|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* neuralyzer : Data Anonymization Library and CLI Tool |
|
4
|
|
|
* |
|
5
|
|
|
* PHP Version 7.1 |
|
6
|
|
|
* |
|
7
|
|
|
* @package edyan/neuralyzer |
|
8
|
|
|
* |
|
9
|
|
|
* @author Emmanuel Dyan |
|
10
|
|
|
* @author Rémi Sauvat |
|
11
|
|
|
* @copyright 2018 Emmanuel Dyan |
|
12
|
|
|
* @license GNU General Public License v2.0 |
|
13
|
|
|
* |
|
14
|
|
|
* @link https://github.com/edyan/neuralyzer |
|
15
|
|
|
*/ |
|
16
|
|
|
|
|
17
|
|
|
namespace Edyan\Neuralyzer\Configuration; |
|
18
|
|
|
|
|
19
|
|
|
use Edyan\Neuralyzer\Anonymizer\DB; |
|
20
|
|
|
use Edyan\Neuralyzer\Exception\NeuralizerConfigurationException; |
|
21
|
|
|
use Edyan\Neuralyzer\GuesserInterface; |
|
22
|
|
|
use Symfony\Component\Config\Definition\Processor; |
|
23
|
|
|
use Symfony\Component\Yaml\Yaml; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Configuration Writer |
|
27
|
|
|
*/ |
|
28
|
|
|
class Writer |
|
29
|
|
|
{ |
|
30
|
|
|
/** |
|
31
|
|
|
* List of tables patterns to ignore |
|
32
|
|
|
* |
|
33
|
|
|
* @var array |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $ignoredTables = []; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* Should I protect the cols ? That will also protect the Primary Keys |
|
39
|
|
|
* |
|
40
|
|
|
* @var boolean |
|
41
|
|
|
*/ |
|
42
|
|
|
protected $protectCols = true; |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* List the cols to protected. Could containe regexp |
|
46
|
|
|
* |
|
47
|
|
|
* @var array |
|
48
|
|
|
*/ |
|
49
|
|
|
protected $protectedCols = ['id', 'parent_id']; |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Store the tables added to the conf |
|
53
|
|
|
* |
|
54
|
|
|
* @var array |
|
55
|
|
|
*/ |
|
56
|
|
|
protected $tablesInConf = []; |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* Doctrine conection handler |
|
60
|
|
|
* @var \Doctrine\DBAL\Connection |
|
61
|
|
|
*/ |
|
62
|
|
|
private $conn; |
|
63
|
|
|
|
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* Generate the configuration by reading tables + cols |
|
67
|
|
|
* |
|
68
|
|
|
* @param DB $db |
|
69
|
|
|
* @param GuesserInterface $guesser |
|
70
|
|
|
* @return array |
|
71
|
|
|
*/ |
|
72
|
8 |
|
public function generateConfFromDB(DB $db, GuesserInterface $guesser): array |
|
73
|
|
|
{ |
|
74
|
8 |
|
$this->conn = $db->getConn(); |
|
75
|
|
|
|
|
76
|
|
|
// First step : get the list of tables |
|
77
|
8 |
|
$tables = $this->getTablesList(); |
|
78
|
8 |
|
if (empty($tables)) { |
|
79
|
2 |
|
throw new NeuralizerConfigurationException('No tables to read in that database'); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
// For each table, read the cols and guess the Faker |
|
83
|
6 |
|
$data = []; |
|
84
|
6 |
|
foreach ($tables as $table) { |
|
85
|
6 |
|
$cols = $this->getColsList($table); |
|
86
|
|
|
// No cols because all are ignored ? |
|
87
|
5 |
|
if (empty($cols)) { |
|
88
|
1 |
|
continue; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
4 |
|
$data[$table]['cols'] = $this->guessColsAnonType($table, $cols, $guesser); |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
5 |
|
if (empty($data)) { |
|
95
|
1 |
|
throw new NeuralizerConfigurationException('All tables or fields have been ignored'); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
$config = [ |
|
99
|
4 |
|
'entities' => $data |
|
100
|
|
|
]; |
|
101
|
|
|
|
|
102
|
4 |
|
$processor = new Processor(); |
|
103
|
|
|
|
|
104
|
4 |
|
return $processor->processConfiguration(new ConfigDefinition, [$config]); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
|
|
108
|
|
|
/** |
|
109
|
|
|
* Get Tables List added to the conf |
|
110
|
|
|
* |
|
111
|
|
|
* @return array |
|
112
|
|
|
*/ |
|
113
|
1 |
|
public function getTablesInConf(): array |
|
114
|
|
|
{ |
|
115
|
1 |
|
return $this->tablesInConf; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Set a flat to protect cols (Primary Key is protected by default) |
|
121
|
|
|
* |
|
122
|
|
|
* @param bool $protectCols |
|
123
|
|
|
* @return Writer |
|
124
|
|
|
*/ |
|
125
|
4 |
|
public function protectCols(bool $protectCols): Writer |
|
126
|
|
|
{ |
|
127
|
4 |
|
$this->protectCols = $protectCols; |
|
128
|
|
|
|
|
129
|
4 |
|
return $this; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* Save the data to the file as YAML |
|
135
|
|
|
* |
|
136
|
|
|
* @param array $data |
|
137
|
|
|
* @param string $filename |
|
138
|
|
|
*/ |
|
139
|
3 |
|
public function save(array $data, string $filename) |
|
140
|
|
|
{ |
|
141
|
3 |
|
if (!is_writeable(dirname($filename))) { |
|
142
|
1 |
|
throw new NeuralizerConfigurationException(dirname($filename) . ' is not writeable.'); |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
2 |
|
file_put_contents($filename, Yaml::dump($data, 4)); |
|
146
|
2 |
|
} |
|
147
|
|
|
|
|
148
|
|
|
|
|
149
|
|
|
/** |
|
150
|
|
|
* Set protected cols |
|
151
|
|
|
* |
|
152
|
|
|
* @param array $ignoredTables |
|
153
|
|
|
* @return Writer |
|
154
|
|
|
*/ |
|
155
|
2 |
|
public function setIgnoredTables(array $ignoredTables): Writer |
|
156
|
|
|
{ |
|
157
|
2 |
|
$this->ignoredTables = $ignoredTables; |
|
158
|
|
|
|
|
159
|
2 |
|
return $this; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
|
|
163
|
|
|
/** |
|
164
|
|
|
* Set protected cols |
|
165
|
|
|
* |
|
166
|
|
|
* @param array $protectedCols |
|
167
|
|
|
* @return Writer |
|
168
|
|
|
*/ |
|
169
|
4 |
|
public function setProtectedCols(array $protectedCols): Writer |
|
170
|
|
|
{ |
|
171
|
4 |
|
$this->protectedCols = $protectedCols; |
|
172
|
|
|
|
|
173
|
4 |
|
return $this; |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
|
|
177
|
|
|
/** |
|
178
|
|
|
* Check if that col has to be ignored |
|
179
|
|
|
* |
|
180
|
|
|
* @param string $table |
|
181
|
|
|
* @param string $col |
|
182
|
|
|
* @return bool |
|
183
|
|
|
*/ |
|
184
|
5 |
|
protected function colIgnored(string $table, string $col): bool |
|
185
|
|
|
{ |
|
186
|
5 |
|
if ($this->protectCols === false) { |
|
187
|
2 |
|
return false; |
|
188
|
|
|
} |
|
189
|
|
|
|
|
190
|
3 |
|
foreach ($this->protectedCols as $protectedCol) { |
|
191
|
3 |
|
if (preg_match("/^$protectedCol\$/", $table . '.' . $col)) { |
|
192
|
3 |
|
return true; |
|
193
|
|
|
} |
|
194
|
|
|
} |
|
195
|
|
|
|
|
196
|
2 |
|
return false; |
|
197
|
|
|
} |
|
198
|
|
|
|
|
199
|
|
|
|
|
200
|
|
|
/** |
|
201
|
|
|
* Get the cols lists from a connection + table |
|
202
|
|
|
* |
|
203
|
|
|
* @param string $table |
|
204
|
|
|
* @return array |
|
205
|
|
|
*/ |
|
206
|
6 |
|
protected function getColsList(string $table): array |
|
207
|
|
|
{ |
|
208
|
6 |
|
$schema = $this->conn->getSchemaManager(); |
|
209
|
6 |
|
$tableDetails = $schema->listTableDetails($table); |
|
210
|
|
|
// No primary ? Exception ! |
|
211
|
6 |
|
if ($tableDetails->hasPrimaryKey() === false) { |
|
212
|
1 |
|
throw new NeuralizerConfigurationException("Can't work with $table, it has no primary key."); |
|
213
|
|
|
} |
|
214
|
5 |
|
$primaryKey = $tableDetails->getPrimaryKey()->getColumns()[0]; |
|
215
|
|
|
|
|
216
|
5 |
|
$cols = $schema->listTableColumns($table); |
|
217
|
5 |
|
$colsInfo = []; |
|
218
|
5 |
|
foreach ($cols as $col) { |
|
219
|
|
|
// If the col has to be ignored: just leave |
|
220
|
5 |
|
if ($primaryKey === $col->getName() || $this->colIgnored($table, $col->getName())) { |
|
221
|
5 |
|
continue; |
|
222
|
|
|
} |
|
223
|
|
|
|
|
224
|
4 |
|
$colsInfo[] = [ |
|
225
|
4 |
|
'name' => $col->getName(), |
|
226
|
4 |
|
'type' => strtolower((string) $col->getType()), |
|
227
|
4 |
|
'len' => $col->getLength(), |
|
228
|
|
|
]; |
|
229
|
|
|
} |
|
230
|
|
|
|
|
231
|
5 |
|
return $colsInfo; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
|
|
235
|
|
|
/** |
|
236
|
|
|
* Get the table lists from a connection |
|
237
|
|
|
* |
|
238
|
|
|
* @return array |
|
239
|
|
|
*/ |
|
240
|
8 |
|
protected function getTablesList(): array |
|
241
|
|
|
{ |
|
242
|
8 |
|
$schemaManager = $this->conn->getSchemaManager(); |
|
243
|
|
|
|
|
244
|
8 |
|
$tablesInDB = $schemaManager->listTables(); |
|
245
|
8 |
|
$tables = []; |
|
246
|
8 |
|
foreach ($tablesInDB as $table) { |
|
247
|
7 |
|
if ($this->tableIgnored($table->getName())) { |
|
248
|
1 |
|
continue; |
|
249
|
|
|
} |
|
250
|
|
|
|
|
251
|
6 |
|
$tables[] = $this->tablesInConf[] = $table->getName(); |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
8 |
|
return array_values($tables); |
|
255
|
|
|
} |
|
256
|
|
|
|
|
257
|
|
|
|
|
258
|
|
|
/** |
|
259
|
|
|
* Guess the cols with the guesser |
|
260
|
|
|
* |
|
261
|
|
|
* @param string $table |
|
262
|
|
|
* @param array $cols |
|
263
|
|
|
* @param GuesserInterface $guesser |
|
264
|
|
|
* @return array |
|
265
|
|
|
*/ |
|
266
|
4 |
|
protected function guessColsAnonType(string $table, array $cols, GuesserInterface $guesser): array |
|
267
|
|
|
{ |
|
268
|
4 |
|
$mapping = []; |
|
269
|
4 |
|
foreach ($cols as $props) { |
|
270
|
4 |
|
$mapping[$props['name']] = $guesser->mapCol($table, $props['name'], $props['type'], $props['len']); |
|
271
|
|
|
} |
|
272
|
|
|
|
|
273
|
4 |
|
return $mapping; |
|
274
|
|
|
} |
|
275
|
|
|
|
|
276
|
|
|
|
|
277
|
|
|
/** |
|
278
|
|
|
* Check if that table has to be ignored |
|
279
|
|
|
* |
|
280
|
|
|
* @param string $table |
|
281
|
|
|
* @return bool |
|
282
|
|
|
*/ |
|
283
|
7 |
|
protected function tableIgnored(string $table): bool |
|
284
|
|
|
{ |
|
285
|
7 |
|
foreach ($this->ignoredTables as $ignoredTable) { |
|
286
|
1 |
|
if (preg_match("/^$ignoredTable\$/", $table)) { |
|
287
|
1 |
|
return true; |
|
288
|
|
|
} |
|
289
|
|
|
} |
|
290
|
|
|
|
|
291
|
6 |
|
return false; |
|
292
|
|
|
} |
|
293
|
|
|
} |
|
294
|
|
|
|