AbstractAnonymizer::generateFakeData() - Code Metrics - Inspection of "Allow the usage of unique faker data" - edyan/neuralyzer - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#15)

by Jeroen

created 2018-11-27 15:48 UTC

AbstractAnonymizer::generateFakeData() B

↳ Parent: AbstractAnonymizer

Complexity

Conditions	7
Paths	4

Size

Total Lines	30
Code Lines	22

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	17
CRAP Score	7

Importance

Changes

Metric	Value
cc	7
eloc	22
nc	4
nop	0
dl	0
loc	30
ccs	17
cts	17
cp	1
crap	7
rs	8.6346
c	0
b	0
f	0

<?php
/**
 * neuralyzer : Data Anonymization Library and CLI Tool
 *
 * PHP Version 7.1
 *
 * @author    Emmanuel Dyan
 * @author    Rémi Sauvat
 * @copyright 2018 Emmanuel Dyan
 *
 * @package edyan/neuralyzer
 *
 * @license GNU General Public License v2.0
 *
 * @link https://github.com/edyan/neuralyzer
 */

namespace Edyan\Neuralyzer\Anonymizer;

use Edyan\Neuralyzer\Configuration\Reader;
use Edyan\Neuralyzer\Exception\NeuralyzerConfigurationException;

/**
 * Abstract Anonymizer, that can be implemented as DB Anonymizer for example
 * Its goal is only to anonymize any data, from a simple array
 * not to write or read it from anywhere
 *
 */
abstract class AbstractAnonymizer
{
    /**
     * Update data into table
     */
    public const UPDATE_TABLE = 1;

    /**
     * Insert data into table
     */
    public const INSERT_TABLE = 2;

    /**
     * Set the batch size for updates
     *
     * @var int
     */
    protected $batchSize = 1000;

    /**
     * Contains the configuration object
     *
     * @var Reader
     */
    protected $configuration;

    /**
     * Configuration of entities
     *
     * @var array
     */
    protected $configEntities = [];

    /**
     * List of used fakers
     *
     * @var array
     */
    protected $fakers = [];

    /**
     * Current table (entity) to process
     *
     * @var string
     */
    protected $entity;

    /**
     * Current table (entity) Columns
     *
     * @var array
     */
    protected $entityCols;

    /**
     * Limit the number of updates or create
     *
     * @var int
     */
    protected $limit = 0;

    /**
     * Pretend we do the update, but do nothing
     *
     * @var bool
     */
    protected $pretend = true;

    /**
     * Return the generated SQL
     *
     * @var bool
     */
    protected $returnRes = false;

    /**
     * @var \Faker\Generator
     */
    protected $faker;


    /**
     * Process the entity according to the anonymizer type
     *
     * @param string        $entity   Entity's name
     * @param callable|null $callback Callback function with current row num as parameter
     *
     * @return array
     */
    abstract public function processEntity(
        string $entity,
        callable $callback = null
    ): array;


    /**
     * Set the configuration
     *
     * @param Reader $configuration
     */
    public function setConfiguration(Reader $configuration): void
    {
        $this->configuration = $configuration;
        $this->configEntities = $configuration->getConfigValues()['entities'];
        $this->initFaker();
    }


    /**
     * Limit of fake generated records for updates and creates
     *
     * @param int $limit
     *
     * @return mixed
     */
    public function setLimit(int $limit)
    {
        $this->limit = $limit;
        if ($this->limit < $this->batchSize) {
            $this->batchSize = $this->limit;
        }

        return $this;
    }


    /**
     * Activate or deactivate the pretending mode (dry run)
     *
     * @param  bool $pretend
     *
     * @return mixed
     */
    public function setPretend(bool $pretend)
    {
        $this->pretend = $pretend;

        return $this;
    }


    /**
     * Return or not a result (like an SQL Query that has
     * been generated with fake data)
     *
     * @param  bool $returnRes
     *
     * @return mixed
     */
    public function setReturnRes(bool $returnRes)
    {
        $this->returnRes = $returnRes;

        return $this;
    }


    /**
     * Evaluate, from the configuration if I have to update or Truncate the table
     *
     * @return int
     * @throws NeuralyzerConfigurationException
     */
    protected function whatToDoWithEntity(): int
    {
        $this->checkEntityIsInConfig();

        $entityConfig = $this->configEntities[$this->entity];

        $actions = 0;
        if (array_key_exists('cols', $entityConfig)) {
            switch ($entityConfig['action']) {
                case 'update':
                    $actions |= self::UPDATE_TABLE;
                    break;
                case 'insert':
                    $actions |= self::INSERT_TABLE;
                    break;
            }
        }

        return $actions;
    }

    /**
     * Generate fake data for an entity and return it as an Array
     *
     * @return array
     * @throws NeuralyzerConfigurationException
     */
    protected function generateFakeData(): array
    {
        $this->checkEntityIsInConfig();
        $language = $this->configuration->getConfigValues()['language'];
        $faker = \Faker\Factory::create($language);
        $faker->addProvider(new \Edyan\Neuralyzer\Faker\Provider\Base($faker));
        $faker->addProvider(new \Edyan\Neuralyzer\Faker\Provider\UniqueWord($faker, $language));
        $colsInConfig = $this->configEntities[$this->entity]['cols'];
        $fakerInstanceUnique = $faker->unique();
        $fakerInstance = $faker;
        $row = [];
        foreach ($colsInConfig as $colName => $colProps) {
            $this->checkColIsInEntity($colName);
            $data = \call_user_func_array(
                [isset($colProps['unique']) && $colProps['unique'] === true ? $fakerInstanceUnique : $fakerInstance, $colProps['method']],
                $colProps['params']
            );
            if (!is_scalar($data)) {
                $msg = "You must use faker methods that generate strings: '{$colProps['method']}' forbidden";
                throw new NeuralyzerConfigurationException($msg);
            }
            $row[$colName] = trim($data);
            $colLength = $this->entityCols[$colName]['length'];
            // Cut the value if too long ...
            if (!empty($colLength) && \strlen($row[$colName]) > $colLength) {
                $row[$colName] = substr($row[$colName], 0, $colLength - 1);
            }
        }

        return $row;
    }


    /**
     * Make sure that entity is defined in the configuration
     *
     * @throws NeuralyzerConfigurationException
     */
    protected function checkEntityIsInConfig(): void
    {
        if (empty($this->configEntities)) {
            throw new NeuralyzerConfigurationException(
                'No entities found. Have you loaded a configuration file ?'
            );
        }
        if (!array_key_exists($this->entity, $this->configEntities)) {
            throw new NeuralyzerConfigurationException(
                "No configuration for that entity ({$this->entity})"
            );
        }
    }

    /**
     * Verify a column is defined in the real entityCols
     *
     * @param string $colName
     *
     * @throws NeuralyzerConfigurationException
     */
    protected function checkColIsInEntity(string $colName): void
    {
        if (!array_key_exists($colName, $this->entityCols)) {
            throw new NeuralyzerConfigurationException("Col $colName does not exist");
        }
    }

    /**
     * Init Faker and add additional methods
     */
    protected function initFaker(): void
    {
        $language = $this->configuration->getConfigValues()['language'];
        $this->faker = \Faker\Factory::create($language);
        $this->faker->addProvider(new \Edyan\Neuralyzer\Faker\Provider\Base($this->faker));
        $this->faker->addProvider(new \Edyan\Neuralyzer\Faker\Provider\UniqueWord($this->faker, $language));
    }
}


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\Anonymizer;
19
20		use Edyan\Neuralyzer\Configuration\Reader;
21		use Edyan\Neuralyzer\Exception\NeuralyzerConfigurationException;
22
23		/**
24		* Abstract Anonymizer, that can be implemented as DB Anonymizer for example
25		* Its goal is only to anonymize any data, from a simple array
26		* not to write or read it from anywhere
27		*
28		*/
29		abstract class AbstractAnonymizer
30		{
31		/**
32		* Update data into table
33		*/
34		public const UPDATE_TABLE = 1;
35
36		/**
37		* Insert data into table
38		*/
39		public const INSERT_TABLE = 2;
40
41		/**
42		* Set the batch size for updates
43		*
44		* @var int
45		*/
46		protected $batchSize = 1000;
47
48		/**
49		* Contains the configuration object
50		*
51		* @var Reader
52		*/
53		protected $configuration;
54
55		/**
56		* Configuration of entities
57		*
58		* @var array
59		*/
60		protected $configEntities = [];
61
62		/**
63		* List of used fakers
64		*
65		* @var array
66		*/
67		protected $fakers = [];
68
69		/**
70		* Current table (entity) to process
71		*
72		* @var string
73		*/
74		protected $entity;
75
76		/**
77		* Current table (entity) Columns
78		*
79		* @var array
80		*/
81		protected $entityCols;
82
83		/**
84		* Limit the number of updates or create
85		*
86		* @var int
87		*/
88		protected $limit = 0;
89
90		/**
91		* Pretend we do the update, but do nothing
92		*
93		* @var bool
94		*/
95		protected $pretend = true;
96
97		/**
98		* Return the generated SQL
99		*
100		* @var bool
101		*/
102		protected $returnRes = false;
103
104		/**
105		* @var \Faker\Generator
106		*/
107		protected $faker;
108
109
110		/**
111		* Process the entity according to the anonymizer type
112	32	*
113		* @param string $entity Entity's name
114	32	* @param callable\|null $callback Callback function with current row num as parameter
115	32	*
116	32	* @return array
117		*/
118		abstract public function processEntity(
119		string $entity,
120		callable $callback = null
121		): array;
122
123
124	24	/**
125		* Set the configuration
126	24	*
127	24	* @param Reader $configuration
128	23	*/
129		public function setConfiguration(Reader $configuration): void
130		{
131	24	$this->configuration = $configuration;
132		$this->configEntities = $configuration->getConfigValues()['entities'];
133		$this->initFaker();
134		}
135
136
137		/**
138		* Limit of fake generated records for updates and creates
139		*
140	29	* @param int $limit
141		*
142	29	* @return mixed
143		*/
144	29	public function setLimit(int $limit)
145		{
146		$this->limit = $limit;
147		if ($this->limit < $this->batchSize) {
148		$this->batchSize = $this->limit;
149		}
150
151		return $this;
152		}
153
154	28
155		/**
156	28	* Activate or deactivate the pretending mode (dry run)
157		*
158	28	* @param bool $pretend
159		*
160		* @return mixed
161		*/
162		public function setPretend(bool $pretend)
163		{
164		$this->pretend = $pretend;
165
166		return $this;
167	27	}
168
169	27
170		/**
171	25	* Return or not a result (like an SQL Query that has
172		* been generated with fake data)
173	25	*
174	25	* @param bool $returnRes
175	5	*
176		* @return mixed
177		*/
178	25	public function setReturnRes(bool $returnRes)
179	25	{
180	25	$this->returnRes = $returnRes;
181	19
182	19	return $this;
183	6	}
184	6
185	6
186		/**
187		* Evaluate, from the configuration if I have to update or Truncate the table
188		*
189	25	* @return int
190		* @throws NeuralyzerConfigurationException
191		*/
192		protected function whatToDoWithEntity(): int
193		{
194		$this->checkEntityIsInConfig();
195
196		$entityConfig = $this->configEntities[$this->entity];
197
198	5	$actions = 0;
199		if (array_key_exists('cols', $entityConfig)) {
200	5	switch ($entityConfig['action']) {
201		case 'update':
202	5	$actions \|= self::UPDATE_TABLE;
203	2	break;
204		case 'insert':
205		$actions \|= self::INSERT_TABLE;
206	3	break;
207		}
208		}
209
210		return $actions;
211		}
212
213		/**
214		* Generate fake data for an entity and return it as an Array
215	23	*
216		* @return array
217	23	* @throws NeuralyzerConfigurationException
218		*/
219	23	protected function generateFakeData(): array
220	23	{
221		$this->checkEntityIsInConfig();
222	23	$language = $this->configuration->getConfigValues()['language'];
223	23	$faker = \Faker\Factory::create($language);
224	23	$faker->addProvider(new \Edyan\Neuralyzer\Faker\Provider\Base($faker));
225	23	$faker->addProvider(new \Edyan\Neuralyzer\Faker\Provider\UniqueWord($faker, $language));
226	22	$colsInConfig = $this->configEntities[$this->entity]['cols'];
227	22	$fakerInstanceUnique = $faker->unique();
228	22	$fakerInstance = $faker;
229		$row = [];
230		foreach ($colsInConfig as $colName => $colProps) {
231	22	$this->checkColIsInEntity($colName);
232	1	$data = \call_user_func_array(
233	1	[isset($colProps['unique']) && $colProps['unique'] === true ? $fakerInstanceUnique : $fakerInstance, $colProps['method']],
234		$colProps['params']
235		);
236	22	if (!is_scalar($data)) {
237		$msg = "You must use faker methods that generate strings: '{$colProps['method']}' forbidden";
238	22	throw new NeuralyzerConfigurationException($msg);
239		}
240	22	$row[$colName] = trim($data);
241	22	$colLength = $this->entityCols[$colName]['length'];
242		// Cut the value if too long ...
243		if (!empty($colLength) && \strlen($row[$colName]) > $colLength) {
244		$row[$colName] = substr($row[$colName], 0, $colLength - 1);
245	21	}
246		}
247
248		return $row;
249		}
250
251
252		/**
253		* Make sure that entity is defined in the configuration
254	27	*
255		* @throws NeuralyzerConfigurationException
256	27	*/
257	1	protected function checkEntityIsInConfig(): void
258	1	{
259		if (empty($this->configEntities)) {
260		throw new NeuralyzerConfigurationException(
261	26	'No entities found. Have you loaded a configuration file ?'
262	1	);
263	1	}
264		if (!array_key_exists($this->entity, $this->configEntities)) {
265		throw new NeuralyzerConfigurationException(
266	25	"No configuration for that entity ({$this->entity})"
267		);
268		}
269		}
270
271		/**
272		* Verify a column is defined in the real entityCols
273		*
274	23	* @param string $colName
275		*
276	23	* @throws NeuralyzerConfigurationException
277	1	*/
278		protected function checkColIsInEntity(string $colName): void
279	22	{
280		if (!array_key_exists($colName, $this->entityCols)) {
281		throw new NeuralyzerConfigurationException("Col $colName does not exist");
282		}
283		}
284
285		/**
286		* Init Faker and add additional methods
287		*/
288		protected function initFaker(): void
289		{
290		$language = $this->configuration->getConfigValues()['language'];
291		$this->faker = \Faker\Factory::create($language);
292		$this->faker->addProvider(new \Edyan\Neuralyzer\Faker\Provider\Base($this->faker));
293		$this->faker->addProvider(new \Edyan\Neuralyzer\Faker\Provider\UniqueWord($this->faker, $language));
294		}
295		}
296

edyan / neuralyzer

Pull Request — master (#15)

AbstractAnonymizer::generateFakeData() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like