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

Completed

Push — master ( a3add9...6ca7fe )

by ignace nyamagana

created 2017-07-04 07:52 UTC

Reader::filterHeaderOffset() A

↳ Parent: Reader

Complexity

Conditions	4
Paths	3

Size

Total Lines	10
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	4

Importance

Changes

Metric	Value
cc	4
eloc	5
nc	3
nop	1
dl	0
loc	10
ccs	6
cts	6
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 JsonSerializable;
use League\Csv\Exception\OutOfRangeException;
use League\Csv\Exception\RuntimeException;
use SplFileObject;
use TypeError;

/**
 * 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, JsonSerializable
{
    /**
     * 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;

    /**
     * @inheritdoc
     */
    protected $stream_filter_mode = STREAM_FILTER_READ;

    /**
     * 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 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)) {
            return $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);
        if (empty($header = $this->document->current())) {
            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();
    }

    /**
     * @inheritdoc
     */
    public function jsonSerialize(): array
    {
        return iterator_to_array($this->getRecords(), false);
    }

    /**
     * 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 string[] $header an optional header to use instead of the CSV document header
     *
     * @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);
    }

    /**
     * Returns the header to be used for iteration
     *
     * @param string[] $header
     *
     * @throws RuntimeException If the header contains non unique column name
     *
     * @return string[]
     */
    protected function computeHeader(array $header)
    {
        if (empty($header)) {
            $header = $this->getHeader();
        }

        if ($header === array_unique(array_filter($header, 'is_string'))) {
            return $header;
        }

        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 string[] $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, null), 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);
    }

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

        return $this;
    }

    /**
     * Filter Header Offset
     *
     * @param int|null $offset
     *
     * @throws OutOfRangeException if header offset is not a positive integer or null
     */
    protected function filterHeaderOffset($offset)
    {
        if (null !== $offset && !is_int($offset)) {
            throw new TypeError(sprintf('Expected argument to be null or a integer %s given', gettype($offset)));
        }

        if ($offset < 0) {
            throw new OutOfRangeException(sprintf('%s() expects the header offset index to be a positive integer or 0, %s given', __METHOD__, $offset));
        }
    }

    /**
     * @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 JsonSerializable;
23		use League\Csv\Exception\OutOfRangeException;
24		use League\Csv\Exception\RuntimeException;
25		use SplFileObject;
26		use TypeError;
27
28		/**
29		* A class to manage records selection from a CSV document
30		*
31		* @package League.csv
32		* @since 3.0.0
33		*
34		* @method array fetchAll() Returns a sequential array of all CSV records
35		* @method array fetchOne(int $offset = 0) Returns a single record from the CSV
36		* @method Generator fetchColumn(string\|int $column_index) Returns the next value from a single CSV record field
37		* @method Generator fetchPairs(string\|int $offset_index, string\|int $value_index) Fetches the next key-value pairs from the CSV document
38		*/
39		class Reader extends AbstractCsv implements Countable, IteratorAggregate, JsonSerializable
40		{
41		/**
42		* CSV Document header offset
43		*
44		* @var int\|null
45		*/
46		protected $header_offset;
47
48		/**
49		* CSV Document Header record
50		*
51		* @var string[]
52		*/
53		protected $header = [];
54
55		/**
56		* Records count
57		*
58		* @var int
59		*/
60		protected $nb_records = -1;
61
62		/**
63		* @inheritdoc
64		*/
65		protected $stream_filter_mode = STREAM_FILTER_READ;
66
67		/**
68		* Returns the record offset used as header
69		*
70		* If no CSV record is used this method MUST return null
71		*
72		* @return int\|null
73		*/
74	10	public function getHeaderOffset()
75		{
76	10	return $this->header_offset;
77		}
78
79		/**
80		* Returns the CSV record header
81		*
82		* The returned header is represented as an array of string values
83		*
84		* @return string[]
85		*/
86	10	public function getHeader(): array
87		{
88	10	if (null === $this->header_offset) {
89	8	return $this->header;
90		}
91
92	4	if (!empty($this->header)) {
93	2	return $this->header;
94		}
95
96	4	$this->header = $this->setHeader($this->header_offset);
97
98	4	return $this->header;
99		}
100
101		/**
102		* Determine the CSV record header
103		*
104		* @param int $offset
105		*
106		* @throws RuntimeException If the header offset is an integer
107		* and the corresponding record is missing
108		* or is an empty array
109		*
110		* @return string[]
111		*/
112	6	protected function setHeader(int $offset): array
113		{
114	6	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
115	6	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
116	6	$this->document->seek($offset);
117	6	if (empty($header = $this->document->current())) {
118	2	throw new RuntimeException(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
119		}
120
121	4	if (0 === $offset) {
122	2	return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
123		}
124
125	2	return $header;
126		}
127
128		/**
129		* Strip the BOM sequence from a record
130		*
131		* @param string[] $record
132		* @param int $bom_length
133		* @param string $enclosure
134		*
135		* @return string[]
136		*/
137	8	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
138		{
139	8	if (0 == $bom_length) {
140	2	return $record;
141		}
142
143	6	$record[0] = mb_substr($record[0], $bom_length);
144	6	if ($enclosure == mb_substr($record[0], 0, 1) && $enclosure == mb_substr($record[0], -1, 1)) {
145	2	$record[0] = mb_substr($record[0], 1, -1);
146		}
147
148	6	return $record;
149		}
150
151		/**
152		* @inheritdoc
153		*/
154	6	public function __call($method, array $arguments)
155		{
156	6	$whitelisted = ['fetchColumn' => 1, 'fetchPairs' => 1, 'fetchOne' => 1, 'fetchAll' => 1];
157	6	if (isset($whitelisted[$method])) {
158	2	return (new ResultSet($this->getRecords(), $this->getHeader()))->$method(...$arguments);
159		}
160
161	4	throw new BadMethodCallException(sprintf('%s::%s() method does not exist', __CLASS__, $method));
162		}
163
164		/**
165		* @inheritdoc
166		*/
167	2	public function count(): int
168		{
169	2	if (-1 === $this->nb_records) {
170	2	$this->nb_records = iterator_count($this->getRecords());
171		}
172
173	2	return $this->nb_records;
174		}
175
176		/**
177		* @inheritdoc
178		*/
179	2	public function getIterator(): Iterator
180		{
181	2	return $this->getRecords();
182		}
183
184		/**
185		* @inheritdoc
186		*/
187	2	public function jsonSerialize(): array
188		{
189	2	return iterator_to_array($this->getRecords(), false);
190		}
191
192		/**
193		* Returns the CSV records in an iterator object.
194		*
195		* Each CSV record is represented as a simple array of string or null values.
196		*
197		* If the CSV document has a header record then each record is combined
198		* to each header record and the header record is removed from the iterator.
199		*
200		* If the CSV document is inconsistent. Missing record fields are
201		* filled with null values while extra record fields are strip from
202		* the returned object.
203		*
204		* @param string[] $header an optional header to use instead of the CSV document header
205		*
206		* @return Iterator
207		*/
208	12	public function getRecords(array $header = []): Iterator
209		{
210	12	$header = $this->computeHeader($header);
211		$normalized = function ($record): bool {
212	10	return is_array($record) && $record != [null];
213	5	};
214	10	$bom = $this->getInputBOM();
215	10	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
216	10	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
217
218	10	$records = $this->stripBOM(new CallbackFilterIterator($this->document, $normalized), $bom);
219	10	if (null !== $this->header_offset) {
220		$records = new CallbackFilterIterator($records, function (array $record, int $offset): bool {
221	4	return $offset !== $this->header_offset;
222	4	});
223		}
224
225	10	return $this->combineHeader($records, $header);
226		}
227
228		/**
229		* Returns the header to be used for iteration
230		*
231		* @param string[] $header
232		*
233		* @throws RuntimeException If the header contains non unique column name
234		*
235		* @return string[]
236		*/
237	16	protected function computeHeader(array $header)
238		{
239	16	if (empty($header)) {
240	14	$header = $this->getHeader();
241		}
242
243	16	if ($header === array_unique(array_filter($header, 'is_string'))) {
244	14	return $header;
245		}
246
247	2	throw new RuntimeException('The header record must be empty or a flat array with unique string values');
248		}
249
250		/**
251		* Add the CSV header if present and valid
252		*
253		* @param Iterator $iterator
254		* @param string[] $header
255		*
256		* @return Iterator
257		*/
258	20	protected function combineHeader(Iterator $iterator, array $header): Iterator
259		{
260	20	if (empty($header)) {
261	14	return $iterator;
262		}
263
264	8	$field_count = count($header);
265		$mapper = function (array $record) use ($header, $field_count): array {
266	8	if (count($record) != $field_count) {
267	4	$record = array_slice(array_pad($record, $field_count, null), 0, $field_count);
268		}
269
270	8	return array_combine($header, $record);
271	8	};
272
273	8	return new MapIterator($iterator, $mapper);
274		}
275
276		/**
277		* Strip the BOM sequence if present
278		*
279		* @param Iterator $iterator
280		* @param string $bom
281		*
282		* @return Iterator
283		*/
284	16	protected function stripBOM(Iterator $iterator, string $bom): Iterator
285		{
286	16	if ('' === $bom) {
287	10	return $iterator;
288		}
289
290	6	$bom_length = mb_strlen($bom);
291	6	$mapper = function (array $record, int $index) use ($bom_length): array {
292	6	if (0 != $index) {
293	2	return $record;
294		}
295
296	6	return $this->removeBOM($record, $bom_length, $this->enclosure);
297	6	};
298
299	6	return new MapIterator($iterator, $mapper);
300		}
301
302		/**
303		* Selects the record to be used as the CSV header
304		*
305		* Because of the header is represented as an array, to be valid
306		* a header MUST contain only unique string value.
307		*
308		* @param int\|null $offset the header record offset
309		*
310		* @return static
311		*/
312	10	public function setHeaderOffset($offset): self
313		{
314	10	$this->filterHeaderOffset($offset);
315	10	if ($offset !== $this->header_offset) {
316	2	$this->header_offset = $offset;
317	2	$this->resetProperties();
318		}
319
320	10	return $this;
321		}
322
323		/**
324		* Filter Header Offset
325		*
326		* @param int\|null $offset
327		*
328		* @throws OutOfRangeException if header offset is not a positive integer or null
329		*/
330	12	protected function filterHeaderOffset($offset)
331		{
332	12	if (null !== $offset && !is_int($offset)) {
333	2	throw new TypeError(sprintf('Expected argument to be null or a integer %s given', gettype($offset)));
334		}
335
336	10	if ($offset < 0) {
337	2	throw new OutOfRangeException(sprintf('%s() expects the header offset index to be a positive integer or 0, %s given', __METHOD__, $offset));
338		}
339	8	}
340
341		/**
342		* @inheritdoc
343		*/
344	2	protected function resetProperties()
345		{
346	2	$this->nb_records = -1;
347	2	$this->header = [];
348	2	}
349		}
350

thephpleague / csv

Push — master ( a3add9...6ca7fe )

Reader::filterHeaderOffset() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like