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 Symfony\Component\Console\Command\Command; |
21
|
|
|
use Symfony\Component\Console\Helper\ProgressBar; |
22
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
23
|
|
|
use Symfony\Component\Console\Input\InputOption; |
24
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
25
|
|
|
use Symfony\Component\Console\Question\Question; |
26
|
|
|
use Symfony\Component\Stopwatch\Stopwatch; |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* Command to launch an anonymization based on a config file |
30
|
|
|
*/ |
31
|
|
|
class RunCommand extends Command |
32
|
|
|
{ |
33
|
|
|
/** |
34
|
|
|
* Store the DB Object |
35
|
|
|
* |
36
|
|
|
* @var \Edyan\Neuralyzer\Anonymizer\DB |
37
|
|
|
*/ |
38
|
|
|
private $db; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* Set the command shortcut to be used in configuration |
42
|
|
|
* |
43
|
|
|
* @var string |
44
|
|
|
*/ |
45
|
|
|
private $command = 'run'; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* Symfony's Input Class for parameters and options |
49
|
|
|
* |
50
|
|
|
* @var InputInterface |
51
|
|
|
*/ |
52
|
|
|
private $input; |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* Symfony's Output Class to display info |
56
|
|
|
* |
57
|
|
|
* @var OutputInterface |
58
|
|
|
*/ |
59
|
|
|
private $output; |
60
|
|
|
|
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* Configure the command |
64
|
|
|
* |
65
|
|
|
* @return void |
66
|
|
|
*/ |
67
|
|
|
protected function configure() |
68
|
|
|
{ |
69
|
|
|
// First command : Test the DB Connexion |
70
|
|
|
$this->setName($this->command) |
71
|
|
|
->setDescription('Generate configuration for the Anonymizer') |
72
|
|
|
->setHelp( |
73
|
|
|
'This command will connect to a DB and run the anonymizer from a yaml config' . PHP_EOL . |
74
|
|
|
"Usage: ./bin/neuralyzer <info>{$this->command} -u app -p app -f neuralyzer.yml</info>" |
75
|
13 |
|
)->addOption( |
76
|
|
|
'driver', |
77
|
|
|
'D', |
78
|
13 |
|
InputOption::VALUE_REQUIRED, |
79
|
13 |
|
'Driver (check Doctrine documentation to have the list)', |
80
|
13 |
|
'pdo_mysql' |
81
|
13 |
|
)->addOption( |
82
|
13 |
|
'host', |
83
|
13 |
|
'H', |
84
|
13 |
|
InputOption::VALUE_REQUIRED, |
85
|
13 |
|
'Host', |
86
|
13 |
|
'127.0.0.1' |
87
|
13 |
|
)->addOption( |
88
|
13 |
|
'db', |
89
|
13 |
|
'd', |
90
|
13 |
|
InputOption::VALUE_REQUIRED, |
91
|
13 |
|
'Database Name' |
92
|
13 |
|
)->addOption( |
93
|
13 |
|
'user', |
94
|
13 |
|
'u', |
95
|
13 |
|
InputOption::VALUE_REQUIRED, |
96
|
13 |
|
'User Name', |
97
|
13 |
|
get_current_user() |
98
|
13 |
|
)->addOption( |
99
|
13 |
|
'password', |
100
|
13 |
|
'p', |
101
|
13 |
|
InputOption::VALUE_REQUIRED, |
102
|
13 |
|
'Password (or prompted)' |
103
|
13 |
|
)->addOption( |
104
|
13 |
|
'config', |
105
|
13 |
|
'c', |
106
|
13 |
|
InputOption::VALUE_REQUIRED, |
107
|
13 |
|
'Configuration File', |
108
|
13 |
|
'neuralyzer.yml' |
109
|
13 |
|
)->addOption( |
110
|
13 |
|
'table', |
111
|
13 |
|
't', |
112
|
13 |
|
InputOption::VALUE_REQUIRED, |
113
|
13 |
|
'Do a single table' |
114
|
13 |
|
)->addOption( |
115
|
13 |
|
'pretend', |
116
|
13 |
|
null, |
117
|
13 |
|
InputOption::VALUE_NONE, |
118
|
13 |
|
"Don't run the queries" |
119
|
13 |
|
)->addOption( |
120
|
13 |
|
'sql', |
121
|
13 |
|
null, |
122
|
13 |
|
InputOption::VALUE_NONE, |
123
|
13 |
|
'Display the SQL' |
124
|
13 |
|
); |
125
|
13 |
|
} |
126
|
13 |
|
|
127
|
13 |
|
/** |
128
|
13 |
|
* Execute the command |
129
|
13 |
|
* |
130
|
13 |
|
* @param InputInterface $input Symfony's Input Class for parameters and options |
131
|
13 |
|
* @param OutputInterface $output Symfony's Output Class to display infos |
132
|
13 |
|
* |
133
|
13 |
|
* @return void |
134
|
13 |
|
*/ |
135
|
13 |
|
protected function execute(InputInterface $input, OutputInterface $output) |
136
|
13 |
|
{ |
137
|
|
|
// Throw an exception immediately if we dont have the required DB parameter |
138
|
13 |
|
if (empty($input->getOption('db'))) { |
139
|
|
|
throw new \InvalidArgumentException('Database name is required (--db)'); |
140
|
|
|
} |
141
|
|
|
|
142
|
|
|
$password = $input->getOption('password'); |
143
|
|
|
if (is_null($password)) { |
144
|
|
|
$question = new Question('Password: '); |
145
|
|
|
$question->setHidden(true)->setHiddenFallback(false); |
146
|
|
|
|
147
|
|
|
$password = $this->getHelper('question')->ask($input, $output, $question); |
|
|
|
|
148
|
7 |
|
} |
149
|
|
|
|
150
|
|
|
$this->input = $input; |
151
|
7 |
|
$this->output = $output; |
152
|
1 |
|
|
153
|
|
|
// Anon READER |
154
|
|
|
$reader = new \Edyan\Neuralyzer\Configuration\Reader($input->getOption('config')); |
155
|
6 |
|
|
156
|
6 |
|
// Now work on the DB |
157
|
2 |
|
$this->db = new \Edyan\Neuralyzer\Anonymizer\DB([ |
158
|
2 |
|
'driver' => $input->getOption('driver'), |
159
|
|
|
'host' => $input->getOption('host'), |
160
|
2 |
|
'dbname' => $input->getOption('db'), |
161
|
|
|
'user' => $input->getOption('user'), |
162
|
|
|
'password' => $password, |
163
|
6 |
|
]); |
164
|
6 |
|
$this->db->setConfiguration($reader); |
165
|
|
|
|
166
|
|
|
$stopwatch = new Stopwatch(); |
167
|
6 |
|
$stopwatch->start('Neuralyzer'); |
168
|
|
|
// Get tables |
169
|
|
|
$tables = empty($input->getOption('table')) ? $reader->getEntities() : [$input->getOption('table')]; |
170
|
6 |
|
foreach ($tables as $table) { |
171
|
6 |
|
$this->anonymizeTable($table, $input, $output); |
|
|
|
|
172
|
6 |
|
} |
173
|
6 |
|
|
174
|
6 |
|
// Get memory and execution time information |
175
|
6 |
|
$event = $stopwatch->stop('Neuralyzer'); |
176
|
|
|
$memory = round($event->getMemory() / 1024 / 1024, 2); |
177
|
5 |
|
$time = round($event->getDuration() / 1000, 2); |
178
|
|
|
$time = ($time > 180 ? round($time / 60, 2) . 'mins' : "$time sec"); |
179
|
5 |
|
|
180
|
5 |
|
$output->writeln("<info>Done in $time using $memory Mb of memory</info>"); |
181
|
|
|
} |
182
|
5 |
|
|
183
|
5 |
|
/** |
184
|
5 |
|
* Anonmyze a specific table and display info about the job |
185
|
5 |
|
* |
186
|
|
|
* @param string $table |
187
|
|
|
*/ |
188
|
|
|
private function anonymizeTable(string $table) |
189
|
3 |
|
{ |
190
|
3 |
|
$total = $this->countRecords($table); |
191
|
3 |
|
if ($total === 0) { |
192
|
3 |
|
$this->output->writeln("<info>$table is empty</info>"); |
193
|
|
|
return; |
194
|
3 |
|
} |
195
|
3 |
|
|
196
|
|
|
$bar = new ProgressBar($this->output, $total); |
197
|
|
|
$bar->setRedrawFrequency($total > 100 ? 100 : 0); |
198
|
|
|
|
199
|
|
|
$this->output->writeln("<info>Anonymizing $table</info>"); |
200
|
|
|
|
201
|
|
|
try { |
202
|
5 |
|
$queries = $this->db->processEntity($table, function () use ($bar) { |
203
|
|
|
$bar->advance(); |
204
|
5 |
|
}, $this->input->getOption('pretend'), $this->input->getOption('sql')); |
205
|
3 |
|
// @codeCoverageIgnoreStart |
206
|
1 |
|
} catch (\Exception $e) { |
207
|
1 |
|
$msg = "<error>Error anonymizing $table. Message was : " . $e->getMessage() . "</error>"; |
208
|
|
|
$this->output->writeln(PHP_EOL . $msg . PHP_EOL); |
209
|
|
|
return; |
210
|
2 |
|
} |
211
|
2 |
|
// @codeCoverageIgnoreEnd |
212
|
|
|
|
213
|
2 |
|
$this->output->writeln(PHP_EOL); |
214
|
|
|
|
215
|
|
|
if ($this->input->getOption('sql')) { |
216
|
2 |
|
$this->output->writeln('<comment>Queries:</comment>'); |
217
|
2 |
|
$this->output->writeln(implode(PHP_EOL, $queries)); |
218
|
2 |
|
$this->output->writeln(PHP_EOL); |
219
|
|
|
} |
220
|
|
|
} |
221
|
|
|
|
222
|
|
|
/** |
223
|
|
|
* Count records on a table |
224
|
|
|
* @param string $table |
225
|
|
|
* @return int |
226
|
|
|
*/ |
227
|
2 |
|
private function countRecords(string $table): int |
228
|
|
|
{ |
229
|
2 |
|
try { |
230
|
1 |
|
$stmt = $this->db->getConn()->prepare("SELECT COUNT(1) AS total FROM $table"); |
231
|
1 |
|
$stmt->execute(); |
232
|
1 |
|
} catch (\Exception $e) { |
233
|
|
|
$msg = "Could not count records in '$table' from your config : " . $e->getMessage(); |
234
|
2 |
|
throw new \InvalidArgumentException($msg); |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
$data = $stmt->fetchAll(); |
238
|
|
|
|
239
|
|
|
return (int)$data[0]['total']; |
240
|
|
|
} |
241
|
|
|
} |
242
|
|
|
|
Let’s take a look at an example:
In the above example, the authenticate() method works fine as long as you just pass instances of MyUser. However, if you now also want to pass a different implementation of User which does not have a getDisplayName() method, the code will break.
Available Fixes
Change the type-hint for the parameter:
Add an additional type-check:
Add the method to the interface: