|
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
|
|
|
use OCA\Analytics\Service\VariableService; |
|
17
|
|
|
|
|
18
|
|
|
class ExternalFile implements IDatasource |
|
19
|
|
|
{ |
|
20
|
|
|
private $logger; |
|
21
|
|
|
private $userId; |
|
22
|
|
|
private $l10n; |
|
23
|
|
|
private $VariableService; |
|
24
|
|
|
|
|
25
|
|
|
public function __construct( |
|
26
|
|
|
$userId, |
|
27
|
|
|
IL10N $l10n, |
|
28
|
|
|
LoggerInterface $logger, |
|
29
|
|
|
VariableService $VariableService |
|
30
|
|
|
) |
|
31
|
|
|
{ |
|
32
|
|
|
$this->userId = $userId; |
|
33
|
|
|
$this->l10n = $l10n; |
|
34
|
|
|
$this->logger = $logger; |
|
35
|
|
|
$this->VariableService = $VariableService; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @return string Display Name of the datasource |
|
40
|
|
|
*/ |
|
41
|
|
|
public function getName(): string |
|
42
|
|
|
{ |
|
43
|
|
|
return $this->l10n->t('External file') . ': csv'; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @return int digit unique data source id |
|
48
|
|
|
*/ |
|
49
|
|
|
public function getId(): int |
|
50
|
|
|
{ |
|
51
|
|
|
return 4; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* @return array available options of the datasoure |
|
56
|
|
|
*/ |
|
57
|
|
|
public function getTemplate(): array |
|
58
|
|
|
{ |
|
59
|
|
|
$template = array(); |
|
60
|
|
|
$template[] = ['id' => 'link', 'name' => $this->l10n->t('External URL'), 'placeholder' => 'url']; |
|
61
|
|
|
$template[] = ['id' => 'hasHeader', 'name' => $this->l10n->t('Header row'), 'placeholder' => 'true-' . $this->l10n->t('Yes').'/false-'.$this->l10n->t('No'), 'type' => 'tf']; |
|
62
|
|
|
$template[] = ['id' => 'offset', 'name' => $this->l10n->t('Ignore leading rows'), 'placeholder' => $this->l10n->t('Number of rows'), 'type' => 'number']; |
|
63
|
|
|
$template[] = ['id' => 'columns', 'name' => $this->l10n->t('Select columns'), 'placeholder' => $this->l10n->t('e.g. 1,2,4 or leave empty'), 'type' => 'columnPicker']; |
|
64
|
|
|
return $template; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* Read the Data |
|
69
|
|
|
* @param $option |
|
70
|
|
|
* @return array available options of the datasoure |
|
71
|
|
|
*/ |
|
72
|
|
|
public function readData($option): array |
|
73
|
|
|
{ |
|
74
|
|
|
$url = htmlspecialchars_decode($option['link'], ENT_NOQUOTES); |
|
75
|
|
|
$ch = curl_init(); |
|
76
|
|
|
if ($ch !== false) { |
|
77
|
|
|
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, FALSE); |
|
78
|
|
|
curl_setopt($ch, CURLOPT_HEADER, false); |
|
79
|
|
|
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true); |
|
80
|
|
|
curl_setopt($ch, CURLOPT_URL, $url); |
|
81
|
|
|
curl_setopt($ch, CURLOPT_REFERER, $url); |
|
82
|
|
|
curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE); |
|
83
|
|
|
curl_setopt($ch, CURLOPT_USERAGENT, 'Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.13) Gecko/20080311 Firefox/2.0.0.13'); |
|
84
|
|
|
$curlResult = curl_exec($ch); |
|
85
|
|
|
$http_code = curl_getinfo($ch, CURLINFO_HTTP_CODE); |
|
86
|
|
|
curl_close($ch); |
|
87
|
|
|
} else { |
|
88
|
|
|
$curlResult = ''; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
$rows = str_getcsv($curlResult, "\n"); |
|
|
|
|
|
|
92
|
|
|
|
|
93
|
|
|
// remove x number of rows from the beginning |
|
94
|
|
|
if (isset($option['offset']) and is_numeric($option['offset'])) { |
|
95
|
|
|
$rows = array_slice($rows, $option['offset']); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
$selectedColumns = array(); |
|
99
|
|
|
if (isset($option['columns']) && strlen($option['columns']) > 0) { |
|
100
|
|
|
$selectedColumns = str_getcsv($option['columns'], ','); |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
// get the delimiter by reading the first row |
|
104
|
|
|
$delimiter = $this->detectDelimiter($rows[0]); |
|
105
|
|
|
|
|
106
|
|
|
// the first row will define the column headers, even if it is not a real header |
|
107
|
|
|
$header = str_getcsv($rows[0], $delimiter); |
|
108
|
|
|
|
|
109
|
|
|
// if the data has a real header, remove the first row |
|
110
|
|
|
if (!isset($option['hasHeader']) or $option['hasHeader'] !== 'false') { |
|
111
|
|
|
$rows = array_slice($rows, 1); |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
$data = array(); |
|
115
|
|
|
if (count($selectedColumns) !== 0) { |
|
116
|
|
|
// if only a subset of columns or fixed column values are set, they are replaced here |
|
117
|
|
|
$header = $this->minimizeRow($selectedColumns, $header); |
|
118
|
|
|
foreach ($rows as $row) { |
|
119
|
|
|
$data[] = $this->minimizeRow($selectedColumns, str_getcsv($row, $delimiter)); |
|
120
|
|
|
} |
|
121
|
|
|
} else { |
|
122
|
|
|
foreach ($rows as $row) { |
|
123
|
|
|
$data[] = str_getcsv($row, $delimiter); |
|
124
|
|
|
} |
|
125
|
|
|
} |
|
126
|
|
|
unset($rows); |
|
127
|
|
|
|
|
128
|
|
|
return [ |
|
129
|
|
|
'header' => $header, |
|
130
|
|
|
'dimensions' => array_slice($header, 0, count($header) - 1), |
|
131
|
|
|
'data' => $data, |
|
132
|
|
|
'rawdata' => $curlResult, |
|
133
|
|
|
'error' => ($http_code>=200 && $http_code<300) ? 0 : 'HTTP response code: '.$http_code, |
|
|
|
|
|
|
134
|
|
|
]; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
private function minimizeRow($selectedColumns, $row) |
|
138
|
|
|
{ |
|
139
|
|
|
$rowMinimized = array(); |
|
140
|
|
|
foreach ($selectedColumns as $selectedColumn) { |
|
141
|
|
|
if (is_numeric($selectedColumn)) { |
|
142
|
|
|
$rowMinimized[] = $row[$selectedColumn - 1]; |
|
143
|
|
|
} else { |
|
144
|
|
|
// if columns contain replacement variables, they are processed here |
|
145
|
|
|
$rowMinimized[] = $this->VariableService->replaceDatasourceColumns($selectedColumn); |
|
146
|
|
|
} |
|
147
|
|
|
} |
|
148
|
|
|
return $rowMinimized; |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
private function detectDelimiter($data) |
|
152
|
|
|
{ |
|
153
|
|
|
$delimiters = ["\t", ";", "|", ","]; |
|
154
|
|
|
$data_2 = array(); |
|
155
|
|
|
$delimiter = $delimiters[0]; |
|
156
|
|
|
foreach ($delimiters as $d) { |
|
157
|
|
|
//$firstRow = str_getcsv($data, "\n")[0]; |
|
158
|
|
|
$data_1 = str_getcsv($data, $d); |
|
159
|
|
|
if (sizeof($data_1) > sizeof($data_2)) { |
|
160
|
|
|
$delimiter = $d; |
|
161
|
|
|
$data_2 = $data_1; |
|
162
|
|
|
} |
|
163
|
|
|
} |
|
164
|
|
|
return $delimiter; |
|
165
|
|
|
} |
|
166
|
|
|
} |
Rello /
analytics
Loading history...