|
|
|
* The TYPO3 project - inspiring people to share! |
|
16
|
|
|
*/ |
|
17
|
|
|
|
|
18
|
|
|
|
|
19
|
|
|
use AOE\Crawler\Domain\Repository\QueueRepository; |
|
20
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
21
|
|
|
use Psr\Http\Message\ServerRequestInterface; |
|
22
|
|
|
use Psr\Http\Server\MiddlewareInterface; |
|
23
|
|
|
use Psr\Http\Server\RequestHandlerInterface; |
|
24
|
|
|
use TYPO3\CMS\Core\Context\Context; |
|
25
|
|
|
use TYPO3\CMS\Core\Http\Response; |
|
26
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
|
27
|
|
|
use TYPO3\CMS\Extbase\Object\ObjectManager; |
|
28
|
|
|
use TYPO3\CMS\Frontend\Controller\ErrorController; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* Evaluates HTTP headers and checks if Crawler should register itself. |
|
32
|
|
|
* Needs to be run after TSFE initialization AND Frontend User Authentication. |
|
33
|
|
|
* |
|
34
|
|
|
* Once done, the queue is fetched, and then the frontend request runs through. |
|
35
|
|
|
* |
|
36
|
|
|
* Finally, at the very end, if the crawler is still running, output the data and replace the response. |
|
37
|
|
|
*/ |
|
38
|
|
|
class CrawlerInitialization implements MiddlewareInterface |
|
39
|
|
|
{ |
|
40
|
|
|
/** |
|
41
|
|
|
* @var string |
|
42
|
|
|
*/ |
|
43
|
|
|
protected $headerName = 'X-T3CRAWLER'; |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* @var Context |
|
47
|
|
|
*/ |
|
48
|
|
|
protected $context; |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* @var QueueRepository |
|
52
|
|
|
*/ |
|
53
|
|
|
protected $queueRepository; |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* @param Context|null $context |
|
57
|
|
|
*/ |
|
58
|
|
|
public function __construct(Context $context = null) |
|
59
|
|
|
{ |
|
60
|
|
|
$this->context = $context ?? GeneralUtility::makeInstance(Context::class); |
|
61
|
|
|
$this->queueRepository = GeneralUtility::makeInstance(ObjectManager::class)->get(QueueRepository::class); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* @param ServerRequestInterface $request |
|
66
|
|
|
* @param RequestHandlerInterface $handler |
|
67
|
|
|
* @return ResponseInterface |
|
68
|
|
|
* @throws \TYPO3\CMS\Core\Context\Exception\AspectNotFoundException |
|
69
|
|
|
* @throws \TYPO3\CMS\Core\Error\Http\ServiceUnavailableException |
|
70
|
|
|
*/ |
|
71
|
|
|
public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface |
|
72
|
|
|
{ |
|
73
|
|
|
$crawlerInformation = $request->getHeaderLine($this->headerName) ?? null; |
|
74
|
|
|
if (empty($crawlerInformation)) { |
|
75
|
|
|
return $handler->handle($request); |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
// Authenticate crawler request: |
|
79
|
|
|
//@todo: ask service to exclude current call for special reasons: for example no relevance because the language version is not affected |
|
80
|
|
|
[$queueId, $hash] = explode(':', $crawlerInformation); |
|
|
|
|
|
|
81
|
|
|
$queueRec = $this->queueRepository->findByQueueId($queueId); |
|
82
|
|
|
|
|
83
|
|
|
// If a crawler record was found and hash was matching, set it up |
|
84
|
|
|
if (!$this->isRequestHashMatchingQueueRecord($queueRec, $hash)) { |
|
85
|
|
|
return GeneralUtility::makeInstance(ErrorController::class)->unavailableAction($request, 'No crawler entry found'); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
$queueParameters = unserialize($queueRec['parameters']); |
|
89
|
|
|
$GLOBALS['TSFE']->applicationData['tx_crawler']['running'] = true; |
|
90
|
|
|
$GLOBALS['TSFE']->applicationData['tx_crawler']['parameters'] = $queueParameters; |
|
91
|
|
|
$GLOBALS['TSFE']->applicationData['tx_crawler']['log'] = []; |
|
92
|
|
|
$request = $request->withAttribute('tx_crawler', $queueParameters); |
|
93
|
|
|
|
|
94
|
|
|
// Now ensure to set the proper user groups |
|
95
|
|
|
$grList = $queueParameters['feUserGroupList']; |
|
96
|
|
|
if ($grList) { |
|
97
|
|
|
if (!is_array($GLOBALS['TSFE']->fe_user->user)) { |
|
98
|
|
|
$GLOBALS['TSFE']->fe_user->user = []; |
|
99
|
|
|
} |
|
100
|
|
|
$GLOBALS['TSFE']->fe_user->user['usergroup'] = $grList; |
|
101
|
|
|
$GLOBALS['TSFE']->applicationData['tx_crawler']['log'][] = 'User Groups: ' . $grList; |
|
102
|
|
|
// @todo: set the frontend user aspect again. |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
// Execute the frontend request as is |
|
106
|
|
|
$handler->handle($request); |
|
107
|
|
|
|
|
108
|
|
|
$GLOBALS['TSFE']->applicationData['tx_crawler']['vars'] = [ |
|
109
|
|
|
'id' => $GLOBALS['TSFE']->id, |
|
110
|
|
|
'gr_list' => implode(',', $this->context->getAspect('frontend.user')->getGroupIds()), |
|
111
|
|
|
'no_cache' => $GLOBALS['TSFE']->no_cache, |
|
112
|
|
|
]; |
|
113
|
|
|
|
|
114
|
|
|
$this->runPollSuccessHooks(); |
|
115
|
|
|
|
|
116
|
|
|
// Output log data for crawler (serialized content): |
|
117
|
|
|
$content = serialize($GLOBALS['TSFE']->applicationData['tx_crawler']); |
|
118
|
|
|
$response = new Response(); |
|
119
|
|
|
$response->getBody()->write($content); |
|
120
|
|
|
return $response; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
protected function isRequestHashMatchingQueueRecord(?array $queueRec, string $hash): bool |
|
124
|
|
|
{ |
|
125
|
|
|
return is_array($queueRec) && $hash === md5($queueRec['qid'] . '|' . $queueRec['set_id'] . '|' . $GLOBALS['TYPO3_CONF_VARS']['SYS']['encryptionKey']); |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* Required because some extensions (staticpub) might never be requested to run due to some Core side effects |
|
130
|
|
|
* and since this is considered as error the crawler should handle it properly |
|
131
|
|
|
*/ |
|
132
|
|
|
protected function runPollSuccessHooks() |
|
133
|
|
|
{ |
|
134
|
|
|
if (!is_array($GLOBALS['TSFE']->applicationData['tx_crawler']['content']['parameters']['procInstructions'])) { |
|
135
|
|
|
return; |
|
136
|
|
|
} |
|
137
|
|
|
foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['pollSuccess'] ?? [] as $pollable) { |
|
138
|
|
|
if (in_array($pollable, $GLOBALS['TSFE']->applicationData['tx_crawler']['content']['parameters']['procInstructions'])) { |
|
139
|
|
|
if (empty($GLOBALS['TSFE']->applicationData['tx_crawler']['success'][$pollable])) { |
|
140
|
|
|
$GLOBALS['TSFE']->applicationData['tx_crawler']['errorlog'][] = 'Error: Pollable extension (' . $pollable . ') did not complete successfully.'; |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
|
|
} |
|
147
|
|
|
|
AOEpeople /
crawler
Loading history...
This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug.