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

Completed

Push — master ( 147747...e61ddd )

by ignace nyamagana

created 2017-04-21 11:43 UTC

Reader::getRecords() A

↳ Parent: Reader

Complexity

Conditions	1
Paths	1

Size

Total Lines	4
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	2
CRAP Score	1

Importance

Changes

Metric	Value
cc	1
eloc	2
nc	1
nop	0
dl	0
loc	4
ccs	2
cts	2
cp	1
crap	1
rs	10
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 CallbackFilterIterator;
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
 *
 */
class Reader extends AbstractCsv implements IteratorAggregate
{
    /**
     * @inheritdoc
     */
    protected $stream_filter_mode = STREAM_FILTER_READ;

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

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

    /**
     * Tell whether the header needs to be re-generated
     *
     * @var bool
     */
    protected $is_header_loaded = false;

    /**
     * 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 a CSV records collection
     *
     * @param Statement $stmt
     *
     * @return ResultSet
     */
    public function select(Statement $stmt): ResultSet
    {
        return $stmt->process($this);
    }

    /**
     * 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, 'The number of records to consider must be a valid positive integer');
        $delimiters = array_unique(array_filter($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);
        $this->document->setCsvControl($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);
    }

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

        $iterator = $this->combineHeader(new CallbackFilterIterator($this->document, $normalized), $header);

        return $this->stripBOM($iterator, $bom);
    }

    /**
     * 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.
     *
     * @see Reader::getIterator()
     *
     * @throws RuntimeException If the header contains non unique column name
     *
     * @return Iterator
     */
    public function getRecords(): Iterator
    {
        return $this->getIterator();
    }

    /**
     * Returns the CSV record header
     *
     * The returned header is represented as an array of string values
     *
     * @throws RuntimeException If the header offset is an integer
     *                          and the corresponding record is missing
     *                          or is an empty array
     *
     * @return string[]
     */
    public function getHeader(): array
    {
        if ($this->is_header_loaded) {
            return $this->header;
        }

        $this->is_header_loaded = true;
        if (null === $this->header_offset) {
            $this->header = [];

            return $this->header;
        }

        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
        $this->document->seek($this->header_offset);
        $this->header = $this->document->current();
        if (empty($this->header)) {
            throw new RuntimeException(sprintf('The header record does not exist or is empty at offset: `%s`', $this->header_offset));
        }

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

        return $this->header;
    }

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

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

        $mapper = function (array $record) use ($header_count, $header): array {
            if ($header_count != count($record)) {
                $record = array_slice(array_pad($record, $header_count, null), 0, $header_count);
            }

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

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

    /**
     * Validates the array to be used by the fetchAssoc method
     *
     * @param array $keys
     *
     * @throws RuntimeException If the submitted array fails the assertion
     *
     * @return array
     */
    protected function filterColumnNames(array $keys): array
    {
        if (empty($keys) || $keys === array_unique(array_filter($keys, 'is_string'))) {
            return $keys;
        }

        throw new RuntimeException('Use a flat array with unique string values');
    }

    /**
     * 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);
    }

    /**
     * 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;
    }

    /**
     * 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, 'The header offset index must be a positive integer or 0');
        }

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

        return $this;
    }

    /**
     * @inheritdoc
     */
    protected function resetProperties()
    {
        return $this->is_header_loaded = false;
    }
}


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 CallbackFilterIterator;
18		use Iterator;
19		use IteratorAggregate;
20		use League\Csv\Exception\RuntimeException;
21		use LimitIterator;
22		use SplFileObject;
23
24		/**
25		* A class to manage records selection from a CSV document
26		*
27		* @package League.csv
28		* @since 3.0.0
29		*
30		*/
31		class Reader extends AbstractCsv implements IteratorAggregate
32		{
33		/**
34		* @inheritdoc
35		*/
36		protected $stream_filter_mode = STREAM_FILTER_READ;
37
38		/**
39		* CSV Document header offset
40		*
41		* @var int\|null
42		*/
43		protected $header_offset;
44
45		/**
46		* CSV Document Header record
47		*
48		* @var string[]
49		*/
50		protected $header = [];
51
52		/**
53		* Tell whether the header needs to be re-generated
54		*
55		* @var bool
56		*/
57		protected $is_header_loaded = false;
58
59		/**
60		* Returns the record offset used as header
61		*
62		* If no CSV record is used this method MUST return null
63		*
64		* @return int\|null
65		*/
66	2	public function getHeaderOffset()
67		{
68	2	return $this->header_offset;
69		}
70
71		/**
72		* Returns a CSV records collection
73		*
74		* @param Statement $stmt
75		*
76		* @return ResultSet
77		*/
78	2	public function select(Statement $stmt): ResultSet
79		{
80	2	return $stmt->process($this);
81		}
82
83		/**
84		* Detect Delimiters occurences in the CSV
85		*
86		* Returns a associative array where each key represents
87		* a valid delimiter and each value the number of occurences
88		*
89		* @param string[] $delimiters the delimiters to consider
90		* @param int $nb_records Detection is made using $nb_records of the CSV
91		*
92		* @return array
93		*/
94	8	public function fetchDelimitersOccurrence(array $delimiters, int $nb_records = 1): array
95		{
96		$filter = function ($value): bool {
97	6	return 1 == strlen($value);
98	4	};
99
100	8	$nb_records = $this->filterMinRange($nb_records, 1, 'The number of records to consider must be a valid positive integer');
101	6	$delimiters = array_unique(array_filter($delimiters, $filter));
102		$reducer = function (array $res, string $delimiter) use ($nb_records): array {
103	6	$res[$delimiter] = $this->getCellCount($delimiter, $nb_records);
104
105	6	return $res;
106	6	};
107
108	6	$res = array_reduce($delimiters, $reducer, []);
109	6	arsort($res, SORT_NUMERIC);
110
111	6	return $res;
112		}
113
114		/**
115		* Returns the cell count for a specified delimiter
116		* and a specified number of records
117		*
118		* @param string $delimiter CSV delimiter
119		* @param int $nb_records CSV records to consider
120		*
121		* @return int
122		*/
123	6	protected function getCellCount(string $delimiter, int $nb_records): int
124		{
125		$filter = function ($record): bool {
126	6	return is_array($record) && count($record) > 1;
127	3	};
128
129	6	$this->document->setFlags(SplFileObject::READ_CSV);
130	6	$this->document->setCsvControl($delimiter, $this->enclosure, $this->escape);
131	6	$iterator = new CallbackFilterIterator(new LimitIterator($this->document, 0, $nb_records), $filter);
132
133	6	return count(iterator_to_array($iterator, false), COUNT_RECURSIVE);
134		}
135
136		/**
137		* @inheritdoc
138		*/
139	108	public function getIterator(): Iterator
140		{
141	108	$bom = $this->getInputBOM();
142	108	$header = $this->getHeader();
143	106	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
144	106	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
145		$normalized = function ($record): bool {
146	90	return is_array($record) && $record != [null];
147	53	};
148
149	106	$iterator = $this->combineHeader(new CallbackFilterIterator($this->document, $normalized), $header);
150
151	104	return $this->stripBOM($iterator, $bom);
152		}
153
154		/**
155		* Returns the CSV records in an iterator object.
156		*
157		* Each CSV record is represented as a simple array of string or null values.
158		*
159		* If the CSV document has a header record then each record is combined
160		* to each header record and the header record is removed from the iterator.
161		*
162		* If the CSV document is inconsistent. Missing record fields are
163		* filled with null values while extra record fields are strip from
164		* the returned object.
165		*
166		* @see Reader::getIterator()
167		*
168		* @throws RuntimeException If the header contains non unique column name
169		*
170		* @return Iterator
171		*/
172	102	public function getRecords(): Iterator
173		{
174	102	return $this->getIterator();
175		}
176
177		/**
178		* Returns the CSV record header
179		*
180		* The returned header is represented as an array of string values
181		*
182		* @throws RuntimeException If the header offset is an integer
183		* and the corresponding record is missing
184		* or is an empty array
185		*
186		* @return string[]
187		*/
188	110	public function getHeader(): array
189		{
190	110	if ($this->is_header_loaded) {
191	102	return $this->header;
192		}
193
194	110	$this->is_header_loaded = true;
195	110	if (null === $this->header_offset) {
196	76	$this->header = [];
197
198	76	return $this->header;
199		}
200
201	36	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
202	36	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
203	36	$this->document->seek($this->header_offset);
204	36	$this->header = $this->document->current();
205	36	if (empty($this->header)) {
206	2	throw new RuntimeException(sprintf('The header record does not exist or is empty at offset: `%s`', $this->header_offset));
207		}
208
209	34	if (0 === $this->header_offset) {
210	30	$this->header = $this->removeBOM($this->header, mb_strlen($this->getInputBOM()), $this->enclosure);
211		}
212
213	34	return $this->header;
214		}
215
216		/**
217		* Add the CSV header if present and valid
218		*
219		* @param Iterator $iterator
220		* @param string[] $header
221		*
222		* @return Iterator
223		*/
224	106	protected function combineHeader(Iterator $iterator, array $header): Iterator
225		{
226	106	if (null === $this->header_offset) {
227	74	return $iterator;
228		}
229
230	32	$header = $this->filterColumnNames($header);
231	30	$header_count = count($header);
232		$iterator = new CallbackFilterIterator($iterator, function (array $record, int $offset): bool {
233	22	return $offset != $this->header_offset;
234	30	});
235
236		$mapper = function (array $record) use ($header_count, $header): array {
237	22	if ($header_count != count($record)) {
238	2	$record = array_slice(array_pad($record, $header_count, null), 0, $header_count);
239		}
240
241	22	return array_combine($header, $record);
242	30	};
243
244	30	return new MapIterator($iterator, $mapper);
245		}
246
247		/**
248		* Validates the array to be used by the fetchAssoc method
249		*
250		* @param array $keys
251		*
252		* @throws RuntimeException If the submitted array fails the assertion
253		*
254		* @return array
255		*/
256	32	protected function filterColumnNames(array $keys): array
257		{
258	32	if (empty($keys) \|\| $keys === array_unique(array_filter($keys, 'is_string'))) {
259	30	return $keys;
260		}
261
262	2	throw new RuntimeException('Use a flat array with unique string values');
263		}
264
265		/**
266		* Strip the BOM sequence if present
267		*
268		* @param Iterator $iterator
269		* @param string $bom
270		*
271		* @return Iterator
272		*/
273	104	protected function stripBOM(Iterator $iterator, string $bom): Iterator
274		{
275	104	if ('' === $bom) {
276	84	return $iterator;
277		}
278
279	20	$bom_length = mb_strlen($bom);
280	20	$mapper = function (array $record, int $index) use ($bom_length): array {
281	20	if (0 != $index) {
282	18	return $record;
283		}
284
285	12	return $this->removeBOM($record, $bom_length, $this->enclosure);
286	20	};
287
288	20	return new MapIterator($iterator, $mapper);
289		}
290
291		/**
292		* Strip the BOM sequence from a record
293		*
294		* @param string[] $record
295		* @param int $bom_length
296		* @param string $enclosure
297		*
298		* @return string[]
299		*/
300	42	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
301		{
302	42	if (0 == $bom_length) {
303	22	return $record;
304		}
305
306	20	$record[0] = mb_substr($record[0], $bom_length);
307	20	if ($enclosure == mb_substr($record[0], 0, 1) && $enclosure == mb_substr($record[0], -1, 1)) {
308	10	$record[0] = mb_substr($record[0], 1, -1);
309		}
310
311	20	return $record;
312		}
313
314		/**
315		* Selects the record to be used as the CSV header
316		*
317		* Because of the header is represented as an array, to be valid
318		* a header MUST contain only unique string value.
319		*
320		* @param int\|null $offset the header record offset
321		*
322		* @return static
323		*/
324	40	public function setHeaderOffset($offset): self
325		{
326	40	if (null !== $offset) {
327	36	$offset = $this->filterMinRange($offset, 0, 'The header offset index must be a positive integer or 0');
328		}
329
330	40	if ($offset !== $this->header_offset) {
331	36	$this->header_offset = $offset;
332	36	$this->resetProperties();
333		}
334
335	40	return $this;
336		}
337
338		/**
339		* @inheritdoc
340		*/
341	52	protected function resetProperties()
342		{
343	52	return $this->is_header_loaded = false;
344		}
345		}
346

thephpleague / csv

Push — master ( 147747...e61ddd )

Reader::getRecords() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like