|
1
|
|
|
<?php |
|
2
|
|
|
namespace TYPO3\CMS\IndexedSearch\Hook; |
|
3
|
|
|
|
|
4
|
|
|
/* |
|
5
|
|
|
* This file is part of the TYPO3 CMS project. |
|
6
|
|
|
* |
|
7
|
|
|
* It is free software; you can redistribute it and/or modify it under |
|
8
|
|
|
* the terms of the GNU General Public License, either version 2 |
|
9
|
|
|
* of the License, or any later version. |
|
10
|
|
|
* |
|
11
|
|
|
* For the full copyright and license information, please read the |
|
12
|
|
|
* LICENSE.txt file that was distributed with this source code. |
|
13
|
|
|
* |
|
14
|
|
|
* The TYPO3 project - inspiring people to share! |
|
15
|
|
|
*/ |
|
16
|
|
|
|
|
17
|
|
|
use TYPO3\CMS\Core\Charset\CharsetConverter; |
|
18
|
|
|
use TYPO3\CMS\Core\Charset\UnknownCharsetException; |
|
19
|
|
|
use TYPO3\CMS\Core\Configuration\ExtensionConfiguration; |
|
20
|
|
|
use TYPO3\CMS\Core\Context\Context; |
|
21
|
|
|
use TYPO3\CMS\Core\Context\LanguageAspect; |
|
22
|
|
|
use TYPO3\CMS\Core\TimeTracker\TimeTracker; |
|
23
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
|
24
|
|
|
use TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController; |
|
25
|
|
|
use TYPO3\CMS\IndexedSearch\Indexer; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Hooks for \TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController (TSFE). |
|
29
|
|
|
* @internal this is a TYPO3-internal hook implementation and not part of TYPO3's Core API. |
|
30
|
|
|
*/ |
|
31
|
|
|
class TypoScriptFrontendHook |
|
32
|
|
|
{ |
|
33
|
|
|
/** |
|
34
|
|
|
* Frontend hook: If the page is not being re-generated this is our chance to force it to be |
|
35
|
|
|
* (because re-generation of the page is required in order to have the indexer called!) |
|
36
|
|
|
* |
|
37
|
|
|
* @param array $params Parameters from frontend |
|
38
|
|
|
* @param TypoScriptFrontendController $tsfe TSFE object |
|
39
|
|
|
*/ |
|
40
|
|
|
public function headerNoCache(array &$params, TypoScriptFrontendController $tsfe) |
|
41
|
|
|
{ |
|
42
|
|
|
// Requirements are that the crawler is loaded, a crawler session is running and re-indexing requested as processing instruction: |
|
43
|
|
|
if (in_array('tx_indexedsearch_reindex', $tsfe->applicationData['tx_crawler']['parameters']['procInstructions'] ?? [], true)) { |
|
44
|
|
|
// Disables a look-up for cached page data - thus resulting in re-generation of the page even if cached. |
|
45
|
|
|
$params['disableAcquireCacheData'] = true; |
|
46
|
|
|
// Enable indexing |
|
47
|
|
|
$tsfe->applicationData['forceIndexing'] = true; |
|
48
|
|
|
} |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Trigger indexing of content, after evaluating if this page could / should be indexed. |
|
53
|
|
|
* |
|
54
|
|
|
* @param TypoScriptFrontendController $tsfe |
|
55
|
|
|
*/ |
|
56
|
|
|
public function hook_indexContent(TypoScriptFrontendController $tsfe) |
|
|
|
|
|
|
57
|
|
|
{ |
|
58
|
|
|
// Determine if page should be indexed, and if so, configure and initialize indexer |
|
59
|
|
|
if (!$tsfe->config['config']['index_enable']) { |
|
60
|
|
|
return; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
// Indexer configuration from Extension Manager interface: |
|
64
|
|
|
$disableFrontendIndexing = (bool)GeneralUtility::makeInstance(ExtensionConfiguration::class)->get('indexed_search', 'disableFrontendIndexing'); |
|
65
|
|
|
$forceIndexing = $tsfe->applicationData['forceIndexing'] ?? false; |
|
66
|
|
|
|
|
67
|
|
|
$timeTracker = GeneralUtility::makeInstance(TimeTracker::class); |
|
68
|
|
|
$timeTracker->push('Index page'); |
|
69
|
|
|
if ($disableFrontendIndexing && !$forceIndexing) { |
|
70
|
|
|
$timeTracker->setTSlogMessage('Index page? No, Ordinary Frontend indexing during rendering is disabled.'); |
|
71
|
|
|
return; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
if ($tsfe->page['no_search']) { |
|
75
|
|
|
$timeTracker->setTSlogMessage('Index page? No, The "No Search" flag has been set in the page properties!'); |
|
76
|
|
|
return; |
|
77
|
|
|
} |
|
78
|
|
|
if ($tsfe->no_cache) { |
|
79
|
|
|
$timeTracker->setTSlogMessage('Index page? No, Ordinary Frontend indexing during rendering is disabled.'); |
|
80
|
|
|
return; |
|
81
|
|
|
} |
|
82
|
|
|
/** @var LanguageAspect $languageAspect */ |
|
83
|
|
|
$languageAspect = GeneralUtility::makeInstance(Context::class)->getAspect('language'); |
|
84
|
|
|
if ($languageAspect->getId() !== $languageAspect->getContentId()) { |
|
85
|
|
|
$timeTracker->setTSlogMessage('Index page? No, languageId was different from contentId which indicates that the page contains fall-back content and that would be falsely indexed as localized content.'); |
|
86
|
|
|
return; |
|
87
|
|
|
} |
|
88
|
|
|
// Init and start indexing |
|
89
|
|
|
$indexer = GeneralUtility::makeInstance(Indexer::class); |
|
90
|
|
|
$indexer->forceIndexing = $forceIndexing; |
|
91
|
|
|
$indexer->init($this->initializeIndexerConfiguration($tsfe, $languageAspect)); |
|
92
|
|
|
$indexer->indexTypo3PageContent(); |
|
93
|
|
|
$timeTracker->pull(); |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* Setting up internal configuration from config array based on TypoScriptFrontendController |
|
98
|
|
|
* Information about page for which the indexing takes place |
|
99
|
|
|
* |
|
100
|
|
|
* @param TypoScriptFrontendController $tsfe |
|
101
|
|
|
* @param LanguageAspect $languageAspect |
|
102
|
|
|
* @return array |
|
103
|
|
|
*/ |
|
104
|
|
|
protected function initializeIndexerConfiguration(TypoScriptFrontendController $tsfe, LanguageAspect $languageAspect): array |
|
105
|
|
|
{ |
|
106
|
|
|
$pageArguments = $tsfe->getPageArguments(); |
|
107
|
|
|
$configuration = [ |
|
108
|
|
|
// Page id |
|
109
|
|
|
'id' => $tsfe->id, |
|
110
|
|
|
// Page type |
|
111
|
|
|
'type'=> $tsfe->type, |
|
|
|
|
|
|
112
|
|
|
// sys_language UID of the language of the indexing. |
|
113
|
|
|
'sys_language_uid' => $languageAspect->getId(), |
|
114
|
|
|
// MP variable, if any (Mount Points) |
|
115
|
|
|
'MP' => $tsfe->MP, |
|
116
|
|
|
// Group list |
|
117
|
|
|
'gr_list' => implode(',', GeneralUtility::makeInstance(Context::class)->getPropertyFromAspect('frontend.user', 'groupIds', [0, -1])), |
|
118
|
|
|
// page arguments array |
|
119
|
|
|
'staticPageArguments' => $pageArguments ? $pageArguments->getStaticArguments() : [], |
|
|
|
|
|
|
120
|
|
|
// The creation date of the TYPO3 page |
|
121
|
|
|
'crdate' => $tsfe->page['crdate'], |
|
122
|
|
|
'rootline_uids' => [], |
|
123
|
|
|
]; |
|
124
|
|
|
|
|
125
|
|
|
// Root line uids |
|
126
|
|
|
foreach ($tsfe->config['rootLine'] as $rlkey => $rldat) { |
|
127
|
|
|
$configuration['rootline_uids'][$rlkey] = $rldat['uid']; |
|
128
|
|
|
} |
|
129
|
|
|
// Content of page |
|
130
|
|
|
$configuration['content'] = $tsfe->content; |
|
131
|
|
|
// Content string (HTML of TYPO3 page) |
|
132
|
|
|
$configuration['indexedDocTitle'] = $this->convOutputCharset($tsfe->indexedDocTitle, $tsfe->metaCharset); |
|
133
|
|
|
// Alternative title for indexing |
|
134
|
|
|
$configuration['metaCharset'] = $tsfe->metaCharset; |
|
135
|
|
|
// Character set of content (will be converted to utf-8 during indexing) |
|
136
|
|
|
$configuration['mtime'] = $tsfe->register['SYS_LASTCHANGED'] ?? $tsfe->page['SYS_LASTCHANGED']; |
|
137
|
|
|
// Most recent modification time (seconds) of the content on the page. Used to evaluate whether it should be re-indexed. |
|
138
|
|
|
// Configuration of behavior |
|
139
|
|
|
$configuration['index_externals'] = $tsfe->config['config']['index_externals']; |
|
140
|
|
|
// Whether to index external documents like PDF, DOC etc. (if possible) |
|
141
|
|
|
$configuration['index_descrLgd'] = $tsfe->config['config']['index_descrLgd']; |
|
142
|
|
|
// Length of description text (max 250, default 200) |
|
143
|
|
|
$configuration['index_metatags'] = $tsfe->config['config']['index_metatags'] ?? true; |
|
144
|
|
|
// Set to zero |
|
145
|
|
|
$configuration['recordUid'] = 0; |
|
146
|
|
|
$configuration['freeIndexUid'] = 0; |
|
147
|
|
|
$configuration['freeIndexSetId'] = 0; |
|
148
|
|
|
return $configuration; |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
/** |
|
152
|
|
|
* Converts input string from utf-8 to metaCharset IF the two charsets are different. |
|
153
|
|
|
* |
|
154
|
|
|
* @param string $content Content to be converted. |
|
155
|
|
|
* @param string $metaCharset |
|
156
|
|
|
* @return string Converted content string. |
|
157
|
|
|
*/ |
|
158
|
|
|
protected function convOutputCharset(string $content, string $metaCharset): string |
|
159
|
|
|
{ |
|
160
|
|
|
if ($metaCharset !== 'utf-8') { |
|
161
|
|
|
$charsetConverter = GeneralUtility::makeInstance(CharsetConverter::class); |
|
162
|
|
|
try { |
|
163
|
|
|
$content = $charsetConverter->conv($content, 'utf-8', $metaCharset); |
|
164
|
|
|
} catch (UnknownCharsetException $e) { |
|
165
|
|
|
throw new \RuntimeException('Invalid config.metaCharset: ' . $e->getMessage(), 1508916285); |
|
166
|
|
|
} |
|
167
|
|
|
} |
|
168
|
|
|
return $content; |
|
169
|
|
|
} |
|
170
|
|
|
} |
|
171
|
|
|
|
TYPO3 /
TYPO3.CMS
Loading history...