|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* neuralyzer : Data Anonymization Library and CLI Tool |
|
4
|
|
|
* |
|
5
|
|
|
* PHP Version 7.1 |
|
6
|
|
|
* |
|
7
|
|
|
* @author Emmanuel Dyan |
|
8
|
|
|
* @author Rémi Sauvat |
|
9
|
|
|
* @copyright 2018 Emmanuel Dyan |
|
10
|
|
|
* |
|
11
|
|
|
* @package edyan/neuralyzer |
|
12
|
|
|
* |
|
13
|
|
|
* @license GNU General Public License v2.0 |
|
14
|
|
|
* |
|
15
|
|
|
* @link https://github.com/edyan/neuralyzer |
|
16
|
|
|
*/ |
|
17
|
|
|
|
|
18
|
|
|
namespace Edyan\Neuralyzer; |
|
19
|
|
|
|
|
20
|
|
|
use Edyan\Neuralyzer\Exception\NeuralizerGuesserException; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* Guesser to map field type to Faker Class |
|
24
|
|
|
*/ |
|
25
|
|
|
class Guesser implements GuesserInterface |
|
26
|
|
|
{ |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Returns the version of your guesser |
|
30
|
|
|
* |
|
31
|
|
|
* @return string |
|
32
|
|
|
*/ |
|
33
|
47 |
|
public function getVersion(): string |
|
34
|
|
|
{ |
|
35
|
47 |
|
return '3.0'; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Returns an array of fieldName => Faker class |
|
40
|
|
|
* |
|
41
|
|
|
* @return array |
|
42
|
|
|
*/ |
|
43
|
15 |
|
public function getColsNameMapping(): array |
|
44
|
|
|
{ |
|
45
|
|
|
// can contain regexp |
|
46
|
|
|
return [ |
|
47
|
|
|
// Internet |
|
48
|
15 |
|
'.*email.*' => ['method' => 'email'], |
|
49
|
|
|
'.*url' => ['method' => 'url'], |
|
50
|
|
|
|
|
51
|
|
|
// Adress and coordinates |
|
52
|
|
|
'.*address.*' => ['method' => 'streetAddress'], |
|
53
|
|
|
'.*street.*' => ['method' => 'streetAddress'], |
|
54
|
|
|
'.*postalcode.*' => ['method' => 'postcode'], |
|
55
|
|
|
'.*city.*' => ['method' => 'city'], |
|
56
|
|
|
'.*state.*' => ['method' => 'state'], |
|
57
|
|
|
'.*country.*' => ['method' => 'country'], |
|
58
|
|
|
'.*phone.*' => ['method' => 'phoneNumber'], |
|
59
|
|
|
|
|
60
|
|
|
// Text |
|
61
|
|
|
'.*\.(comments|description)' => ['method' => 'sentence', 'params' => [20]], |
|
62
|
|
|
|
|
63
|
|
|
// Person |
|
64
|
|
|
'.*first_?name' => ['method' => 'firstName'], |
|
65
|
|
|
'.*last_?name' => ['method' => 'lastName'], |
|
66
|
|
|
]; |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* Returns an array of fieldType => Faker method |
|
72
|
|
|
* @param mixed $length Field's length |
|
73
|
|
|
* |
|
74
|
|
|
* @return array |
|
75
|
|
|
*/ |
|
76
|
9 |
|
public function getColsTypeMapping($length): array |
|
77
|
|
|
{ |
|
78
|
|
|
return [ |
|
79
|
|
|
// Strings |
|
80
|
9 |
|
'string' => ['method' => 'sentence', 'params' => [$length]], |
|
81
|
|
|
|
|
82
|
|
|
// Text & Blobs |
|
83
|
|
|
'text' => ['method' => 'sentence', 'params' => [20]], |
|
84
|
|
|
'blob' => ['method' => 'sentence', 'params' => [20]], |
|
85
|
|
|
|
|
86
|
|
|
// DateTime |
|
87
|
|
|
'date' => ['method' => 'date', 'params' => ['Y-m-d']], |
|
88
|
|
|
'datetime' => ['method' => 'date', 'params' => ['Y-m-d H:i:s']], |
|
89
|
|
|
'time' => ['method' => 'time', 'params' => ['H:i:s']], |
|
90
|
|
|
|
|
91
|
|
|
// Integer |
|
92
|
|
|
'boolean' => ['method' => 'boolean', 'params' => [4]], |
|
93
|
|
|
'smallint' => ['method' => 'randomNumber', 'params' => [4]], |
|
94
|
|
|
'integer' => ['method' => 'randomNumber', 'params' => [9]], |
|
95
|
9 |
|
'bigint' => ['method' => 'randomNumber', 'params' => [strlen(mt_getrandmax()) - 1]], |
|
96
|
|
|
|
|
97
|
|
|
// Decimal |
|
98
|
|
|
'float' => ['method' => 'randomFloat', 'params' => [2, 0, 999999]], |
|
99
|
|
|
'decimal' => ['method' => 'randomFloat', 'params' => [2, 0, 999999]], |
|
100
|
|
|
]; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* Will map cols first by looking for field name then by looking for field type |
|
105
|
|
|
* if the first returned nothing |
|
106
|
|
|
* |
|
107
|
|
|
* @param string $table |
|
108
|
|
|
* @param string $name |
|
109
|
|
|
* @param string $type |
|
110
|
|
|
* @param mixed $len Used to get options from enum (stored in length) |
|
111
|
|
|
* |
|
112
|
|
|
* @return array |
|
113
|
|
|
*/ |
|
114
|
14 |
|
public function mapCol(string $table, string $name, string $type, string $len = null): array |
|
115
|
|
|
{ |
|
116
|
|
|
// Try to find by colsName |
|
117
|
14 |
|
$colsName = $this->getColsNameMapping(); |
|
118
|
14 |
|
foreach ($colsName as $colRegex => $params) { |
|
119
|
14 |
|
preg_match("/^$colRegex\$/i", $table. '.' . $name, $matches); |
|
120
|
14 |
|
if (!empty($matches)) { |
|
121
|
14 |
|
return $params; |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
// Hardcoded types |
|
126
|
9 |
|
if ($type === 'enum') { |
|
127
|
|
|
return [ |
|
128
|
1 |
|
'method' => 'randomElement', |
|
129
|
1 |
|
'params' => [explode("','", substr($len, 1, -1))] |
|
130
|
|
|
]; |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
// Try to find by fieldType |
|
134
|
8 |
|
$colsType = $this->getColsTypeMapping($len); |
|
135
|
8 |
|
if (!array_key_exists($type, $colsType)) { |
|
136
|
1 |
|
$msg = "Can't guess the type $type ({$table}.{$name})"; |
|
137
|
1 |
|
throw new NeuralizerGuesserException($msg); |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
7 |
|
return $colsType[$type]; |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
|
edyan /
neuralyzer
