|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Inet Data Anonymization |
|
4
|
|
|
* |
|
5
|
|
|
* PHP Version 5.3 -> 7.0 |
|
6
|
|
|
* |
|
7
|
|
|
* @author Emmanuel Dyan |
|
8
|
|
|
* @author Rémi Sauvat |
|
9
|
|
|
* @copyright 2005-2015 iNet Process |
|
10
|
|
|
* |
|
11
|
|
|
* @package inetprocess/neuralyzer |
|
12
|
|
|
* |
|
13
|
|
|
* @license GNU General Public License v2.0 |
|
14
|
|
|
* |
|
15
|
|
|
* @link http://www.inetprocess.com |
|
16
|
|
|
*/ |
|
17
|
|
|
|
|
18
|
|
|
namespace Inet\Neuralyzer; |
|
19
|
|
|
|
|
20
|
|
|
use Inet\Neuralyzer\Exception\InetAnonGuesserException; |
|
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
|
8 |
|
public function getVersion() |
|
34
|
|
|
{ |
|
35
|
8 |
|
return '1.0.0b'; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Returns an array of fieldName => Faker class |
|
40
|
|
|
* |
|
41
|
|
|
* @return array |
|
42
|
|
|
*/ |
|
43
|
11 |
|
public function getColsNameMapping() |
|
44
|
|
|
{ |
|
45
|
|
|
// can contain regexp |
|
46
|
|
|
return array( |
|
47
|
|
|
// Adress and coordinates |
|
48
|
|
|
'.*\..*street.*' => array( |
|
49
|
11 |
|
'method' => 'streetAddress', |
|
50
|
11 |
|
), |
|
51
|
|
|
'.*\..*postalcode.*' => array( |
|
52
|
11 |
|
'method' => 'postcode', |
|
53
|
11 |
|
), |
|
54
|
|
|
'.*\..*city.*' => array( |
|
55
|
11 |
|
'method' => 'city', |
|
56
|
11 |
|
), |
|
57
|
|
|
'.*\..*state.*' => array( |
|
58
|
11 |
|
'method' => 'state', |
|
59
|
11 |
|
), |
|
60
|
|
|
'.*\..*country.*' => array( |
|
61
|
11 |
|
'method' => 'country', |
|
62
|
11 |
|
), |
|
63
|
|
|
'.*\..*phone.*' => array( |
|
64
|
11 |
|
'method' => 'phoneNumber', |
|
65
|
11 |
|
), |
|
66
|
|
|
// Internet |
|
67
|
|
|
'.*\.email_address' => array( |
|
68
|
11 |
|
'method' => 'email', |
|
69
|
11 |
|
), |
|
70
|
|
|
'.*\.url' => array( |
|
71
|
11 |
|
'method' => 'url', |
|
72
|
11 |
|
), |
|
73
|
|
|
// Text |
|
74
|
|
|
'.*\.(comments|description)' => array( |
|
75
|
11 |
|
'method' => 'sentence', |
|
76
|
11 |
|
'params' => array(20), |
|
77
|
11 |
|
), |
|
78
|
|
|
// Person |
|
79
|
|
|
'.*\.first_name' => array( |
|
80
|
11 |
|
'method' => 'firstName', |
|
81
|
11 |
|
), |
|
82
|
|
|
'.*\.last_name' => array( |
|
83
|
11 |
|
'method' => 'lastName', |
|
84
|
11 |
|
), |
|
85
|
11 |
|
); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* Retruns an array of fieldType => Faker class |
|
90
|
|
|
* |
|
91
|
|
|
* @return array |
|
92
|
|
|
*/ |
|
93
|
9 |
|
public function getColsTypeMapping() |
|
94
|
|
|
{ |
|
95
|
|
|
return array( |
|
96
|
|
|
// Strings |
|
97
|
|
|
'char' => array( |
|
98
|
9 |
|
'method' => 'sentence', |
|
99
|
9 |
|
'params' => array(8), |
|
100
|
9 |
|
), |
|
101
|
|
|
'varchar' => array( |
|
102
|
9 |
|
'method' => 'sentence', |
|
103
|
9 |
|
'params' => array(8), |
|
104
|
9 |
|
), |
|
105
|
|
|
// Text & Blobs |
|
106
|
|
|
'text' => array( |
|
107
|
9 |
|
'method' => 'sentence', |
|
108
|
9 |
|
'params' => array(20), |
|
109
|
9 |
|
), |
|
110
|
|
|
'blob' => array( |
|
111
|
9 |
|
'method' => 'sentence', |
|
112
|
9 |
|
'params' => array(20), |
|
113
|
9 |
|
), |
|
114
|
|
|
'longtext' => array( |
|
115
|
9 |
|
'method' => 'sentence', |
|
116
|
9 |
|
'params' => array(70), |
|
117
|
9 |
|
), |
|
118
|
|
|
'longblob' => array( |
|
119
|
9 |
|
'method' => 'sentence', |
|
120
|
9 |
|
'params' => array(70), |
|
121
|
9 |
|
), |
|
122
|
|
|
// DateTime |
|
123
|
|
|
'date' => array( |
|
124
|
9 |
|
'method' => 'date', |
|
125
|
9 |
|
'params' => array('Y-m-d', 'now') |
|
126
|
9 |
|
), |
|
127
|
|
|
'datetime' => array( |
|
128
|
9 |
|
'method' => 'date', |
|
129
|
9 |
|
'params' => array('Y-m-d H:i:s', 'now') |
|
130
|
9 |
|
), |
|
131
|
|
|
'time' => array( |
|
132
|
9 |
|
'method' => 'time', |
|
133
|
9 |
|
), |
|
134
|
|
|
// Integer |
|
135
|
|
|
'tinyint' => array( |
|
136
|
9 |
|
'method' => 'randomNumber', |
|
137
|
9 |
|
'params' => array(2), |
|
138
|
9 |
|
), |
|
139
|
|
|
'smallint' => array( |
|
140
|
9 |
|
'method' => 'randomNumber', |
|
141
|
9 |
|
'params' => array(4), |
|
142
|
9 |
|
), |
|
143
|
|
|
'mediumint' => array( |
|
144
|
9 |
|
'method' => 'randomNumber', |
|
145
|
9 |
|
'params' => array(6), |
|
146
|
9 |
|
), |
|
147
|
|
|
'int' => array( |
|
148
|
9 |
|
'method' => 'randomNumber', |
|
149
|
9 |
|
'params' => array(9), |
|
150
|
9 |
|
), |
|
151
|
|
|
'bigint' => array( |
|
152
|
9 |
|
'method' => 'randomNumber', |
|
153
|
9 |
|
'params' => array(15), |
|
154
|
9 |
|
), |
|
155
|
|
|
// Decimal |
|
156
|
|
|
'float' => array( |
|
157
|
9 |
|
'method' => 'randomFloat', |
|
158
|
9 |
|
'params' => array(2, 0, 999999), |
|
159
|
9 |
|
), |
|
160
|
|
|
'decimal' => array( |
|
161
|
9 |
|
'method' => 'randomFloat', |
|
162
|
9 |
|
'params' => array(2, 0, 999999), |
|
163
|
9 |
|
), |
|
164
|
|
|
'double' => array( |
|
165
|
9 |
|
'method' => 'randomFloat', |
|
166
|
9 |
|
'params' => array(2, 0, 999999), |
|
167
|
9 |
|
), |
|
168
|
9 |
|
); |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
/** |
|
172
|
|
|
* Will map cols first by looking for field name then by looking for field type |
|
173
|
|
|
* if the first returned nothing |
|
174
|
|
|
* |
|
175
|
|
|
* @param string $table |
|
176
|
|
|
* @param string $name |
|
177
|
|
|
* @param string $type |
|
178
|
|
|
* |
|
179
|
|
|
* @return array |
|
180
|
|
|
*/ |
|
181
|
10 |
|
public function mapCol($table, $name, $type, $len) |
|
182
|
|
|
{ |
|
183
|
|
|
// Try to find by colsName |
|
184
|
10 |
|
$colsName = $this->getColsNameMapping(); |
|
185
|
10 |
|
foreach ($colsName as $colRegex => $params) { |
|
186
|
10 |
|
preg_match("/^$colRegex\$/", $table. '.' . $name, $matches); |
|
187
|
10 |
|
if (!empty($matches)) { |
|
188
|
1 |
|
return $params; |
|
189
|
|
|
} |
|
190
|
9 |
|
} |
|
191
|
|
|
|
|
192
|
|
|
// Hardcoded types |
|
193
|
9 |
|
if ($type === 'enum') { |
|
194
|
|
|
return array( |
|
195
|
1 |
|
'method' => 'randomElement', |
|
196
|
1 |
|
'params' => array(explode("','", substr($len, 1, -1))) |
|
197
|
1 |
|
); |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
// Try to find by fieldType |
|
201
|
8 |
|
$colsType = $this->getColsTypeMapping(); |
|
202
|
8 |
|
if (!array_key_exists($type, $colsType)) { |
|
203
|
1 |
|
throw new InetAnonGuesserException("Can't guess the type $type"); |
|
204
|
|
|
} |
|
205
|
|
|
|
|
206
|
7 |
|
return $colsType[$type]; |
|
207
|
|
|
} |
|
208
|
|
|
} |
|
209
|
|
|
|
edyan /
neuralyzer
