|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Analytics |
|
4
|
|
|
* |
|
5
|
|
|
* This file is licensed under the Affero General Public License version 3 or |
|
6
|
|
|
* later. See the LICENSE.md file. |
|
7
|
|
|
* |
|
8
|
|
|
* @author Marcel Scherello <[email protected]> |
|
9
|
|
|
* @copyright 2020 Marcel Scherello |
|
10
|
|
|
*/ |
|
11
|
|
|
|
|
12
|
|
|
namespace OCA\Analytics\Datasource; |
|
13
|
|
|
|
|
14
|
|
|
use OCP\IL10N; |
|
15
|
|
|
use OCP\ILogger; |
|
16
|
|
|
|
|
17
|
|
|
class RegexService |
|
18
|
|
|
{ |
|
19
|
|
|
private $logger; |
|
20
|
|
|
private $l10n; |
|
21
|
|
|
|
|
22
|
|
|
public function __construct( |
|
23
|
|
|
IL10N $l10n, |
|
24
|
|
|
ILogger $logger |
|
25
|
|
|
) |
|
26
|
|
|
{ |
|
27
|
|
|
$this->l10n = $l10n; |
|
28
|
|
|
$this->logger = $logger; |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
public function getName(): string |
|
32
|
|
|
{ |
|
33
|
|
|
return $this->l10n->t('HTML grabber'); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* Grab data from external html via regex |
|
38
|
|
|
* |
|
39
|
|
|
* @NoAdminRequired |
|
40
|
|
|
* @param array $option |
|
41
|
|
|
* @return array |
|
42
|
|
|
*/ |
|
43
|
|
|
public function read($option) |
|
44
|
|
|
{ |
|
45
|
|
|
// http headers for requests |
|
46
|
|
|
$headers = array( |
|
|
|
|
|
|
47
|
|
|
'Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', |
|
48
|
|
|
'Accept-Language: en-US,en;q=0.5', |
|
49
|
|
|
'Connection: keep-alive', |
|
50
|
|
|
'DNT: 1', // :) |
|
51
|
|
|
); |
|
52
|
|
|
|
|
53
|
|
|
$regex = $option['regex']; |
|
54
|
|
|
$url = $option['url']; |
|
55
|
|
|
|
|
56
|
|
|
$context = stream_context_create( |
|
57
|
|
|
array( |
|
58
|
|
|
"http" => array( |
|
59
|
|
|
"header" => "User-Agent: NextCloud Analytics APP" |
|
60
|
|
|
) |
|
61
|
|
|
) |
|
62
|
|
|
); |
|
63
|
|
|
|
|
64
|
|
|
$html = file_get_contents($url, false, $context); |
|
65
|
|
|
preg_match_all($regex, $html, $matches); |
|
66
|
|
|
|
|
67
|
|
|
$data = array(); |
|
68
|
|
|
$count = count($matches['dimension']); |
|
69
|
|
|
for ($i = 0; $i < $count; $i++) { |
|
70
|
|
|
if (isset($option['limit'])) { |
|
71
|
|
|
if ($i === (int)$option['limit'] AND (int)$option['limit'] !== 0) break; |
|
72
|
|
|
} |
|
73
|
|
|
array_push($data, ['', $matches['dimension'][$i], $matches['value'][$i]]); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
$header = array(); |
|
77
|
|
|
$header[0] = ''; |
|
78
|
|
|
$header[1] = 'Dimension2'; |
|
79
|
|
|
$header[2] = 'Count'; |
|
80
|
|
|
|
|
81
|
|
|
return [ |
|
82
|
|
|
'header' => $header, |
|
83
|
|
|
'data' => $data, |
|
84
|
|
|
'error' => 0, |
|
85
|
|
|
]; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* template for options & settings |
|
90
|
|
|
* |
|
91
|
|
|
* @NoAdminRequired |
|
92
|
|
|
* @return array |
|
93
|
|
|
*/ |
|
94
|
|
|
public function getTemplate() |
|
95
|
|
|
{ |
|
96
|
|
|
$template = array(); |
|
97
|
|
|
array_push($template, ['id' => 'url', 'name' => 'URL', 'placeholder' => 'url']); |
|
98
|
|
|
array_push($template, ['id' => 'regex', 'name' => 'valid regex', 'placeholder' => '//']); |
|
99
|
|
|
array_push($template, ['id' => 'limit', 'name' => 'Limit', 'placeholder' => 'Number of records']); |
|
100
|
|
|
array_push($template, ['id' => 'timestamp', 'name' => 'Timestamp of dataload', 'placeholder' => 'true/false']); |
|
101
|
|
|
array_push($template, ['id' => 'delete', 'name' => 'Delete all data before load', 'placeholder' => 'true/false']); |
|
102
|
|
|
return $template; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
private function backup() |
|
|
|
|
|
|
106
|
|
|
{ |
|
107
|
|
|
/** |
|
108
|
|
|
* $ch = curl_init(); |
|
109
|
|
|
* curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, FALSE); |
|
110
|
|
|
* curl_setopt($ch, CURLOPT_COOKIESESSION, true ); |
|
111
|
|
|
* curl_setopt($ch, CURLOPT_COOKIEFILE, ''); |
|
112
|
|
|
* curl_setopt($ch, CURLOPT_ENCODING, 'gzip, deflate'); |
|
113
|
|
|
* curl_setopt($ch, CURLOPT_HEADER, false); |
|
114
|
|
|
* curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true); |
|
115
|
|
|
* curl_setopt($ch, CURLOPT_URL, $url); |
|
116
|
|
|
* curl_setopt($ch, CURLOPT_REFERER, $url); |
|
117
|
|
|
* curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE); |
|
118
|
|
|
* curl_setopt($ch, CURLOPT_USERAGENT, 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/75.0.3770.100 Safari/537.36'); |
|
119
|
|
|
* //$html = curl_exec($ch); |
|
120
|
|
|
* curl_close($ch); |
|
121
|
|
|
* |
|
122
|
|
|
* //$this->logger->debug($result); |
|
123
|
|
|
* //$html = str_get_html($result, true, false); |
|
124
|
|
|
* //$this->logger->debug($html); |
|
125
|
|
|
* //$value = $html->find('a[href=/gp/bestsellers/books/405436/ref=pd_zg_hrsr_books]', 0)->parent->parent->first_child->innertext; |
|
126
|
|
|
* //ocument.querySelectorAll("a[href='/gp/bestsellers/books/405436/ref=pd_zg_hrsr_books']")[0].parentNode.parentNode.firstElementChild.innerText |
|
127
|
|
|
* //$string = 'http://www.amazon.de/dp/3964433578'; |
|
128
|
|
|
* //$filter = '/(<span class=\"zg_hrsr_rank\">Nr. )(.*)(<\/span><span class="zg_hrsr_ladder">in <a href="\/gp\/bestsellers)(.*)(hrsr_books">)(.*)(<\/a><\/span>)/'; |
|
129
|
|
|
* //$filter = '/(<span class="zg_hrsr_rank">Nr. )(.*)(<\/span><span class="zg_hrsr_ladder">in <a href="\/gp\/bestsellers\/books\/405436\/ref=pd_zg_hrsr_books">Vietnamesisch)/'; |
|
130
|
|
|
* //$filter = '/(<span class="zg_hrsr_rank">Nr. )(.*)(<\/span>)(.*)(hrsr_books">)/'; |
|
131
|
|
|
* //$filter = '/(<span class="zg_hrsr_rank">Nr. )(?<value>.*)(<\/span>\n )(.*)(?<dimension>Vietnamesisch lernen)/'; |
|
132
|
|
|
* //$this->logger->debug('values all: '. json_encode($values)); |
|
133
|
|
|
* //$values = $values['value']; |
|
134
|
|
|
* //$this->logger->debug('values first array: '. json_encode($values));**/ |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
|
Rello /
analytics
Loading history...