Reader::getDocument() - Code Metrics - Inspection of "Adding support for the escape char with an empty s..." - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#309)

by ignace nyamagana

created 2018-09-21 10:41 UTC

Reader::getDocument() A

↳ Parent: Reader

Complexity

Conditions	3
Paths	2

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	8
CRAP Score	3

Importance

Changes

Metric	Value
cc	3
nc	2
nop	0
dl	0
loc	14
ccs	8
cts	8
cp	1
crap	3
rs	9.7998
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 9.1.5
 * @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 select records from a CSV document.
 *
 * @package League.csv
 * @since  3.0.0
 *
 * @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)
    {
        $document = $this->getDocument();
        foreach ($document as $index => $record) {
            if ($offset === $index) {
                return $record;
            }
        }

        return false;
    }

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

            return (new RFC4180Iterator($this->document))->getIterator();
        }

        $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 9.1.5
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 select records from a CSV document.
45		*
46		* @package League.csv
47		* @since 3.0.0
48		*
49		* @method array fetchOne(int $nth_record = 0) Returns a single record from the CSV
50		* @method Generator fetchColumn(string\|int $column_index) Returns the next value from a single CSV record field
51		* @method Generator fetchPairs(string\|int $offset_index = 0, string\|int $value_index = 1) Fetches the next key-value pairs from the CSV document
52		*/
53		class Reader extends AbstractCsv implements Countable, IteratorAggregate, JsonSerializable
54		{
55		/**
56		* header offset.
57		*
58		* @var int\|null
59		*/
60		protected $header_offset;
61
62		/**
63		* header record.
64		*
65		* @var string[]
66		*/
67		protected $header = [];
68
69		/**
70		* records count.
71		*
72		* @var int
73		*/
74		protected $nb_records = -1;
75
76		/**
77		* {@inheritdoc}
78		*/
79		protected $stream_filter_mode = STREAM_FILTER_READ;
80
81		/**
82		* {@inheritdoc}
83		*/
84	3	public static function createFromPath(string $path, string $open_mode = 'r', $context = null)
85		{
86	3	return parent::createFromPath($path, $open_mode, $context);
87		}
88
89		/**
90		* {@inheritdoc}
91		*/
92	24	protected function resetProperties()
93		{
94	24	parent::resetProperties();
95	24	$this->nb_records = -1;
96	24	$this->header = [];
97	24	}
98
99		/**
100		* Returns the header offset.
101		*
102		* If no CSV header offset is set this method MUST return null
103		*
104		* @return int\|null
105		*/
106	15	public function getHeaderOffset()
107		{
108	15	return $this->header_offset;
109		}
110
111		/**
112		* Returns the CSV record used as header.
113		*
114		* The returned header is represented as an array of string values
115		*
116		* @return string[]
117		*/
118	15	public function getHeader(): array
119		{
120	15	if (null === $this->header_offset) {
121	12	return $this->header;
122		}
123
124	6	if ([] !== $this->header) {
125	3	return $this->header;
126		}
127
128	6	$this->header = $this->setHeader($this->header_offset);
129
130	6	return $this->header;
131		}
132
133		/**
134		* Determine the CSV record header.
135		*
136		* @throws Exception If the header offset is set and no record is found or is the empty array
137		*
138		* @return string[]
139		*/
140	12	protected function setHeader(int $offset): array
141		{
142	12	$header = $this->seekRow($offset);
143	12	if (false === $header \|\| [] === $header) {
144	6	throw new Exception(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
145		}
146
147	6	if (0 === $offset) {
148	3	return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
149		}
150
151	3	return $header;
152		}
153
154		/**
155		* Returns the row at a given offset.
156		*
157		* @return array\|false
158		*/
159	6	protected function seekRow(int $offset)
160		{
161	6	$document = $this->getDocument();
162	6	foreach ($document as $index => $record) {
163	6	if ($offset === $index) {
164	6	return $record;
165		}
166		}
167
168	6	return false;
169		}
170
171		/**
172		* Returns the document as an Iterator.
173		*/
174	15	protected function getDocument(): Iterator
175		{
176	15	if ('' === $this->escape && PHP_VERSION_ID < 70400) {
177	6	$this->document->setCsvControl($this->delimiter, $this->enclosure);
178
179	6	return (new RFC4180Iterator($this->document))->getIterator();
180		}
181
182	9	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
183	9	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
184	9	$this->document->rewind();
185
186	9	return $this->document;
187		}
188
189		/**
190		* Strip the BOM sequence from a record.
191		*
192		* @param string[] $record
193		*
194		* @return string[]
195		*/
196	12	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
197		{
198	12	if (0 === $bom_length) {
199	3	return $record;
200		}
201
202	9	$record[0] = mb_substr($record[0], $bom_length);
203	9	if ($enclosure.$enclosure != substr($record[0].$record[0], strlen($record[0]) - 1, 2)) {
204	6	return $record;
205		}
206
207	3	$record[0] = substr($record[0], 1, -1);
208
209	3	return $record;
210		}
211
212		/**
213		* {@inheritdoc}
214		*/
215	9	public function __call($method, array $arguments)
216		{
217	9	static $whitelisted = ['fetchColumn' => 1, 'fetchOne' => 1, 'fetchPairs' => 1];
218	9	if (isset($whitelisted[$method])) {
219	3	return (new ResultSet($this->getRecords(), $this->getHeader()))->$method(...$arguments);
220		}
221
222	6	throw new BadMethodCallException(sprintf('%s::%s() method does not exist', static::class, $method));
223		}
224
225		/**
226		* {@inheritdoc}
227		*/
228	3	public function count(): int
229		{
230	3	if (-1 === $this->nb_records) {
231	3	$this->nb_records = iterator_count($this->getRecords());
232		}
233
234	3	return $this->nb_records;
235		}
236
237		/**
238		* {@inheritdoc}
239		*/
240	3	public function getIterator(): Iterator
241		{
242	3	return $this->getRecords();
243		}
244
245		/**
246		* {@inheritdoc}
247		*/
248	3	public function jsonSerialize(): array
249		{
250	3	return iterator_to_array($this->getRecords(), false);
251		}
252
253		/**
254		* Returns the CSV records as an iterator object.
255		*
256		* Each CSV record is represented as a simple array containig strings or null values.
257		*
258		* If the CSV document has a header record then each record is combined
259		* to the header record and the header record is removed from the iterator.
260		*
261		* If the CSV document is inconsistent. Missing record fields are
262		* filled with null values while extra record fields are strip from
263		* the returned object.
264		*
265		* @param string[] $header an optional header to use instead of the CSV document header
266		*/
267	18	public function getRecords(array $header = []): Iterator
268		{
269	18	$header = $this->computeHeader($header);
270	15	$normalized = function ($record): bool {
271	15	return is_array($record) && $record != [null];
272	15	};
273	15	$bom = $this->getInputBOM();
274	15	$document = $this->getDocument();
275
276	15	$records = $this->stripBOM(new CallbackFilterIterator($document, $normalized), $bom);
277	15	if (null !== $this->header_offset) {
278	6	$records = new CallbackFilterIterator($records, function (array $record, int $offset): bool {
279	6	return $offset !== $this->header_offset;
280	6	});
281		}
282
283	15	return $this->combineHeader($records, $header);
284		}
285
286		/**
287		* Returns the header to be used for iteration.
288		*
289		* @param string[] $header
290		*
291		* @throws Exception If the header contains non unique column name
292		*
293		* @return string[]
294		*/
295	24	protected function computeHeader(array $header)
296		{
297	24	if ([] === $header) {
298	21	$header = $this->getHeader();
299		}
300
301	24	if ($header === array_unique(array_filter($header, 'is_string'))) {
302	21	return $header;
303		}
304
305	3	throw new Exception('The header record must be empty or a flat array with unique string values');
306		}
307
308		/**
309		* Combine the CSV header to each record if present.
310		*
311		* @param string[] $header
312		*/
313	30	protected function combineHeader(Iterator $iterator, array $header): Iterator
314		{
315	30	if ([] === $header) {
316	21	return $iterator;
317		}
318
319	12	$field_count = count($header);
320	12	$mapper = function (array $record) use ($header, $field_count): array {
321	12	if (count($record) != $field_count) {
322	6	$record = array_slice(array_pad($record, $field_count, null), 0, $field_count);
323		}
324
325	12	return array_combine($header, $record);
326	12	};
327
328	12	return new MapIterator($iterator, $mapper);
329		}
330
331		/**
332		* Strip the BOM sequence from the returned records if necessary.
333		*/
334	24	protected function stripBOM(Iterator $iterator, string $bom): Iterator
335		{
336	24	if ('' === $bom) {
337	15	return $iterator;
338		}
339
340	9	$bom_length = mb_strlen($bom);
341	9	$mapper = function (array $record, int $index) use ($bom_length): array {
342	9	if (0 != $index) {
343	3	return $record;
344		}
345
346	9	return $this->removeBOM($record, $bom_length, $this->enclosure);
347	9	};
348
349	9	return new MapIterator($iterator, $mapper);
350		}
351
352		/**
353		* Selects the record to be used as the CSV header.
354		*
355		* Because the header is represented as an array, to be valid
356		* a header MUST contain only unique string value.
357		*
358		* @param int\|null $offset the header record offset
359		*
360		* @throws Exception if the offset is a negative integer
361		*
362		* @return static
363		*/
364	21	public function setHeaderOffset($offset): self
365		{
366	21	if ($offset === $this->header_offset) {
367	12	return $this;
368		}
369
370	9	if (!is_nullable_int($offset)) {
371	3	throw new TypeError(sprintf(__METHOD__.'() expects 1 Argument to be null or an integer %s given', gettype($offset)));
372		}
373
374	6	if (null !== $offset && 0 > $offset) {
375	3	throw new Exception(__METHOD__.'() expects 1 Argument to be greater or equal to 0');
376		}
377
378	3	$this->header_offset = $offset;
379	3	$this->resetProperties();
380
381	3	return $this;
382		}
383		}
384

thephpleague / csv

Pull Request — master (#309)

Reader::getDocument() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like