1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* neuralyzer : Data Anonymization Library and CLI Tool |
4
|
|
|
* |
5
|
|
|
* PHP Version 7.0 |
6
|
|
|
* |
7
|
|
|
* @author Emmanuel Dyan |
8
|
|
|
* @author Rémi Sauvat |
9
|
|
|
* @copyright 2017 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 Inet\Neuralyzer\Configuration; |
19
|
|
|
|
20
|
|
|
use Inet\Neuralyzer\GuesserInterface; |
21
|
|
|
use Inet\Neuralyzer\Exception\NeuralizerConfigurationException; |
22
|
|
|
use Symfony\Component\Yaml\Yaml; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* Configuration Writer |
26
|
|
|
*/ |
27
|
|
|
class Writer |
28
|
|
|
{ |
29
|
|
|
/** |
30
|
|
|
* Should I protect the cols ? That will also protect the Primary Keys |
31
|
|
|
* |
32
|
|
|
* @var boolean |
33
|
|
|
*/ |
34
|
|
|
protected $protectCols = true; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* List the cols to protected. Could containe regexp |
38
|
|
|
* |
39
|
|
|
* @var array |
40
|
|
|
*/ |
41
|
|
|
protected $protectedCols = ['id', 'parent_id']; |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* List of tables patterns to ignore |
45
|
|
|
* |
46
|
|
|
* @var array |
47
|
|
|
*/ |
48
|
|
|
protected $ignoredTables = []; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* Store the tables added to the conf |
52
|
|
|
* |
53
|
|
|
* @var array |
54
|
|
|
*/ |
55
|
|
|
protected $tablesInConf = []; |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* Set Change Id to ask the writer to ignore (or not) the protectedCols |
59
|
|
|
* |
60
|
|
|
* @param bool |
61
|
|
|
*/ |
62
|
|
|
public function protectCols(bool $protect): Writer |
63
|
|
|
{ |
64
|
|
|
$this->protectCols = $protect; |
65
|
|
|
|
66
|
|
|
return $this; |
67
|
|
|
} |
68
|
|
|
|
69
|
|
|
/** |
70
|
|
|
* Set protected cols |
71
|
|
|
* |
72
|
|
|
* @param array $cols |
73
|
|
|
*/ |
74
|
|
|
public function setProtectedCols(array $cols): Writer |
75
|
|
|
{ |
76
|
|
|
$this->protectedCols = $cols; |
77
|
|
|
|
78
|
|
|
return $this; |
79
|
|
|
} |
80
|
|
|
|
81
|
|
|
/** |
82
|
|
|
* Set protected cols |
83
|
|
|
* |
84
|
|
|
* @param array $tables |
85
|
|
|
*/ |
86
|
|
|
public function setIgnoredTables(array $tables): Writer |
87
|
|
|
{ |
88
|
|
|
$this->ignoredTables = $tables; |
89
|
|
|
|
90
|
|
|
return $this; |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* Get Tables List added to the conf |
95
|
|
|
* |
96
|
|
|
* @return array |
97
|
|
|
*/ |
98
|
|
|
public function getTablesInConf(): array |
99
|
|
|
{ |
100
|
|
|
return $this->tablesInConf; |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* Generate the configuration by reading tables + cols |
105
|
|
|
* |
106
|
|
|
* @param \PDO $pdo |
107
|
|
|
* @param GuesserInterface $guesser |
108
|
|
|
* |
109
|
|
|
* @return array |
110
|
|
|
*/ |
111
|
|
|
public function generateConfFromDB(\PDO $pdo, GuesserInterface $guesser): array |
112
|
|
|
{ |
113
|
|
|
|
114
|
|
|
// First step : get the list of tables |
115
|
|
|
$tables = $this->getTablesList($pdo); |
116
|
|
|
if (empty($tables)) { |
117
|
|
|
throw new NeuralizerConfigurationException('No tables to read in that database'); |
118
|
|
|
} |
119
|
|
|
|
120
|
|
|
// For each table, read the cols and guess the Faker |
121
|
|
|
$data = []; |
122
|
|
|
foreach ($tables as $table) { |
123
|
|
|
$cols = $this->getColsList($pdo, $table); |
124
|
|
|
// No cols because all are ignored ? |
125
|
|
|
if (empty($cols)) { |
126
|
|
|
continue; |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
$data[$table]['cols'] = $this->guessColsAnonType($table, $cols, $guesser); |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
if (empty($data)) { |
133
|
|
|
throw new NeuralizerConfigurationException('All tables or fields have been ignored'); |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
return ['guesser_version' => $guesser->getVersion(), 'entities' => $data]; |
137
|
|
|
} |
138
|
|
|
|
139
|
|
|
/** |
140
|
|
|
* Save the data to the file as YAML |
141
|
|
|
* |
142
|
|
|
* @param array $data |
143
|
|
|
* @param string $filename |
144
|
|
|
*/ |
145
|
|
|
public function save(array $data, string $filename) |
146
|
|
|
{ |
147
|
|
|
if (!is_writeable(dirname($filename))) { |
148
|
|
|
throw new NeuralizerConfigurationException(dirname($filename) . ' is not writeable.'); |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
file_put_contents($filename, Yaml::dump($data, 4)); |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* Check if that col has to be ignored |
156
|
|
|
* |
157
|
|
|
* @param string $table |
158
|
|
|
* @param string $col |
159
|
|
|
* @param bool $isPrimary |
160
|
|
|
* |
161
|
|
|
* @return bool |
162
|
|
|
*/ |
163
|
|
|
protected function colIgnored(string $table, string $col, bool $isPrimary): bool |
164
|
|
|
{ |
165
|
|
|
if ($this->protectCols === false) { |
166
|
|
|
return false; |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
foreach ($this->protectedCols as $protectedCol) { |
170
|
|
|
if ($isPrimary || preg_match("/^$protectedCol\$/", $table. '.' . $col)) { |
171
|
|
|
return true; |
172
|
|
|
} |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
return false; |
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
/** |
179
|
|
|
* Get the table lists from a connection |
180
|
|
|
* |
181
|
|
|
* @param \PDO $pdo |
182
|
|
|
* |
183
|
|
|
* @return array |
184
|
|
|
*/ |
185
|
|
|
protected function getTablesList(\PDO $pdo): array |
186
|
|
|
{ |
187
|
|
|
$result = $pdo->query("SHOW FULL TABLES WHERE Table_type = 'BASE TABLE'"); |
188
|
|
|
|
189
|
|
|
$tables = $result->fetchAll(\PDO::FETCH_COLUMN); |
190
|
|
|
foreach ($tables as $key => $val) { |
191
|
|
|
if ($this->tableIgnored($val)) { |
192
|
|
|
unset($tables[$key]); |
193
|
|
|
continue; |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
$this->tablesInConf[] = $val; |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
return array_values($tables); |
200
|
|
|
} |
201
|
|
|
|
202
|
|
|
/** |
203
|
|
|
* Get the cols lists from a connection + table |
204
|
|
|
* |
205
|
|
|
* @param \PDO $pdo |
206
|
|
|
* @param string $table |
207
|
|
|
* |
208
|
|
|
* @return array |
209
|
|
|
*/ |
210
|
|
|
protected function getColsList(\PDO $pdo, string $table): array |
211
|
|
|
{ |
212
|
|
|
$result = $pdo->query("SHOW COLUMNS FROM $table"); |
213
|
|
|
|
214
|
|
|
$cols = $result->fetchAll(\PDO::FETCH_ASSOC); |
215
|
|
|
$colsInfo = []; |
216
|
|
|
|
217
|
|
|
$hasPrimary = false; |
218
|
|
|
foreach ($cols as $col) { |
219
|
|
|
// get the info that we found a primary key |
220
|
|
|
$isPrimary = ($col['Key'] === 'PRI' ? true : false); |
221
|
|
|
if ($isPrimary) { |
222
|
|
|
$hasPrimary = true; |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
// If the col has to be ignored: just leave |
226
|
|
|
if ($this->colIgnored($table, $col['Field'], $isPrimary)) { |
227
|
|
|
continue; |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
$length = null; |
231
|
|
|
$type = $col['Type']; |
232
|
|
|
preg_match('|^(.*)\((.*)\)|', $col['Type'], $colProps); |
233
|
|
|
if (!empty($colProps)) { |
234
|
|
|
$type = $colProps[1]; |
235
|
|
|
$length = $colProps[2]; |
236
|
|
|
} |
237
|
|
|
|
238
|
|
|
$colsInfo[] = [ |
239
|
|
|
'name' => $col['Field'], |
240
|
|
|
'type' => $type, |
241
|
|
|
'len' => $length, |
242
|
|
|
'id' => $isPrimary, |
243
|
|
|
]; |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
// No primary ? Exception ! |
247
|
|
|
if ($hasPrimary === false) { |
248
|
|
|
throw new NeuralizerConfigurationException("Can't work with $table, it has no primary key."); |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
return $colsInfo; |
252
|
|
|
} |
253
|
|
|
|
254
|
|
|
/** |
255
|
|
|
* Guess the cols with the guesser |
256
|
|
|
* |
257
|
|
|
* @param string $table |
258
|
|
|
* @param array $cols |
259
|
|
|
* @param GuesserInterface $guesser |
260
|
|
|
* |
261
|
|
|
* @return array |
262
|
|
|
*/ |
263
|
|
|
protected function guessColsAnonType(string $table, array $cols, GuesserInterface $guesser): array |
264
|
|
|
{ |
265
|
|
|
$mapping = []; |
266
|
|
|
foreach ($cols as $props) { |
267
|
|
|
$mapping[$props['name']] = $guesser->mapCol($table, $props['name'], $props['type'], $props['len']); |
268
|
|
|
} |
269
|
|
|
|
270
|
|
|
return $mapping; |
271
|
|
|
} |
272
|
|
|
|
273
|
|
|
/** |
274
|
|
|
* Check if that table has to be ignored |
275
|
|
|
* |
276
|
|
|
* @param string $table |
277
|
|
|
* |
278
|
|
|
* @return bool |
279
|
|
|
*/ |
280
|
|
|
protected function tableIgnored(string $table): bool |
281
|
|
|
{ |
282
|
|
|
foreach ($this->ignoredTables as $ignoredTable) { |
283
|
|
|
if (preg_match("/^$ignoredTable\$/", $table)) { |
284
|
|
|
return true; |
285
|
|
|
} |
286
|
|
|
} |
287
|
|
|
|
288
|
|
|
return false; |
289
|
|
|
} |
290
|
|
|
} |
291
|
|
|
|