1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* neuralyzer : Data Anonymization Library and CLI Tool |
4
|
|
|
* |
5
|
|
|
* PHP Version 7.1 |
6
|
|
|
* |
7
|
|
|
* @author Emmanuel Dyan |
8
|
|
|
* @author Rémi Sauvat |
9
|
|
|
* @copyright 2018 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 Edyan\Neuralyzer\Console\Commands; |
19
|
|
|
|
20
|
|
|
use Edyan\Neuralyzer\Configuration\Reader; |
21
|
|
|
use Edyan\Neuralyzer\Utils\DBUtils; |
22
|
|
|
use Symfony\Component\Console\Command\Command; |
23
|
|
|
use Symfony\Component\Console\Helper\ProgressBar; |
24
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
25
|
|
|
use Symfony\Component\Console\Input\InputOption; |
26
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
27
|
|
|
use Symfony\Component\Console\Question\Question; |
28
|
|
|
use Symfony\Component\Stopwatch\Stopwatch; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* Command to launch an anonymization based on a config file |
32
|
|
|
*/ |
33
|
|
|
class RunCommand extends Command |
34
|
|
|
{ |
35
|
|
|
/** |
36
|
|
|
* Store the DB Object |
37
|
|
|
* |
38
|
|
|
* @var \Edyan\Neuralyzer\Anonymizer\DB |
39
|
|
|
*/ |
40
|
|
|
private $db; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* Set the command shortcut to be used in configuration |
44
|
|
|
* |
45
|
|
|
* @var string |
46
|
|
|
*/ |
47
|
|
|
private $command = 'run'; |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* Symfony's Input Class for parameters and options |
51
|
|
|
* |
52
|
|
|
* @var InputInterface |
53
|
|
|
*/ |
54
|
|
|
private $input; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* Symfony's Output Class to display info |
58
|
|
|
* |
59
|
|
|
* @var OutputInterface |
60
|
|
|
*/ |
61
|
|
|
private $output; |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* Neuralyzer reader |
65
|
|
|
* |
66
|
|
|
* @var Reader |
67
|
|
|
*/ |
68
|
|
|
private $reader; |
69
|
|
|
|
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* Configure the command |
73
|
|
|
* |
74
|
|
|
* @return void |
75
|
|
|
*/ |
76
|
27 |
|
protected function configure(): void |
77
|
|
|
{ |
78
|
|
|
// First command : Test the DB Connexion |
79
|
27 |
|
$this->setName($this->command) |
80
|
27 |
|
->setDescription('Run Anonymizer') |
81
|
27 |
|
->setHelp( |
82
|
27 |
|
'This command will connect to a DB and run the anonymizer from a yaml config' . PHP_EOL . |
83
|
27 |
|
"Usage: <info>./bin/neuralyzer {$this->command} -u app -p app -f neuralyzer.yml</info>" |
84
|
27 |
|
)->addOption( |
85
|
27 |
|
'driver', |
86
|
27 |
|
'D', |
87
|
27 |
|
InputOption::VALUE_REQUIRED, |
88
|
27 |
|
'Driver (check Doctrine documentation to have the list)', |
89
|
27 |
|
'pdo_mysql' |
90
|
27 |
|
)->addOption( |
91
|
27 |
|
'host', |
92
|
27 |
|
'H', |
93
|
27 |
|
InputOption::VALUE_REQUIRED, |
94
|
27 |
|
'Host', |
95
|
27 |
|
'127.0.0.1' |
96
|
27 |
|
)->addOption( |
97
|
27 |
|
'db', |
98
|
27 |
|
'd', |
99
|
27 |
|
InputOption::VALUE_REQUIRED, |
100
|
27 |
|
'Database Name' |
101
|
27 |
|
)->addOption( |
102
|
27 |
|
'user', |
103
|
27 |
|
'u', |
104
|
27 |
|
InputOption::VALUE_REQUIRED, |
105
|
27 |
|
'User Name', |
106
|
27 |
|
get_current_user() |
107
|
27 |
|
)->addOption( |
108
|
27 |
|
'password', |
109
|
27 |
|
'p', |
110
|
27 |
|
InputOption::VALUE_REQUIRED, |
111
|
27 |
|
'Password (or prompted)' |
112
|
27 |
|
)->addOption( |
113
|
27 |
|
'config', |
114
|
27 |
|
'c', |
115
|
27 |
|
InputOption::VALUE_REQUIRED, |
116
|
27 |
|
'Configuration File', |
117
|
27 |
|
'neuralyzer.yml' |
118
|
27 |
|
)->addOption( |
119
|
27 |
|
'table', |
120
|
27 |
|
't', |
121
|
27 |
|
InputOption::VALUE_REQUIRED, |
122
|
27 |
|
'Do a single table' |
123
|
27 |
|
)->addOption( |
124
|
27 |
|
'pretend', |
125
|
27 |
|
null, |
126
|
27 |
|
InputOption::VALUE_NONE, |
127
|
27 |
|
"Don't run the queries" |
128
|
27 |
|
)->addOption( |
129
|
27 |
|
'sql', |
130
|
27 |
|
's', |
131
|
27 |
|
InputOption::VALUE_NONE, |
132
|
27 |
|
'Display the SQL' |
133
|
27 |
|
)->addOption( |
134
|
27 |
|
'limit', |
135
|
27 |
|
'l', |
136
|
27 |
|
InputOption::VALUE_REQUIRED, |
137
|
27 |
|
'Limit the number of written records (update or insert). 100 by default for insert' |
138
|
27 |
|
)->addOption( |
139
|
27 |
|
'mode', |
140
|
27 |
|
'm', |
141
|
27 |
|
InputOption::VALUE_REQUIRED, |
142
|
27 |
|
'Set the mode : batch or queries', |
143
|
27 |
|
'batch' |
144
|
|
|
); |
145
|
27 |
|
} |
146
|
|
|
|
147
|
|
|
/** |
148
|
|
|
* Execute the command |
149
|
|
|
* |
150
|
|
|
* @param InputInterface $input Symfony's Input Class for parameters and options |
151
|
|
|
* @param OutputInterface $output Symfony's Output Class to display infos |
152
|
|
|
* |
153
|
|
|
* @return void |
154
|
|
|
*/ |
155
|
17 |
|
protected function execute(InputInterface $input, OutputInterface $output): void |
156
|
|
|
{ |
157
|
|
|
// Throw an exception immediately if we dont have the required DB parameter |
158
|
17 |
|
if (empty($input->getOption('db'))) { |
159
|
1 |
|
throw new \InvalidArgumentException('Database name is required (--db)'); |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
// Throw an exception immediately if we dont have the right mode |
163
|
16 |
|
if (!in_array($input->getOption('mode'), ['queries', 'batch'])) { |
164
|
1 |
|
throw new \InvalidArgumentException('--mode could be only "queries" or "batch"'); |
165
|
|
|
} |
166
|
|
|
|
167
|
15 |
|
$password = $input->getOption('password'); |
168
|
15 |
|
if (is_null($password)) { |
169
|
4 |
|
$question = new Question('Password: '); |
170
|
4 |
|
$question->setHidden(true)->setHiddenFallback(false); |
171
|
|
|
|
172
|
4 |
|
$password = $this->getHelper('question')->ask($input, $output, $question); |
173
|
|
|
} |
174
|
|
|
|
175
|
15 |
|
$this->input = $input; |
176
|
15 |
|
$this->output = $output; |
177
|
|
|
|
178
|
|
|
// Anon READER |
179
|
15 |
|
$this->reader = new Reader($input->getOption('config')); |
180
|
|
|
|
181
|
|
|
// Now work on the DB |
182
|
15 |
|
$this->db = new \Edyan\Neuralyzer\Anonymizer\DB([ |
183
|
15 |
|
'driver' => $input->getOption('driver'), |
184
|
15 |
|
'host' => $input->getOption('host'), |
185
|
15 |
|
'dbname' => $input->getOption('db'), |
186
|
15 |
|
'user' => $input->getOption('user'), |
187
|
15 |
|
'password' => $password |
188
|
|
|
]); |
189
|
14 |
|
$this->db->setConfiguration($this->reader); |
190
|
14 |
|
$this->db->setMode($this->input->getOption('mode')); |
191
|
14 |
|
$this->db->setPretend($this->input->getOption('pretend')); |
192
|
14 |
|
$this->db->setReturnRes($this->input->getOption('sql')); |
193
|
|
|
|
194
|
14 |
|
$stopwatch = new Stopwatch(); |
195
|
14 |
|
$stopwatch->start('Neuralyzer'); |
196
|
|
|
// Get tables |
197
|
14 |
|
$table = $input->getOption('table'); |
198
|
14 |
|
$tables = empty($table) ? $this->reader->getEntities() : [$table]; |
199
|
14 |
|
foreach ($tables as $table) { |
200
|
14 |
|
$this->anonymizeTable($table); |
201
|
|
|
} |
202
|
|
|
|
203
|
|
|
// Get memory and execution time information |
204
|
12 |
|
$event = $stopwatch->stop('Neuralyzer'); |
205
|
12 |
|
$memory = round($event->getMemory() / 1024 / 1024, 2); |
206
|
12 |
|
$time = round($event->getDuration() / 1000, 2); |
207
|
12 |
|
$time = ($time > 180 ? round($time / 60, 2) . 'mins' : "$time sec"); |
208
|
|
|
|
209
|
12 |
|
$output->writeln("<info>Done in $time using $memory Mb of memory</info>"); |
210
|
12 |
|
} |
211
|
|
|
|
212
|
|
|
/** |
213
|
|
|
* Anonmyze a specific table and display info about the job |
214
|
|
|
* |
215
|
|
|
* @param string $table |
216
|
|
|
*/ |
217
|
14 |
|
private function anonymizeTable(string $table): void |
218
|
|
|
{ |
219
|
14 |
|
$total = $this->getTotal($table); |
220
|
12 |
|
if ($total === 0) { |
221
|
2 |
|
$this->output->writeln("<info>$table is empty</info>"); |
222
|
2 |
|
return; |
223
|
|
|
} |
224
|
|
|
|
225
|
10 |
|
$bar = new ProgressBar($this->output, $total); |
226
|
10 |
|
$bar->setRedrawFrequency($total > 100 ? 100 : 0); |
227
|
|
|
|
228
|
10 |
|
$this->output->writeln("<info>Anonymizing $table</info>"); |
229
|
|
|
|
230
|
|
|
try { |
231
|
10 |
|
$queries = $this->db->setLimit($total)->processEntity($table, function () use ($bar) { |
232
|
10 |
|
$bar->advance(); |
233
|
10 |
|
}); |
234
|
|
|
// @codeCoverageIgnoreStart |
235
|
|
|
} catch (\Exception $e) { |
236
|
|
|
$msg = "<error>Error anonymizing $table. Message was : " . $e->getMessage() . "</error>"; |
237
|
|
|
$this->output->writeln(PHP_EOL . $msg . PHP_EOL); |
238
|
|
|
return; |
239
|
|
|
} |
240
|
|
|
// @codeCoverageIgnoreEnd |
241
|
|
|
|
242
|
10 |
|
$this->output->writeln(PHP_EOL); |
243
|
|
|
|
244
|
10 |
|
if ($this->input->getOption('sql')) { |
245
|
2 |
|
$this->output->writeln('<comment>Queries:</comment>'); |
246
|
2 |
|
$this->output->writeln(implode(PHP_EOL, $queries)); |
247
|
2 |
|
$this->output->writeln(PHP_EOL); |
248
|
|
|
} |
249
|
10 |
|
} |
250
|
|
|
|
251
|
|
|
|
252
|
|
|
/** |
253
|
|
|
* Define the total number of records to process for progress bar |
254
|
|
|
* |
255
|
|
|
* @param string $table |
256
|
|
|
* @return int |
257
|
|
|
*/ |
258
|
14 |
|
private function getTotal(string $table): int |
259
|
|
|
{ |
260
|
14 |
|
$limit = (int)$this->input->getOption('limit'); |
261
|
14 |
|
$config = $this->reader->getEntityConfig($table); |
262
|
14 |
|
if ($config['action'] === 'insert') { |
263
|
2 |
|
return empty($limit) ? 100 : $limit; |
264
|
|
|
} |
265
|
|
|
|
266
|
12 |
|
$rows = (new DBUtils($this->db->getConn()))->countResults($table); |
267
|
10 |
|
if (empty($limit)) { |
268
|
6 |
|
return $rows; |
269
|
|
|
} |
270
|
|
|
|
271
|
4 |
|
if (!empty($limit) && $limit > $rows) { |
272
|
2 |
|
return $rows; |
273
|
|
|
} |
274
|
|
|
|
275
|
2 |
|
return $limit; |
276
|
|
|
} |
277
|
|
|
} |
278
|
|
|
|