Reader::computeHeader() - Code Metrics - Inspection of "Improve Reader:getRecords and RFC4180Field" - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 6f5dc8...ace255 )

by ignace nyamagana

created 2017-06-19 08:15 UTC

Reader::computeHeader() A

↳ Parent: Reader

Complexity

Conditions	4
Paths	3

Size

Total Lines	12
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	5
CRAP Score	4

Importance

Changes

Metric	Value
cc	4
eloc	6
nc	3
nop	1
dl	0
loc	12
ccs	5
cts	5
cp	1
crap	4
rs	9.2
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 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;
    }

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

    /**
     * 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.
     *
     *
     * @param  array            $header
     * @throws RuntimeException If the header contains non unique column name
     * @return Iterator
     */
    public function getRecords(array $header = []): Iterator
    {
        $header = $this->computeHeader($header);
        $normalized = function ($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);

        $records = $this->stripBOM(new CallbackFilterIterator($this->document, $normalized), $bom);
        if (null !== $this->header_offset) {
            $records = new CallbackFilterIterator($records, function (array $record, int $offset): bool {
                return $offset !== $this->header_offset;
            });
        }

        return $this->combineHeader($records, $header);
    }

    protected function computeHeader(array $header)
    {
        if (!empty($header) && $header === array_unique(array_filter($header, 'is_string'))) {
            return $header;
        }

        if ($this->supportsHeaderAsRecordKeys()) {
            return $this->getHeader();
        }

        throw new RuntimeException('The header record must be empty or a flat array with unique string values');
    }

    /**
     * Add the CSV header if present and valid
     *
     * @param Iterator $iterator
     * @param array    $header
     *
     * @return Iterator
     */
    protected function combineHeader(Iterator $iterator, array $header): Iterator
    {
        if (empty($header)) {
            return $iterator;
        }

        $field_count = count($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);
    }

    /**
     * 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 = [];
    }
}


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

thephpleague / csv

Push — master ( 6f5dc8...ace255 )

Reader::computeHeader() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like