|
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 2019-2022 Marcel Scherello |
|
10
|
|
|
*/ |
|
11
|
|
|
|
|
12
|
|
|
namespace OCA\Analytics\Datasource; |
|
13
|
|
|
|
|
14
|
|
|
use OCP\IL10N; |
|
15
|
|
|
use Psr\Log\LoggerInterface; |
|
16
|
|
|
|
|
17
|
|
|
class Regex implements IDatasource |
|
18
|
|
|
{ |
|
19
|
|
|
private $logger; |
|
20
|
|
|
private $l10n; |
|
21
|
|
|
|
|
22
|
|
|
public function __construct( |
|
23
|
|
|
IL10N $l10n, |
|
24
|
|
|
LoggerInterface $logger |
|
25
|
|
|
) |
|
26
|
|
|
{ |
|
27
|
|
|
$this->l10n = $l10n; |
|
28
|
|
|
$this->logger = $logger; |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @return string Display Name of the datasource |
|
33
|
|
|
*/ |
|
34
|
|
|
public function getName(): string |
|
35
|
|
|
{ |
|
36
|
|
|
return $this->l10n->t('HTML grabber'); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @return int digit unique datasource id |
|
41
|
|
|
*/ |
|
42
|
|
|
public function getId(): int |
|
43
|
|
|
{ |
|
44
|
|
|
return 5; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* @return array available options of the datasoure |
|
49
|
|
|
*/ |
|
50
|
|
|
public function getTemplate(): array |
|
51
|
|
|
{ |
|
52
|
|
|
$template = array(); |
|
53
|
|
|
array_push($template, ['id' => 'url', 'name' => 'URL', 'placeholder' => 'url']); |
|
54
|
|
|
array_push($template, ['id' => 'name', 'name' => 'Data series description', 'placeholder' => 'optional']); |
|
55
|
|
|
array_push($template, ['id' => 'regex', 'name' => $this->l10n->t('valid regex'), 'placeholder' => '//']); |
|
56
|
|
|
array_push($template, ['id' => 'limit', 'name' => $this->l10n->t('Limit'), 'placeholder' => $this->l10n->t('Number of rows')]); |
|
57
|
|
|
array_push($template, ['id' => 'timestamp', 'name' => $this->l10n->t('Timestamp of data load'), 'placeholder' => $this->l10n->t('true/false'), 'type' => 'tf']); |
|
58
|
|
|
return $template; |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* Read the Data |
|
63
|
|
|
* @param $option |
|
64
|
|
|
* @return array available options of the datasoure |
|
65
|
|
|
*/ |
|
66
|
|
|
public function readData($option): array |
|
67
|
|
|
{ |
|
68
|
|
|
$regex = htmlspecialchars_decode($option['regex'], ENT_NOQUOTES); |
|
69
|
|
|
$url = htmlspecialchars_decode($option['url'], ENT_NOQUOTES); |
|
70
|
|
|
|
|
71
|
|
|
$context = stream_context_create( |
|
72
|
|
|
array( |
|
73
|
|
|
"http" => array( |
|
74
|
|
|
"header" => "User-Agent: NextCloud Analytics APP" |
|
75
|
|
|
) |
|
76
|
|
|
) |
|
77
|
|
|
); |
|
78
|
|
|
|
|
79
|
|
|
$html = file_get_contents($url, false, $context); |
|
80
|
|
|
preg_match_all($regex, $html, $matches); |
|
81
|
|
|
|
|
82
|
|
|
$data = array(); |
|
83
|
|
|
$count = count($matches['dimension']); |
|
84
|
|
|
for ($i = 0; $i < $count; $i++) { |
|
85
|
|
|
if (isset($option['limit'])) { |
|
86
|
|
|
if ($i === (int)$option['limit'] AND (int)$option['limit'] !== 0) break; |
|
87
|
|
|
} |
|
88
|
|
|
array_push($data, [$option['name'], $matches['dimension'][$i], $matches['value'][$i]]); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
$header = array(); |
|
92
|
|
|
$header[0] = ''; |
|
93
|
|
|
$header[1] = 'Dimension2'; |
|
94
|
|
|
$header[2] = 'Count'; |
|
95
|
|
|
|
|
96
|
|
|
return [ |
|
97
|
|
|
'header' => $header, |
|
98
|
|
|
'dimensions' => array_slice($header, 0, count($header) - 1), |
|
99
|
|
|
'data' => $data, |
|
100
|
|
|
'error' => 0, |
|
101
|
|
|
'rawData' => $html, |
|
102
|
|
|
'URL' => $url, |
|
103
|
|
|
]; |
|
104
|
|
|
} |
|
105
|
|
|
} |
|
106
|
|
|
|
Rello /
analytics
