1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
/** |
6
|
|
|
* neuralyzer : Data Anonymization Library and CLI Tool |
7
|
|
|
* |
8
|
|
|
* PHP Version 7.2 |
9
|
|
|
* |
10
|
|
|
* @author Emmanuel Dyan |
11
|
|
|
* @author Rémi Sauvat |
12
|
|
|
* |
13
|
|
|
* @copyright 2020 Emmanuel Dyan |
14
|
|
|
* |
15
|
|
|
* @package edyan/neuralyzer |
16
|
|
|
* |
17
|
|
|
* @license GNU General Public License v2.0 |
18
|
|
|
* |
19
|
|
|
* @link https://github.com/edyan/neuralyzer |
20
|
|
|
*/ |
21
|
|
|
|
22
|
|
|
namespace Edyan\Neuralyzer\Console\Commands; |
23
|
|
|
|
24
|
|
|
use Edyan\Neuralyzer\Anonymizer\DB; |
25
|
|
|
use Edyan\Neuralyzer\Configuration\Reader; |
26
|
|
|
use Edyan\Neuralyzer\Utils\DBUtils; |
27
|
|
|
use Edyan\Neuralyzer\Utils\Expression; |
28
|
|
|
use Edyan\Neuralyzer\Utils\FileLoader; |
29
|
|
|
use Symfony\Component\Console\Command\Command; |
30
|
|
|
use Symfony\Component\Console\Helper\ProgressBar; |
31
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
32
|
|
|
use Symfony\Component\Console\Input\InputOption; |
33
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
34
|
|
|
use Symfony\Component\Console\Question\Question; |
35
|
|
|
use Symfony\Component\Stopwatch\Stopwatch; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* Command to launch an anonymization based on a config file |
39
|
|
|
*/ |
40
|
|
|
class RunCommand extends Command |
41
|
|
|
{ |
42
|
|
|
/** |
43
|
|
|
* Store the DB Object |
44
|
|
|
* |
45
|
|
|
* @var DB |
46
|
|
|
*/ |
47
|
|
|
private $db; |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* Set the command shortcut to be used in configuration |
51
|
|
|
* |
52
|
|
|
* @var string |
53
|
|
|
*/ |
54
|
|
|
private $command = 'run'; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* Symfony's Input Class for parameters and options |
58
|
|
|
* |
59
|
|
|
* @var InputInterface |
60
|
|
|
*/ |
61
|
|
|
private $input; |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* Symfony's Output Class to display info |
65
|
|
|
* |
66
|
|
|
* @var OutputInterface |
67
|
|
|
*/ |
68
|
|
|
private $output; |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* Neuralyzer reader |
72
|
|
|
* |
73
|
|
|
* @var Reader |
74
|
|
|
*/ |
75
|
|
|
private $reader; |
76
|
|
|
|
77
|
|
|
/** |
78
|
|
|
* Store the DBUtils Object (autowiring) |
79
|
|
|
* |
80
|
|
|
* @var DBUtils |
81
|
|
|
*/ |
82
|
|
|
private $dbUtils; |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* Store the Expression Object (autowiring) |
86
|
|
|
* |
87
|
|
|
* @var Expression |
88
|
|
|
*/ |
89
|
|
|
private $expression; |
90
|
|
|
|
91
|
|
|
public function __construct(DBUtils $dbUtils, Expression $expression) |
92
|
|
|
{ |
93
|
28 |
|
parent::__construct(); |
94
|
|
|
|
95
|
28 |
|
$this->dbUtils = $dbUtils; |
96
|
|
|
$this->expression = $expression; |
97
|
28 |
|
} |
98
|
28 |
|
|
99
|
28 |
|
protected function configure(): void |
100
|
|
|
{ |
101
|
|
|
// First command : Test the DB Connexion |
102
|
|
|
$this->setName($this->command) |
103
|
|
|
->setDescription('Run Anonymizer') |
104
|
|
|
->setHelp( |
105
|
|
|
'This command will connect to a DB and run the anonymizer from a yaml config'.PHP_EOL. |
106
|
28 |
|
"Usage: <info>./bin/neuralyzer {$this->command} -u app -p app -f neuralyzer.yml</info>" |
107
|
|
|
)->addOption( |
108
|
|
|
'driver', |
109
|
28 |
|
'D', |
110
|
28 |
|
InputOption::VALUE_REQUIRED, |
111
|
28 |
|
'Driver (check Doctrine documentation to have the list)', |
112
|
28 |
|
'pdo_mysql' |
113
|
28 |
|
)->addOption( |
114
|
28 |
|
'host', |
115
|
28 |
|
'H', |
116
|
28 |
|
InputOption::VALUE_REQUIRED, |
117
|
28 |
|
'Host', |
118
|
28 |
|
'127.0.0.1' |
119
|
28 |
|
)->addOption( |
120
|
28 |
|
'db', |
121
|
28 |
|
'd', |
122
|
28 |
|
InputOption::VALUE_REQUIRED, |
123
|
28 |
|
'Database Name' |
124
|
28 |
|
)->addOption( |
125
|
28 |
|
'user', |
126
|
28 |
|
'u', |
127
|
28 |
|
InputOption::VALUE_REQUIRED, |
128
|
28 |
|
'User Name', |
129
|
28 |
|
get_current_user() |
130
|
28 |
|
)->addOption( |
131
|
28 |
|
'password', |
132
|
28 |
|
'p', |
133
|
28 |
|
InputOption::VALUE_REQUIRED, |
134
|
28 |
|
'Password (or prompted)' |
135
|
28 |
|
)->addOption( |
136
|
28 |
|
'config', |
137
|
28 |
|
'c', |
138
|
28 |
|
InputOption::VALUE_REQUIRED, |
139
|
28 |
|
'Configuration File', |
140
|
28 |
|
'neuralyzer.yml' |
141
|
28 |
|
)->addOption( |
142
|
28 |
|
'table', |
143
|
28 |
|
't', |
144
|
28 |
|
InputOption::VALUE_REQUIRED, |
145
|
28 |
|
'Do a single table' |
146
|
28 |
|
)->addOption( |
147
|
28 |
|
'pretend', |
148
|
28 |
|
null, |
149
|
28 |
|
InputOption::VALUE_NONE, |
150
|
28 |
|
"Don't run queries (pre and post actions will always be executed)" |
151
|
28 |
|
)->addOption( |
152
|
28 |
|
'sql', |
153
|
28 |
|
's', |
154
|
28 |
|
InputOption::VALUE_NONE, |
155
|
28 |
|
'Display the SQL' |
156
|
28 |
|
)->addOption( |
157
|
28 |
|
'mode', |
158
|
28 |
|
'm', |
159
|
28 |
|
InputOption::VALUE_REQUIRED, |
160
|
28 |
|
'Set the mode : batch or queries', |
161
|
28 |
|
'batch' |
162
|
28 |
|
)->addOption( |
163
|
28 |
|
'bootstrap', |
164
|
28 |
|
'b', |
165
|
28 |
|
InputOption::VALUE_REQUIRED, |
166
|
28 |
|
'Provide a bootstrap file to load a custom setup before executing the command. Format /path/to/bootstrap.php' |
167
|
28 |
|
) |
168
|
28 |
|
; |
169
|
28 |
|
} |
170
|
28 |
|
|
171
|
28 |
|
/** |
172
|
28 |
|
* @param InputInterface $input Symfony's Input Class for parameters and options |
173
|
28 |
|
* @param OutputInterface $output Symfony's Output Class to display info |
174
|
28 |
|
* |
175
|
28 |
|
* @return int|null null or 0 if everything went fine, or an error code |
176
|
28 |
|
* |
177
|
28 |
|
* @throws \Doctrine\DBAL\DBALException |
178
|
28 |
|
* @throws \Edyan\Neuralyzer\Exception\NeuralyzerException |
179
|
|
|
*/ |
180
|
|
|
protected function execute(InputInterface $input, OutputInterface $output): int |
181
|
28 |
|
{ |
182
|
|
|
// Throw an exception immediately if we don't have the required DB parameter |
183
|
|
|
if (empty($input->getOption('db'))) { |
184
|
|
|
throw new \InvalidArgumentException('Database name is required (--db)'); |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
// Throw an exception immediately if we don't have the right mode |
188
|
|
|
if (! in_array($input->getOption('mode'), ['queries', 'batch'])) { |
189
|
|
|
throw new \InvalidArgumentException('--mode could be only "queries" or "batch"'); |
190
|
|
|
} |
191
|
18 |
|
|
192
|
|
|
$password = $input->getOption('password'); |
193
|
18 |
|
if ($password === null) { |
194
|
|
|
$question = new Question('Password: '); |
195
|
|
|
$question->setHidden(true)->setHiddenFallback(false); |
196
|
|
|
|
197
|
|
|
$password = $this->getHelper('question')->ask($input, $output, $question); |
198
|
18 |
|
} |
199
|
1 |
|
|
200
|
|
|
$this->input = $input; |
201
|
|
|
$this->output = $output; |
202
|
|
|
|
203
|
17 |
|
// Anon READER |
204
|
1 |
|
$this->reader = new Reader($input->getOption('config')); |
205
|
|
|
if (! empty($this->reader->getDepreciationMessages())) { |
206
|
|
|
foreach ($this->reader->getDepreciationMessages() as $message) { |
207
|
16 |
|
$output->writeLn("<comment>WARNING : ${message}</comment>"); |
208
|
16 |
|
} |
209
|
4 |
|
} |
210
|
4 |
|
|
211
|
|
|
$this->dbUtils->configure([ |
212
|
4 |
|
'driver' => $input->getOption('driver'), |
213
|
|
|
'host' => $input->getOption('host'), |
214
|
|
|
'dbname' => $input->getOption('db'), |
215
|
16 |
|
'user' => $input->getOption('user'), |
216
|
16 |
|
'password' => $password, |
217
|
|
|
]); |
218
|
|
|
|
219
|
16 |
|
$this->db = new DB($this->expression, $this->dbUtils); |
220
|
16 |
|
$this->db->setConfiguration($this->reader); |
221
|
|
|
$this->db->setMode($this->input->getOption('mode')); |
222
|
|
|
$this->db->setPretend($this->input->getOption('pretend')); |
223
|
|
|
$this->db->setReturnRes($this->input->getOption('sql')); |
224
|
|
|
|
225
|
|
|
if (! empty($input->getOption('bootstrap'))) { |
226
|
16 |
|
FileLoader::checkAndLoad($input->getOption('bootstrap')); |
227
|
16 |
|
} |
228
|
16 |
|
|
229
|
16 |
|
$stopwatch = new Stopwatch(); |
230
|
16 |
|
$stopwatch->start('Neuralyzer'); |
231
|
16 |
|
// Get tables |
232
|
|
|
$table = $input->getOption('table'); |
233
|
|
|
$tables = empty($table) ? $this->reader->getEntities() : [$table]; |
234
|
15 |
|
$hasErrors = false; |
235
|
15 |
|
foreach ($tables as $table) { |
236
|
15 |
|
if (! $this->anonymizeTable($table)) { |
237
|
15 |
|
$hasErrors = true; |
238
|
15 |
|
} |
239
|
|
|
} |
240
|
15 |
|
|
241
|
15 |
|
// Get memory and execution time information |
242
|
|
|
$event = $stopwatch->stop('Neuralyzer'); |
243
|
15 |
|
$memory = round($event->getMemory() / 1024 / 1024, 2); |
244
|
15 |
|
$time = round($event->getDuration() / 1000, 2); |
245
|
15 |
|
$time = ($time > 180 ? round($time / 60, 2).'min' : "${time} sec"); |
246
|
15 |
|
|
247
|
15 |
|
$output->writeln("<info>Done in ${time} using ${memory} Mb of memory</info>"); |
248
|
13 |
|
|
249
|
|
|
return $hasErrors ? Command::FAILURE : Command::SUCCESS; |
250
|
|
|
} |
251
|
|
|
|
252
|
|
|
/** |
253
|
13 |
|
* Anonymize a specific table and display info about the job |
254
|
13 |
|
*/ |
255
|
13 |
|
private function anonymizeTable(string $table): bool |
256
|
13 |
|
{ |
257
|
|
|
$total = $this->getTotal($table); |
258
|
13 |
|
if ($total === 0) { |
259
|
|
|
$this->output->writeln("<info>${table} is empty</info>"); |
260
|
13 |
|
|
261
|
|
|
return true; |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
$bar = new ProgressBar($this->output, $total); |
265
|
|
|
$bar->setRedrawFrequency($total > 100 ? 100 : 0); |
266
|
|
|
|
267
|
|
|
$this->output->writeln("<info>Anonymizing ${table}</info>"); |
268
|
|
|
|
269
|
|
|
try { |
270
|
15 |
|
$queries = $this->db->setLimit($total)->processEntity( |
271
|
|
|
$table, |
272
|
15 |
|
static function () use ($bar): void { |
273
|
13 |
|
$bar->advance(); |
274
|
2 |
|
} |
275
|
|
|
); |
276
|
2 |
|
// @codeCoverageIgnoreStart |
277
|
|
|
} catch (\Exception $e) { |
278
|
|
|
$msg = "<error>Error anonymizing ${table}. Message was : " . $e->getMessage() . '</error>'; |
279
|
11 |
|
$this->output->writeln(PHP_EOL . $msg . PHP_EOL); |
280
|
11 |
|
|
281
|
|
|
return false; |
282
|
11 |
|
} |
283
|
|
|
// @codeCoverageIgnoreEnd |
284
|
|
|
|
285
|
11 |
|
$this->output->writeln(PHP_EOL); |
286
|
11 |
|
|
287
|
|
|
if ($this->input->getOption('sql')) { |
288
|
10 |
|
$this->output->writeln('<comment>Queries:</comment>'); |
289
|
11 |
|
$this->output->writeln(implode(PHP_EOL, $queries)); |
290
|
|
|
$this->output->writeln(PHP_EOL); |
291
|
|
|
} |
292
|
|
|
|
293
|
|
|
return true; |
294
|
|
|
} |
295
|
|
|
|
296
|
|
|
/** |
297
|
|
|
* Define the total number of records to process for progress bar |
298
|
|
|
*/ |
299
|
|
|
private function getTotal(string $table): int |
300
|
10 |
|
{ |
301
|
|
|
$config = $this->reader->getEntityConfig($table); |
302
|
10 |
|
$limit = (int) $config['limit']; |
303
|
2 |
|
if ($config['action'] === 'insert') { |
304
|
2 |
|
return empty($limit) ? 100 : $limit; |
305
|
2 |
|
} |
306
|
|
|
|
307
|
|
|
$rows = $this->dbUtils->countResults($table); |
308
|
10 |
|
if (empty($limit)) { |
309
|
|
|
return $rows; |
310
|
|
|
} |
311
|
|
|
|
312
|
|
|
if (! empty($limit) && $limit > $rows) { |
313
|
|
|
return $rows; |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
return $limit; |
317
|
|
|
} |
318
|
|
|
} |
319
|
|
|
|