Reader::removeBOM() - Code Metrics - Inspection of "Bug fix Reader::getHeader method" - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( b4a812...401201 )

by ignace nyamagana

created 2018-09-25 20:41 UTC

Reader::removeBOM() A

↳ Parent: Reader

Complexity

Conditions	3
Paths	3

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	8
CRAP Score	3

Importance

Changes

Metric	Value
cc	3
nc	3
nop	3
dl	0
loc	15
ccs	8
cts	8
cp	1
crap	3
rs	9.7666
c	0
b	0
f	0

<?php

/**
 * League.Csv (https://csv.thephpleague.com).
 *
 * @author  Ignace Nyamagana Butera <[email protected]>
 * @license https://github.com/thephpleague/csv/blob/master/LICENSE (MIT License)
 * @version 9.2.0
 * @link    https://github.com/thephpleague/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 SplFileObject;
use TypeError;
use const STREAM_FILTER_READ;
use function array_combine;
use function array_filter;
use function array_pad;
use function array_slice;
use function array_unique;
use function gettype;
use function is_array;
use function iterator_count;
use function iterator_to_array;
use function mb_strlen;
use function mb_substr;
use function sprintf;
use function strlen;
use function substr;

/**
 * A class to parse and read records from a CSV document.
 *
 * @method array fetchOne(int $nth_record = 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 = 0, string|int $value_index = 1) Fetches the next key-value pairs from the CSV document
 */
class Reader extends AbstractCsv implements Countable, IteratorAggregate, JsonSerializable
{
    /**
     * header offset.
     *
     * @var int|null
     */
    protected $header_offset;

    /**
     * header record.
     *
     * @var string[]
     */
    protected $header = [];

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

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

    /**
     * {@inheritdoc}
     */
    public static function createFromPath(string $path, string $open_mode = 'r', $context = null)
    {
        return parent::createFromPath($path, $open_mode, $context);
    }

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

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

    /**
     * Returns the CSV record used as 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 ([] !== $this->header) {
            return $this->header;
        }

        $this->header = $this->setHeader($this->header_offset);

        return $this->header;
    }

    /**
     * Determine the CSV record header.
     *
     * @throws Exception If the header offset is set and no record is found or is the empty array
     *
     * @return string[]
     */
    protected function setHeader(int $offset): array
    {
        $header = $this->seekRow($offset);
        if (false === $header || [] === $header) {
            throw new Exception(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;
    }

    /**
     * Returns the row at a given offset.
     *
     * @return array|false
     */
    protected function seekRow(int $offset)
    {
        foreach ($this->getDocument() as $index => $record) {
            if ($offset === $index) {
                return [null] === $record ? false : $record;
            }
        }

        return false;
    }

    /**
     * Returns the document as an Iterator.
     */
    protected function getDocument(): Iterator
    {
        if ('' === $this->escape && PHP_VERSION_ID < 70400) {
            return EmptyEscapeParser::parse($this->document, $this->delimiter, $this->enclosure);
        }

        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
        $this->document->rewind();

        return $this->document;
    }

    /**
     * Strip the BOM sequence from a record.
     *
     * @param string[] $record
     *
     * @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.$enclosure != substr($record[0].$record[0], strlen($record[0]) - 1, 2)) {
            return $record;
        }

        $record[0] = substr($record[0], 1, -1);

        return $record;
    }

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

        throw new BadMethodCallException(sprintf('%s::%s() method does not exist', static::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 as an iterator object.
     *
     * Each CSV record is represented as a simple array containig strings or null values.
     *
     * If the CSV document has a header record then each record is combined
     * to the 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
     */
    public function getRecords(array $header = []): Iterator
    {
        $header = $this->computeHeader($header);
        $normalized = function ($record): bool {
            return is_array($record) && $record != [null];
        };
        $bom = $this->getInputBOM();
        $document = $this->getDocument();

        $records = $this->stripBOM(new CallbackFilterIterator($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 Exception If the header contains non unique column name
     *
     * @return string[]
     */
    protected function computeHeader(array $header)
    {
        if ([] === $header) {
            $header = $this->getHeader();
        }

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

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

    /**
     * Combine the CSV header to each record if present.
     *
     * @param string[] $header
     */
    protected function combineHeader(Iterator $iterator, array $header): Iterator
    {
        if ([] === $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 from the returned records if necessary.
     */
    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 the header is represented as an array, to be valid
     * a header MUST contain only unique string value.
     *
     * @param int|null $offset the header record offset
     *
     * @throws Exception if the offset is a negative integer
     *
     * @return static
     */
    public function setHeaderOffset($offset): self
    {
        if ($offset === $this->header_offset) {
            return $this;
        }

        if (!is_nullable_int($offset)) {
            throw new TypeError(sprintf(__METHOD__.'() expects 1 Argument to be null or an integer %s given', gettype($offset)));
        }

        if (null !== $offset && 0 > $offset) {
            throw new Exception(__METHOD__.'() expects 1 Argument to be greater or equal to 0');
        }

        $this->header_offset = $offset;
        $this->resetProperties();

        return $this;
    }
}


1		<?php
2
3		/**
4		* League.Csv (https://csv.thephpleague.com).
5		*
6		* @author Ignace Nyamagana Butera <[email protected]>
7		* @license https://github.com/thephpleague/csv/blob/master/LICENSE (MIT License)
8		* @version 9.2.0
9		* @link https://github.com/thephpleague/csv
10		*
11		* For the full copyright and license information, please view the LICENSE
12		* file that was distributed with this source code.
13		*/
14
15		declare(strict_types=1);
16
17		namespace League\Csv;
18
19		use BadMethodCallException;
20		use CallbackFilterIterator;
21		use Countable;
22		use Iterator;
23		use IteratorAggregate;
24		use JsonSerializable;
25		use SplFileObject;
26		use TypeError;
27		use const STREAM_FILTER_READ;
28		use function array_combine;
29		use function array_filter;
30		use function array_pad;
31		use function array_slice;
32		use function array_unique;
33		use function gettype;
34		use function is_array;
35		use function iterator_count;
36		use function iterator_to_array;
37		use function mb_strlen;
38		use function mb_substr;
39		use function sprintf;
40		use function strlen;
41		use function substr;
42
43		/**
44		* A class to parse and read records from a CSV document.
45		*
46		* @method array fetchOne(int $nth_record = 0) Returns a single record from the CSV
47		* @method Generator fetchColumn(string\|int $column_index) Returns the next value from a single CSV record field
48		* @method Generator fetchPairs(string\|int $offset_index = 0, string\|int $value_index = 1) Fetches the next key-value pairs from the CSV document
49		*/
50		class Reader extends AbstractCsv implements Countable, IteratorAggregate, JsonSerializable
51		{
52		/**
53		* header offset.
54		*
55		* @var int\|null
56		*/
57		protected $header_offset;
58
59		/**
60		* header record.
61		*
62		* @var string[]
63		*/
64		protected $header = [];
65
66		/**
67		* records count.
68		*
69		* @var int
70		*/
71		protected $nb_records = -1;
72
73		/**
74		* {@inheritdoc}
75		*/
76		protected $stream_filter_mode = STREAM_FILTER_READ;
77
78		/**
79		* {@inheritdoc}
80		*/
81	3	public static function createFromPath(string $path, string $open_mode = 'r', $context = null)
82		{
83	3	return parent::createFromPath($path, $open_mode, $context);
84		}
85
86		/**
87		* {@inheritdoc}
88		*/
89	24	protected function resetProperties()
90		{
91	24	parent::resetProperties();
92	24	$this->nb_records = -1;
93	24	$this->header = [];
94	24	}
95
96		/**
97		* Returns the header offset.
98		*
99		* If no CSV header offset is set this method MUST return null
100		*
101		* @return int\|null
102		*/
103	15	public function getHeaderOffset()
104		{
105	15	return $this->header_offset;
106		}
107
108		/**
109		* Returns the CSV record used as header.
110		*
111		* The returned header is represented as an array of string values
112		*
113		* @return string[]
114		*/
115	18	public function getHeader(): array
116		{
117	18	if (null === $this->header_offset) {
118	12	return $this->header;
119		}
120
121	9	if ([] !== $this->header) {
122	3	return $this->header;
123		}
124
125	9	$this->header = $this->setHeader($this->header_offset);
126
127	6	return $this->header;
128		}
129
130		/**
131		* Determine the CSV record header.
132		*
133		* @throws Exception If the header offset is set and no record is found or is the empty array
134		*
135		* @return string[]
136		*/
137	12	protected function setHeader(int $offset): array
138		{
139	12	$header = $this->seekRow($offset);
140	12	if (false === $header \|\| [] === $header) {
141	6	throw new Exception(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
142		}
143
144	6	if (0 === $offset) {
145	3	return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
146		}
147
148	3	return $header;
149		}
150
151		/**
152		* Returns the row at a given offset.
153		*
154		* @return array\|false
155		*/
156	9	protected function seekRow(int $offset)
157		{
158	9	foreach ($this->getDocument() as $index => $record) {
159	9	if ($offset === $index) {
160	9	return [null] === $record ? false : $record;
161		}
162		}
163
164	9	return false;
165		}
166
167		/**
168		* Returns the document as an Iterator.
169		*/
170	15	protected function getDocument(): Iterator
171		{
172	15	if ('' === $this->escape && PHP_VERSION_ID < 70400) {
173	6	return EmptyEscapeParser::parse($this->document, $this->delimiter, $this->enclosure);
174		}
175
176	9	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
177	9	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
178	9	$this->document->rewind();
179
180	9	return $this->document;
181		}
182
183		/**
184		* Strip the BOM sequence from a record.
185		*
186		* @param string[] $record
187		*
188		* @return string[]
189		*/
190	12	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
191		{
192	12	if (0 === $bom_length) {
193	3	return $record;
194		}
195
196	9	$record[0] = mb_substr($record[0], $bom_length);
197	9	if ($enclosure.$enclosure != substr($record[0].$record[0], strlen($record[0]) - 1, 2)) {
198	6	return $record;
199		}
200
201	3	$record[0] = substr($record[0], 1, -1);
202
203	3	return $record;
204		}
205
206		/**
207		* {@inheritdoc}
208		*/
209	9	public function __call($method, array $arguments)
210		{
211	9	static $whitelisted = ['fetchColumn' => 1, 'fetchOne' => 1, 'fetchPairs' => 1];
212	9	if (isset($whitelisted[$method])) {
213	3	return (new ResultSet($this->getRecords(), $this->getHeader()))->$method(...$arguments);
214		}
215
216	6	throw new BadMethodCallException(sprintf('%s::%s() method does not exist', static::class, $method));
217		}
218
219		/**
220		* {@inheritdoc}
221		*/
222	3	public function count(): int
223		{
224	3	if (-1 === $this->nb_records) {
225	3	$this->nb_records = iterator_count($this->getRecords());
226		}
227
228	3	return $this->nb_records;
229		}
230
231		/**
232		* {@inheritdoc}
233		*/
234	3	public function getIterator(): Iterator
235		{
236	3	return $this->getRecords();
237		}
238
239		/**
240		* {@inheritdoc}
241		*/
242	3	public function jsonSerialize(): array
243		{
244	3	return iterator_to_array($this->getRecords(), false);
245		}
246
247		/**
248		* Returns the CSV records as an iterator object.
249		*
250		* Each CSV record is represented as a simple array containig strings or null values.
251		*
252		* If the CSV document has a header record then each record is combined
253		* to the header record and the header record is removed from the iterator.
254		*
255		* If the CSV document is inconsistent. Missing record fields are
256		* filled with null values while extra record fields are strip from
257		* the returned object.
258		*
259		* @param string[] $header an optional header to use instead of the CSV document header
260		*/
261	18	public function getRecords(array $header = []): Iterator
262		{
263	18	$header = $this->computeHeader($header);
264	15	$normalized = function ($record): bool {
265	15	return is_array($record) && $record != [null];
266	15	};
267	15	$bom = $this->getInputBOM();
268	15	$document = $this->getDocument();
269
270	15	$records = $this->stripBOM(new CallbackFilterIterator($document, $normalized), $bom);
271	15	if (null !== $this->header_offset) {
272	6	$records = new CallbackFilterIterator($records, function (array $record, int $offset): bool {
273	6	return $offset !== $this->header_offset;
274	6	});
275		}
276
277	15	return $this->combineHeader($records, $header);
278		}
279
280		/**
281		* Returns the header to be used for iteration.
282		*
283		* @param string[] $header
284		*
285		* @throws Exception If the header contains non unique column name
286		*
287		* @return string[]
288		*/
289	24	protected function computeHeader(array $header)
290		{
291	24	if ([] === $header) {
292	21	$header = $this->getHeader();
293		}
294
295	24	if ($header === array_unique(array_filter($header, 'is_string'))) {
296	21	return $header;
297		}
298
299	3	throw new Exception('The header record must be empty or a flat array with unique string values');
300		}
301
302		/**
303		* Combine the CSV header to each record if present.
304		*
305		* @param string[] $header
306		*/
307	30	protected function combineHeader(Iterator $iterator, array $header): Iterator
308		{
309	30	if ([] === $header) {
310	21	return $iterator;
311		}
312
313	12	$field_count = count($header);
314	12	$mapper = function (array $record) use ($header, $field_count): array {
315	12	if (count($record) != $field_count) {
316	6	$record = array_slice(array_pad($record, $field_count, null), 0, $field_count);
317		}
318
319	12	return array_combine($header, $record);
320	12	};
321
322	12	return new MapIterator($iterator, $mapper);
323		}
324
325		/**
326		* Strip the BOM sequence from the returned records if necessary.
327		*/
328	24	protected function stripBOM(Iterator $iterator, string $bom): Iterator
329		{
330	24	if ('' === $bom) {
331	15	return $iterator;
332		}
333
334	9	$bom_length = mb_strlen($bom);
335	9	$mapper = function (array $record, int $index) use ($bom_length): array {
336	9	if (0 != $index) {
337	3	return $record;
338		}
339
340	9	return $this->removeBOM($record, $bom_length, $this->enclosure);
341	9	};
342
343	9	return new MapIterator($iterator, $mapper);
344		}
345
346		/**
347		* Selects the record to be used as the CSV header.
348		*
349		* Because the header is represented as an array, to be valid
350		* a header MUST contain only unique string value.
351		*
352		* @param int\|null $offset the header record offset
353		*
354		* @throws Exception if the offset is a negative integer
355		*
356		* @return static
357		*/
358	24	public function setHeaderOffset($offset): self
359		{
360	24	if ($offset === $this->header_offset) {
361	12	return $this;
362		}
363
364	12	if (!is_nullable_int($offset)) {
365	3	throw new TypeError(sprintf(__METHOD__.'() expects 1 Argument to be null or an integer %s given', gettype($offset)));
366		}
367
368	9	if (null !== $offset && 0 > $offset) {
369	3	throw new Exception(__METHOD__.'() expects 1 Argument to be greater or equal to 0');
370		}
371
372	6	$this->header_offset = $offset;
373	6	$this->resetProperties();
374
375	6	return $this;
376		}
377		}
378

thephpleague / csv

Push — master ( b4a812...401201 )

Reader::removeBOM() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like