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