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\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
|
|
|
/** |
30
|
|
|
* Command to launch an anonymization based on a config file |
31
|
|
|
*/ |
32
|
|
|
class AnonRunCommand extends Command |
33
|
|
|
{ |
34
|
|
|
use DBTrait; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Anonymizer DB Interface |
38
|
|
|
* @var Inet\Neuralyzer\Anonymizer\DB |
39
|
|
|
*/ |
40
|
|
|
private $anon; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* Set the command shortcut to be used in configuration |
44
|
|
|
* |
45
|
|
|
* @var string |
46
|
|
|
*/ |
47
|
|
|
private $command = 'run'; |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* @var InputInterface |
51
|
|
|
*/ |
52
|
|
|
private $input; |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* @var OutputInterface |
56
|
|
|
*/ |
57
|
|
|
private $output; |
58
|
|
|
|
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* Configure the command |
62
|
|
|
* |
63
|
|
|
* @return void |
64
|
|
|
*/ |
65
|
4 |
|
protected function configure() |
66
|
|
|
{ |
67
|
|
|
// First command : Test the DB Connexion |
68
|
4 |
|
$this->setName($this->command) |
69
|
4 |
|
->setDescription( |
70
|
4 |
|
'Generate configuration for the Anonymizer' |
71
|
4 |
|
)->setHelp( |
72
|
4 |
|
'This command will connect to a DB and run the anonymizer from a yaml config' . PHP_EOL . |
73
|
4 |
|
"Usage: ./bin/anon <info>{$this->command} -u app -p app -f anon.yml</info>" |
74
|
4 |
|
)->addOption( |
75
|
4 |
|
'host', |
76
|
4 |
|
null, |
77
|
4 |
|
InputOption::VALUE_REQUIRED, |
78
|
4 |
|
'Host', |
79
|
4 |
|
'127.0.0.1' |
80
|
4 |
|
)->addOption( |
81
|
4 |
|
'db', |
82
|
4 |
|
'd', |
83
|
4 |
|
InputOption::VALUE_REQUIRED, |
84
|
4 |
|
'Database Name' |
85
|
4 |
|
)->addOption( |
86
|
4 |
|
'user', |
87
|
4 |
|
'u', |
88
|
4 |
|
InputOption::VALUE_REQUIRED, |
89
|
4 |
|
'User Name', |
90
|
|
|
get_current_user() |
91
|
4 |
|
)->addOption( |
92
|
4 |
|
'password', |
93
|
4 |
|
'p', |
94
|
4 |
|
InputOption::VALUE_REQUIRED, |
95
|
4 |
|
'Password (or prompted)' |
96
|
4 |
|
)->addOption( |
97
|
4 |
|
'config', |
98
|
4 |
|
'c', |
99
|
4 |
|
InputOption::VALUE_REQUIRED, |
100
|
4 |
|
'Configuration File', |
101
|
4 |
|
'anon.yml' |
102
|
4 |
|
)->addOption( |
103
|
4 |
|
'pretend', |
104
|
4 |
|
null, |
105
|
4 |
|
InputOption::VALUE_NONE, |
106
|
4 |
|
"Don't run the queries" |
107
|
4 |
|
)->addOption( |
108
|
4 |
|
'sql', |
109
|
4 |
|
null, |
110
|
4 |
|
InputOption::VALUE_NONE, |
111
|
4 |
|
'Display the SQL' |
112
|
|
|
); |
113
|
4 |
|
} |
114
|
|
|
|
115
|
|
|
/** |
116
|
|
|
* Execute the command |
117
|
|
|
* |
118
|
|
|
* @param InputInterface $input |
119
|
|
|
* @param OutputInterface $output |
120
|
|
|
* |
121
|
|
|
* @return void |
122
|
|
|
*/ |
123
|
4 |
|
protected function execute(InputInterface $input, OutputInterface $output) |
124
|
|
|
{ |
125
|
4 |
|
$password = $input->getOption('password'); |
126
|
4 |
|
if (is_null($password)) { |
127
|
|
|
$question = new Question('Password: '); |
128
|
|
|
$question->setHidden(true)->setHiddenFallback(false); |
129
|
|
|
|
130
|
|
|
$password = $this->getHelper('question')->ask($input, $output, $question); |
|
|
|
|
131
|
|
|
} |
132
|
|
|
|
133
|
4 |
|
$this->connectToDB( |
134
|
4 |
|
$input->getOption('host'), |
135
|
4 |
|
$input->getOption('db'), |
136
|
4 |
|
$input->getOption('user'), |
137
|
|
|
$password |
138
|
|
|
); |
139
|
|
|
|
140
|
2 |
|
$this->input = $input; |
141
|
2 |
|
$this->output = $output; |
142
|
|
|
|
143
|
|
|
// Anon READER |
144
|
2 |
|
$reader = new \Inet\Neuralyzer\Configuration\Reader($input->getOption('config')); |
145
|
|
|
|
146
|
|
|
// Now work on the DB |
147
|
2 |
|
$this->anon = new \Inet\Neuralyzer\Anonymizer\DB($this->pdo); |
|
|
|
|
148
|
2 |
|
$this->anon->setConfiguration($reader); |
149
|
|
|
|
150
|
2 |
|
$stopwatch = new Stopwatch(); |
151
|
2 |
|
$stopwatch->start('Neuralyzer'); |
152
|
|
|
// Get tables |
153
|
2 |
|
$tables = $reader->getEntities(); |
154
|
2 |
|
foreach ($tables as $table) { |
155
|
2 |
|
$this->anonymizeTable($table, $input, $output); |
|
|
|
|
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
// Get memory and execution time information |
159
|
1 |
|
$event = $stopwatch->stop('Neuralyzer'); |
160
|
1 |
|
$memory = round($event->getMemory() / 1024 / 1024, 2); |
161
|
1 |
|
$time = round($event->getDuration() / 1000, 2); |
162
|
1 |
|
$time = ($time > 180 ? round($time / 60, 2) . 'mins' : "$time sec"); |
163
|
|
|
|
164
|
1 |
|
$output->writeln("<info>Done in $time using $memory Mb of memory</info>"); |
165
|
1 |
|
} |
166
|
|
|
|
167
|
|
|
/** |
168
|
|
|
* Anonmyze a specific table and display info about the job |
169
|
|
|
* |
170
|
|
|
* @param string $table |
171
|
|
|
*/ |
172
|
2 |
|
private function anonymizeTable(string $table) |
173
|
|
|
{ |
174
|
2 |
|
$total = $this->countRecords($table); |
175
|
1 |
|
if ($total === 0) { |
176
|
|
|
$this->output->writeln("<info>$table is empty</info>"); |
177
|
|
|
return; |
178
|
|
|
} |
179
|
|
|
|
180
|
1 |
|
$bar = new ProgressBar($this->output, $total); |
181
|
1 |
|
$bar->setRedrawFrequency($total > 100 ? 100 : 0); |
182
|
|
|
|
183
|
1 |
|
$this->output->writeln("<info>Anonymizing $table</info>"); |
184
|
|
|
|
185
|
|
|
try { |
186
|
1 |
|
$queries = $this->anon->processEntity($table, function () use ($bar) { |
187
|
1 |
|
$bar->advance(); |
188
|
1 |
|
}, $this->input->getOption('pretend'), $this->input->getOption('sql')); |
189
|
|
|
} catch (\Exception $e) { |
190
|
|
|
$msg = "<error>Error anonymizing $table. Message was : " . $e->getMessage() . "</error>"; |
191
|
|
|
$this->output->writeln(PHP_EOL . $msg . PHP_EOL); |
192
|
|
|
return; |
193
|
|
|
} |
194
|
|
|
|
195
|
1 |
|
$this->output->writeln(PHP_EOL); |
196
|
|
|
|
197
|
1 |
|
if ($this->input->getOption('sql')) { |
198
|
|
|
$this->output->writeln('<comment>Queries:</comment>'); |
199
|
|
|
$this->output->writeln(implode(PHP_EOL, $queries)); |
200
|
|
|
$this->output->writeln(PHP_EOL); |
201
|
|
|
} |
202
|
1 |
|
} |
203
|
|
|
|
204
|
|
|
/** |
205
|
|
|
* Count records on a table |
206
|
|
|
* @param string $table |
207
|
|
|
* @return int |
208
|
|
|
*/ |
209
|
2 |
|
private function countRecords(string $table): int |
210
|
|
|
{ |
211
|
|
|
try { |
212
|
2 |
|
$result = $this->pdo->query("SELECT COUNT(1) FROM $table"); |
213
|
1 |
|
} catch (\Exception $e) { |
214
|
1 |
|
$msg = "Could not count records in table '$table' defined in your config"; |
215
|
1 |
|
throw new \InvalidArgumentException($msg); |
216
|
|
|
} |
217
|
|
|
|
218
|
1 |
|
$data = $result->fetchAll(\PDO::FETCH_COLUMN); |
219
|
|
|
|
220
|
1 |
|
return (int)$data[0]; |
221
|
|
|
} |
222
|
|
|
} |
223
|
|
|
|
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: