Reader::setRecords() - Code Metrics - Inspection of "Internal code improved" - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#234)

by ignace nyamagana

created 2017-06-13 07:30 UTC

Reader::setRecords() A

↳ Parent: Reader

Complexity

Conditions	3
Paths	2

Size

Total Lines	15
Code Lines	9

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	9
CRAP Score	3

Importance

Changes

Metric	Value
cc	3
eloc	9
nc	2
nop	0
dl	0
loc	15
ccs	9
cts	9
cp	1
crap	3
rs	9.4285
c	0
b	0
f	0

<?php
/**
* This file is part of the League.csv library
*
* @license http://opensource.org/licenses/MIT
* @link https://github.com/thephpleague/csv/
* @version 9.0.0
* @package League.csv
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
declare(strict_types=1);

namespace League\Csv;

use BadMethodCallException;
use CallbackFilterIterator;
use Countable;
use Iterator;
use IteratorAggregate;
use League\Csv\Exception\RuntimeException;
use LimitIterator;
use SplFileObject;

/**
 * A class to manage records selection from a CSV document
 *
 * @package League.csv
 * @since  3.0.0
 *
 * @method array fetchAll() Returns a sequential array of all CSV records
 * @method array fetchOne(int $offset = 0) Returns a single record from the CSV
 * @method Generator fetchColumn(string|int $column_index) Returns the next value from a single CSV record field
 * @method Generator fetchPairs(string|int $offset_index, string|int $value_index) Fetches the next key-value pairs from the CSV document
 */
class Reader extends AbstractCsv implements Countable, IteratorAggregate
{
    /**
     * @inheritdoc
     */
    protected $stream_filter_mode = STREAM_FILTER_READ;

    /**
     * The value to pad if the record is less than header size.
     *
     * @var mixed
     */
    protected $record_padding_value;

    /**
     * CSV Document header offset
     *
     * @var int|null
     */
    protected $header_offset;

    /**
     * CSV Document Header record
     *
     * @var string[]
     */
    protected $header = [];

    /**
     * Records Iterator
     *
     * @var Iterator
     */
    protected $records;

    /**
     * Records count
     *
     * @var int
     */
    protected $nb_records = -1;

    /**
     * Detect Delimiters occurences in the CSV
     *
     * Returns a associative array where each key represents
     * a valid delimiter and each value the number of occurences
     *
     * @param string[] $delimiters the delimiters to consider
     * @param int      $nb_records Detection is made using $nb_records of the CSV
     *
     * @return array
     */
    public function fetchDelimitersOccurrence(array $delimiters, int $nb_records = 1): array
    {
        $filter = function ($value): bool {
            return 1 == strlen($value);
        };

        $nb_records = $this->filterMinRange($nb_records, 1, __METHOD__.'() expects the number of records to consider to be a valid positive integer, %s given');
        $delimiters = array_unique(array_filter($delimiters, $filter));
        $reducer = function (array $res, string $delimiter) use ($nb_records): array {
            $res[$delimiter] = $this->getCellCount($delimiter, $nb_records);

            return $res;
        };

        $res = array_reduce($delimiters, $reducer, []);
        arsort($res, SORT_NUMERIC);

        return $res;
    }

    /**
     * Returns the cell count for a specified delimiter
     * and a specified number of records
     *
     * @param string $delimiter  CSV delimiter
     * @param int    $nb_records CSV records to consider
     *
     * @return int
     */
    protected function getCellCount(string $delimiter, int $nb_records): int
    {
        $filter = function ($record): bool {
            return is_array($record) && count($record) > 1;
        };

        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($delimiter, $this->enclosure, $this->escape);
        $iterator = new CallbackFilterIterator(new LimitIterator($this->document, 0, $nb_records), $filter);

        return count(iterator_to_array($iterator, false), COUNT_RECURSIVE);
    }

    /**
     * Returns the record padding value
     *
     * @return mixed
     */
    public function getRecordPaddingValue()
    {
        return $this->record_padding_value;
    }

    /**
     * Returns the record offset used as header
     *
     * If no CSV record is used this method MUST return null
     *
     * @return int|null
     */
    public function getHeaderOffset()
    {
        return $this->header_offset;
    }

    /**
     * Returns wether the selected header can be combine to each record
     *
     * A valid header must be empty or contains unique string field names
     *
     * @return bool
     */
    public function supportsHeaderAsRecordKeys(): bool
    {
        $header = $this->getHeader();

        return empty($header) || $header === array_unique(array_filter($header, 'is_string'));
    }

    /**
     * Returns the CSV record header
     *
     * The returned header is represented as an array of string values
     *
     * @return string[]
     */
    public function getHeader(): array
    {
        if (null === $this->header_offset) {
            return $this->header;
        }

        if (empty($this->header)) {
            $this->header = $this->setHeader($this->header_offset);
        }

        return $this->header;
    }

    /**
     * Determine the CSV record header
     *
     * @param int $offset
     *
     * @throws RuntimeException If the header offset is an integer
     *                          and the corresponding record is missing
     *                          or is an empty array
     *
     * @return string[]
     */
    protected function setHeader(int $offset): array
    {
        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
        $this->document->seek($offset);
        $header = $this->document->current();
        if (empty($header)) {
            throw new RuntimeException(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
        }

        if (0 === $offset) {
            return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
        }

        return $header;
    }

    /**
     * Strip the BOM sequence from a record
     *
     * @param string[] $record
     * @param int      $bom_length
     * @param string   $enclosure
     *
     * @return string[]
     */
    protected function removeBOM(array $record, int $bom_length, string $enclosure): array
    {
        if (0 == $bom_length) {
            return $record;
        }

        $record[0] = mb_substr($record[0], $bom_length);
        if ($enclosure == mb_substr($record[0], 0, 1) && $enclosure == mb_substr($record[0], -1, 1)) {
            $record[0] = mb_substr($record[0], 1, -1);
        }

        return $record;
    }

    /**
     * @inheritdoc
     */
    public function __call($method, array $arguments)
    {
        $whitelisted = ['fetchColumn' => 1, 'fetchPairs' => 1, 'fetchOne' => 1, 'fetchAll' => 1];
        if (isset($whitelisted[$method])) {
            return (new ResultSet($this->getRecords(), $this->getHeader()))->$method(...$arguments);
        }

        throw new BadMethodCallException(sprintf('%s::%s() method does not exist', __CLASS__, $method));
    }

    /**
     * @inheritdoc
     */
    public function count(): int
    {
        if (-1 === $this->nb_records) {
            $this->nb_records = iterator_count($this->getRecords());
        }

        return $this->nb_records;
    }

    /**
     * Returns the CSV records in an iterator object.
     *
     * @return Iterator
     */
    public function getRecords(): Iterator
    {
        $this->records = $this->records ?? $this->setRecords();

        return $this->records;
    }

    /**
     * Returns the CSV records in an iterator object.
     *
     * Each CSV record is represented as a simple array of string or null values.
     *
     * If the CSV document has a header record then each record is combined
     * to each header record and the header record is removed from the iterator.
     *
     * If the CSV document is inconsistent. Missing record fields are
     * filled with null values while extra record fields are strip from
     * the returned object.
     *
     * @throws RuntimeException If the header contains non unique column name
     *
     * @return Iterator
     */
    protected function setRecords(): Iterator
    {
        if (!$this->supportsHeaderAsRecordKeys()) {
            throw new RuntimeException('The header record must be empty or a flat array with unique string values');
        }

        $normalized = function ($record): bool {
            return is_array($record) && $record != [null];
        };
        $bom = $this->getInputBOM();
        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);

        return $this->combineHeader($this->stripBOM(new CallbackFilterIterator($this->document, $normalized), $bom));
    }

    /**
     * Add the CSV header if present and valid
     *
     * @param Iterator $iterator
     *
     * @return Iterator
     */
    protected function combineHeader(Iterator $iterator): Iterator
    {
        if (null === $this->header_offset) {
            return $iterator;
        }

        $iterator = new CallbackFilterIterator($iterator, function (array $record, int $offset): bool {
            return $offset != $this->header_offset;
        });

        $header = $this->getHeader();
        $field_count = count($header);
        $mapper = function (array $record) use ($header, $field_count): array {
            if (count($record) != $field_count) {
                $record = array_slice(array_pad($record, $field_count, $this->record_padding_value), 0, $field_count);
            }

            return array_combine($header, $record);
        };

        return new MapIterator($iterator, $mapper);
    }

    /**
     * Strip the BOM sequence if present
     *
     * @param Iterator $iterator
     * @param string   $bom
     *
     * @return Iterator
     */
    protected function stripBOM(Iterator $iterator, string $bom): Iterator
    {
        if ('' === $bom) {
            return $iterator;
        }

        $bom_length = mb_strlen($bom);
        $mapper = function (array $record, int $index) use ($bom_length): array {
            if (0 != $index) {
                return $record;
            }

            return $this->removeBOM($record, $bom_length, $this->enclosure);
        };

        return new MapIterator($iterator, $mapper);
    }

    /**
     * @inheritdoc
     */
    public function getIterator(): Iterator
    {
        return $this->getRecords();
    }

    /**
     * Set the record padding value
     *
     * @param mixed $record_padding_value
     *
     * @return static
     */
    public function setRecordPaddingValue($record_padding_value): self
    {
        $this->record_padding_value = $record_padding_value;

        return $this;
    }

    /**
     * Selects the record to be used as the CSV header
     *
     * Because of the header is represented as an array, to be valid
     * a header MUST contain only unique string value.
     *
     * @param int|null $offset the header record offset
     *
     * @return static
     */
    public function setHeaderOffset($offset): self
    {
        if (null !== $offset) {
            $offset = $this->filterMinRange($offset, 0, __METHOD__.'() expects the header offset index to be a positive integer or 0, %s given');
        }

        if ($offset !== $this->header_offset) {
            $this->header_offset = $offset;
            $this->resetProperties();
        }

        return $this;
    }

    /**
     * @inheritdoc
     */
    protected function resetProperties()
    {
        $this->nb_records = -1;
        $this->header = [];
        $this->records = null;
    }
}


1		<?php
2		/**
3		* This file is part of the League.csv library
4		*
5		* @license http://opensource.org/licenses/MIT
6		* @link https://github.com/thephpleague/csv/
7		* @version 9.0.0
8		* @package League.csv
9		*
10		* For the full copyright and license information, please view the LICENSE
11		* file that was distributed with this source code.
12		*/
13		declare(strict_types=1);
14
15		namespace League\Csv;
16
17		use BadMethodCallException;
18		use CallbackFilterIterator;
19		use Countable;
20		use Iterator;
21		use IteratorAggregate;
22		use League\Csv\Exception\RuntimeException;
23		use LimitIterator;
24		use SplFileObject;
25
26		/**
27		* A class to manage records selection from a CSV document
28		*
29		* @package League.csv
30		* @since 3.0.0
31		*
32		* @method array fetchAll() Returns a sequential array of all CSV records
33		* @method array fetchOne(int $offset = 0) Returns a single record from the CSV
34		* @method Generator fetchColumn(string\|int $column_index) Returns the next value from a single CSV record field
35		* @method Generator fetchPairs(string\|int $offset_index, string\|int $value_index) Fetches the next key-value pairs from the CSV document
36		*/
37		class Reader extends AbstractCsv implements Countable, IteratorAggregate
38		{
39		/**
40		* @inheritdoc
41		*/
42		protected $stream_filter_mode = STREAM_FILTER_READ;
43
44		/**
45		* The value to pad if the record is less than header size.
46		*
47		* @var mixed
48		*/
49		protected $record_padding_value;
50
51		/**
52		* CSV Document header offset
53		*
54		* @var int\|null
55		*/
56		protected $header_offset;
57
58		/**
59		* CSV Document Header record
60		*
61		* @var string[]
62		*/
63		protected $header = [];
64
65		/**
66		* Records Iterator
67		*
68		* @var Iterator
69		*/
70		protected $records;
71
72		/**
73		* Records count
74		*
75		* @var int
76		*/
77		protected $nb_records = -1;
78
79		/**
80		* Detect Delimiters occurences in the CSV
81		*
82		* Returns a associative array where each key represents
83		* a valid delimiter and each value the number of occurences
84		*
85		* @param string[] $delimiters the delimiters to consider
86		* @param int $nb_records Detection is made using $nb_records of the CSV
87		*
88		* @return array
89		*/
90	6	public function fetchDelimitersOccurrence(array $delimiters, int $nb_records = 1): array
91		{
92		$filter = function ($value): bool {
93	4	return 1 == strlen($value);
94	3	};
95
96	6	$nb_records = $this->filterMinRange($nb_records, 1, __METHOD__.'() expects the number of records to consider to be a valid positive integer, %s given');
97	4	$delimiters = array_unique(array_filter($delimiters, $filter));
98		$reducer = function (array $res, string $delimiter) use ($nb_records): array {
99	4	$res[$delimiter] = $this->getCellCount($delimiter, $nb_records);
100
101	4	return $res;
102	4	};
103
104	4	$res = array_reduce($delimiters, $reducer, []);
105	4	arsort($res, SORT_NUMERIC);
106
107	4	return $res;
108		}
109
110		/**
111		* Returns the cell count for a specified delimiter
112		* and a specified number of records
113		*
114		* @param string $delimiter CSV delimiter
115		* @param int $nb_records CSV records to consider
116		*
117		* @return int
118		*/
119	2	protected function getCellCount(string $delimiter, int $nb_records): int
120		{
121		$filter = function ($record): bool {
122	2	return is_array($record) && count($record) > 1;
123	1	};
124
125	2	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
126	2	$this->document->setCsvControl($delimiter, $this->enclosure, $this->escape);
127	2	$iterator = new CallbackFilterIterator(new LimitIterator($this->document, 0, $nb_records), $filter);
128
129	2	return count(iterator_to_array($iterator, false), COUNT_RECURSIVE);
130		}
131
132		/**
133		* Returns the record padding value
134		*
135		* @return mixed
136		*/
137	2	public function getRecordPaddingValue()
138		{
139	2	return $this->record_padding_value;
140		}
141
142		/**
143		* Returns the record offset used as header
144		*
145		* If no CSV record is used this method MUST return null
146		*
147		* @return int\|null
148		*/
149	2	public function getHeaderOffset()
150		{
151	2	return $this->header_offset;
152		}
153
154		/**
155		* Returns wether the selected header can be combine to each record
156		*
157		* A valid header must be empty or contains unique string field names
158		*
159		* @return bool
160		*/
161	2	public function supportsHeaderAsRecordKeys(): bool
162		{
163	2	$header = $this->getHeader();
164
165	2	return empty($header) \|\| $header === array_unique(array_filter($header, 'is_string'));
166		}
167
168		/**
169		* Returns the CSV record header
170		*
171		* The returned header is represented as an array of string values
172		*
173		* @return string[]
174		*/
175	4	public function getHeader(): array
176		{
177	4	if (null === $this->header_offset) {
178	2	return $this->header;
179		}
180
181	4	if (empty($this->header)) {
182	4	$this->header = $this->setHeader($this->header_offset);
183		}
184
185	4	return $this->header;
186		}
187
188		/**
189		* Determine the CSV record header
190		*
191		* @param int $offset
192		*
193		* @throws RuntimeException If the header offset is an integer
194		* and the corresponding record is missing
195		* or is an empty array
196		*
197		* @return string[]
198		*/
199	6	protected function setHeader(int $offset): array
200		{
201	6	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
202	6	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
203	6	$this->document->seek($offset);
204	6	$header = $this->document->current();
205	6	if (empty($header)) {
206	2	throw new RuntimeException(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
207		}
208
209	4	if (0 === $offset) {
210	2	return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
211		}
212
213	2	return $header;
214		}
215
216		/**
217		* Strip the BOM sequence from a record
218		*
219		* @param string[] $record
220		* @param int $bom_length
221		* @param string $enclosure
222		*
223		* @return string[]
224		*/
225	8	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
226		{
227	8	if (0 == $bom_length) {
228	2	return $record;
229		}
230
231	6	$record[0] = mb_substr($record[0], $bom_length);
232	6	if ($enclosure == mb_substr($record[0], 0, 1) && $enclosure == mb_substr($record[0], -1, 1)) {
233	2	$record[0] = mb_substr($record[0], 1, -1);
234		}
235
236	6	return $record;
237		}
238
239		/**
240		* @inheritdoc
241		*/
242	6	public function __call($method, array $arguments)
243		{
244	6	$whitelisted = ['fetchColumn' => 1, 'fetchPairs' => 1, 'fetchOne' => 1, 'fetchAll' => 1];
245	6	if (isset($whitelisted[$method])) {
246	2	return (new ResultSet($this->getRecords(), $this->getHeader()))->$method(...$arguments);
247		}
248
249	4	throw new BadMethodCallException(sprintf('%s::%s() method does not exist', __CLASS__, $method));
250		}
251
252		/**
253		* @inheritdoc
254		*/
255	2	public function count(): int
256		{
257	2	if (-1 === $this->nb_records) {
258	2	$this->nb_records = iterator_count($this->getRecords());
259		}
260
261	2	return $this->nb_records;
262		}
263
264		/**
265		* Returns the CSV records in an iterator object.
266		*
267		* @return Iterator
268		*/
269	6	public function getRecords(): Iterator
270		{
271	6	$this->records = $this->records ?? $this->setRecords();
272
273	4	return $this->records;
274		}
275
276		/**
277		* Returns the CSV records in an iterator object.
278		*
279		* Each CSV record is represented as a simple array of string or null values.
280		*
281		* If the CSV document has a header record then each record is combined
282		* to each header record and the header record is removed from the iterator.
283		*
284		* If the CSV document is inconsistent. Missing record fields are
285		* filled with null values while extra record fields are strip from
286		* the returned object.
287		*
288		* @throws RuntimeException If the header contains non unique column name
289		*
290		* @return Iterator
291		*/
292	4	protected function setRecords(): Iterator
293		{
294	4	if (!$this->supportsHeaderAsRecordKeys()) {
295	2	throw new RuntimeException('The header record must be empty or a flat array with unique string values');
296		}
297
298		$normalized = function ($record): bool {
299	2	return is_array($record) && $record != [null];
300	1	};
301	2	$bom = $this->getInputBOM();
302	2	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
303	2	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
304
305	2	return $this->combineHeader($this->stripBOM(new CallbackFilterIterator($this->document, $normalized), $bom));
306		}
307
308		/**
309		* Add the CSV header if present and valid
310		*
311		* @param Iterator $iterator
312		*
313		* @return Iterator
314		*/
315	14	protected function combineHeader(Iterator $iterator): Iterator
316		{
317	14	if (null === $this->header_offset) {
318	8	return $iterator;
319		}
320
321		$iterator = new CallbackFilterIterator($iterator, function (array $record, int $offset): bool {
322	8	return $offset != $this->header_offset;
323	8	});
324
325	8	$header = $this->getHeader();
326	8	$field_count = count($header);
327		$mapper = function (array $record) use ($header, $field_count): array {
328	8	if (count($record) != $field_count) {
329	4	$record = array_slice(array_pad($record, $field_count, $this->record_padding_value), 0, $field_count);
330		}
331
332	8	return array_combine($header, $record);
333	8	};
334
335	8	return new MapIterator($iterator, $mapper);
336		}
337
338		/**
339		* Strip the BOM sequence if present
340		*
341		* @param Iterator $iterator
342		* @param string $bom
343		*
344		* @return Iterator
345		*/
346	10	protected function stripBOM(Iterator $iterator, string $bom): Iterator
347		{
348	10	if ('' === $bom) {
349	4	return $iterator;
350		}
351
352	6	$bom_length = mb_strlen($bom);
353	6	$mapper = function (array $record, int $index) use ($bom_length): array {
354	6	if (0 != $index) {
355	2	return $record;
356		}
357
358	6	return $this->removeBOM($record, $bom_length, $this->enclosure);
359	6	};
360
361	6	return new MapIterator($iterator, $mapper);
362		}
363
364		/**
365		* @inheritdoc
366		*/
367	2	public function getIterator(): Iterator
368		{
369	2	return $this->getRecords();
370		}
371
372		/**
373		* Set the record padding value
374		*
375		* @param mixed $record_padding_value
376		*
377		* @return static
378		*/
379	2	public function setRecordPaddingValue($record_padding_value): self
380		{
381	2	$this->record_padding_value = $record_padding_value;
382
383	2	return $this;
384		}
385
386		/**
387		* Selects the record to be used as the CSV header
388		*
389		* Because of the header is represented as an array, to be valid
390		* a header MUST contain only unique string value.
391		*
392		* @param int\|null $offset the header record offset
393		*
394		* @return static
395		*/
396	2	public function setHeaderOffset($offset): self
397		{
398	2	if (null !== $offset) {
399	2	$offset = $this->filterMinRange($offset, 0, __METHOD__.'() expects the header offset index to be a positive integer or 0, %s given');
400		}
401
402	2	if ($offset !== $this->header_offset) {
403	2	$this->header_offset = $offset;
404	2	$this->resetProperties();
405		}
406
407	2	return $this;
408		}
409
410		/**
411		* @inheritdoc
412		*/
413	4	protected function resetProperties()
414		{
415	4	$this->nb_records = -1;
416	4	$this->header = [];
417	4	$this->records = null;
418	4	}
419		}
420

thephpleague / csv

Pull Request — master (#234)

Reader::setRecords() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like