Reader::stripBOM() - Code Metrics - Inspection of "Adding empty records process controls" - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#354)

by ignace nyamagana

created 2019-08-28 07:36 UTC

Reader::stripBOM() A

↳ Parent: Reader

Complexity

Conditions	3
Paths	2

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	10
CRAP Score	3

Importance

Changes

Metric	Value
cc	3
nc	2
nop	2
dl	0
loc	17
ccs	10
cts	10
cp	1
crap	3
rs	9.7
c	0
b	0
f	0

<?php

/**
 * League.Csv (https://csv.thephpleague.com)
 *
 * (c) Ignace Nyamagana Butera <[email protected]>
 *
 * 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 Generator;
use Iterator;
use IteratorAggregate;
use JsonSerializable;
use League\Csv\Polyfill\EmptyEscapeParser;
use SplFileObject;
use TypeError;
use function array_combine;
use function array_filter;
use function array_pad;
use function array_slice;
use function array_unique;
use function count;
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;
use const PHP_VERSION_ID;
use const STREAM_FILTER_READ;

/**
 * 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;

    /**
     * @var bool
     */
    protected $is_empty_records_skipped = true;

    /**
     * {@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 || [null] === $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 $record;
            }
        }

        return false;
    }

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

            return EmptyEscapeParser::parse($this->document);
        }

        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD);
        $this->document->setCsvControl($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 containing 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($record) && (!$this->is_empty_records_skipped || $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;
            });
        }

        if (!$this->is_empty_records_skipped) {
            $normalized_empty_records = static function (array $record): array {
                if ([null] === $record) {
                    return [];
                }

                return $record;
            };

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

        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 = static 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;
    }

    /**
     * Enable skipping empty records.
     */
    public function enableEmptyRecordsSkipping(): self
    {
        if (!$this->is_empty_records_skipped) {
            $this->is_empty_records_skipped = true;
            $this->nb_records = -1;
        }

        return $this;
    }

    /**
     * Disable skipping empty records.
     */
    public function disableEmptyRecordsSkipping(): self
    {
        if ($this->is_empty_records_skipped) {
            $this->is_empty_records_skipped = false;
            $this->nb_records = -1;
        }

        return $this;
    }

    /**
     * Tells whether empty records are skipped by the instance.
     */
    public function isEmptyRecordsSkipped(): bool
    {
        return $this->is_empty_records_skipped;
    }
}


1		<?php
2
3		/**
4		* League.Csv (https://csv.thephpleague.com)
5		*
6		* (c) Ignace Nyamagana Butera <[email protected]>
7		*
8		* For the full copyright and license information, please view the LICENSE
9		* file that was distributed with this source code.
10		*/
11
12		declare(strict_types=1);
13
14		namespace League\Csv;
15
16		use BadMethodCallException;
17		use CallbackFilterIterator;
18		use Countable;
19		use Generator;
20		use Iterator;
21		use IteratorAggregate;
22		use JsonSerializable;
23		use League\Csv\Polyfill\EmptyEscapeParser;
24		use SplFileObject;
25		use TypeError;
26		use function array_combine;
27		use function array_filter;
28		use function array_pad;
29		use function array_slice;
30		use function array_unique;
31		use function count;
32		use function gettype;
33		use function is_array;
34		use function iterator_count;
35		use function iterator_to_array;
36		use function mb_strlen;
37		use function mb_substr;
38		use function sprintf;
39		use function strlen;
40		use function substr;
41		use const PHP_VERSION_ID;
42		use const STREAM_FILTER_READ;
43
44		/**
45		* A class to parse and read records from a CSV document.
46		*
47		* @method array fetchOne(int $nth_record = 0) Returns a single record from the CSV
48		* @method Generator fetchColumn(string\|int $column_index) Returns the next value from a single CSV record field
49		* @method Generator fetchPairs(string\|int $offset_index = 0, string\|int $value_index = 1) Fetches the next key-value pairs from the CSV document
50		*/
51		class Reader extends AbstractCsv implements Countable, IteratorAggregate, JsonSerializable
52		{
53		/**
54		* header offset.
55		*
56		* @var int\|null
57		*/
58		protected $header_offset;
59
60		/**
61		* header record.
62		*
63		* @var string[]
64		*/
65		protected $header = [];
66
67		/**
68		* records count.
69		*
70		* @var int
71		*/
72		protected $nb_records = -1;
73
74		/**
75		* {@inheritdoc}
76		*/
77		protected $stream_filter_mode = STREAM_FILTER_READ;
78
79		/**
80		* @var bool
81		*/
82		protected $is_empty_records_skipped = true;
83
84		/**
85		* {@inheritdoc}
86		*/
87	3	public static function createFromPath(string $path, string $open_mode = 'r', $context = null)
88		{
89	3	return parent::createFromPath($path, $open_mode, $context);
90		}
91
92		/**
93		* {@inheritdoc}
94		*/
95	24	protected function resetProperties()
96		{
97	24	parent::resetProperties();
98	24	$this->nb_records = -1;
99	24	$this->header = [];
100	24	}
101
102		/**
103		* Returns the header offset.
104		*
105		* If no CSV header offset is set this method MUST return null
106		*
107		* @return int\|null
108		*/
109	15	public function getHeaderOffset()
110		{
111	15	return $this->header_offset;
112		}
113
114		/**
115		* Returns the CSV record used as header.
116		*
117		* The returned header is represented as an array of string values
118		*
119		* @return string[]
120		*/
121	18	public function getHeader(): array
122		{
123	18	if (null === $this->header_offset) {
124	12	return $this->header;
125		}
126
127	9	if ([] !== $this->header) {
128	3	return $this->header;
129		}
130
131	9	$this->header = $this->setHeader($this->header_offset);
132
133	6	return $this->header;
134		}
135
136		/**
137		* Determine the CSV record header.
138		*
139		* @throws Exception If the header offset is set and no record is found or is the empty array
140		*
141		* @return string[]
142		*/
143	12	protected function setHeader(int $offset): array
144		{
145	12	$header = $this->seekRow($offset);
146	12	if (false === $header \|\| [] === $header \|\| [null] === $header) {
147	6	throw new Exception(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
148		}
149
150	6	if (0 === $offset) {
151	3	return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
152		}
153
154	3	return $header;
155		}
156
157		/**
158		* Returns the row at a given offset.
159		*
160		* @return array\|false
161		*/
162	12	protected function seekRow(int $offset)
163		{
164	12	foreach ($this->getDocument() as $index => $record) {
165	12	if ($offset === $index) {
166	10	return $record;
167		}
168		}
169
170	6	return false;
171		}
172
173		/**
174		* Returns the document as an Iterator.
175		*/
176	15	protected function getDocument(): Iterator
177		{
178	15	if (70400 > PHP_VERSION_ID && '' === $this->escape) {
179	6	$this->document->setCsvControl($this->delimiter, $this->enclosure);
180
181	6	return EmptyEscapeParser::parse($this->document);
182		}
183
184	9	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD);
185	9	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
186	9	$this->document->rewind();
187
188	9	return $this->document;
189		}
190
191		/**
192		* Strip the BOM sequence from a record.
193		*
194		* @param string[] $record
195		*
196		* @return string[]
197		*/
198	12	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
199		{
200	12	if (0 === $bom_length) {
201	3	return $record;
202		}
203
204	9	$record[0] = mb_substr($record[0], $bom_length);
205	9	if ($enclosure.$enclosure != substr($record[0].$record[0], strlen($record[0]) - 1, 2)) {
206	6	return $record;
207		}
208
209	3	$record[0] = substr($record[0], 1, -1);
210
211	3	return $record;
212		}
213
214		/**
215		* {@inheritdoc}
216		*/
217	9	public function __call($method, array $arguments)
218		{
219	9	static $whitelisted = ['fetchColumn' => 1, 'fetchOne' => 1, 'fetchPairs' => 1];
220	9	if (isset($whitelisted[$method])) {
221	3	return (new ResultSet($this->getRecords(), $this->getHeader()))->$method(...$arguments);
222		}
223
224	6	throw new BadMethodCallException(sprintf('%s::%s() method does not exist', static::class, $method));
225		}
226
227		/**
228		* {@inheritdoc}
229		*/
230	3	public function count(): int
231		{
232	3	if (-1 === $this->nb_records) {
233	3	$this->nb_records = iterator_count($this->getRecords());
234		}
235
236	3	return $this->nb_records;
237		}
238
239		/**
240		* {@inheritdoc}
241		*/
242	3	public function getIterator(): Iterator
243		{
244	3	return $this->getRecords();
245		}
246
247		/**
248		* {@inheritdoc}
249		*/
250	3	public function jsonSerialize(): array
251		{
252	3	return iterator_to_array($this->getRecords(), false);
253		}
254
255		/**
256		* Returns the CSV records as an iterator object.
257		*
258		* Each CSV record is represented as a simple array containing strings or null values.
259		*
260		* If the CSV document has a header record then each record is combined
261		* to the header record and the header record is removed from the iterator.
262		*
263		* If the CSV document is inconsistent. Missing record fields are
264		* filled with null values while extra record fields are strip from
265		* the returned object.
266		*
267		* @param string[] $header an optional header to use instead of the CSV document header
268		*/
269	30	public function getRecords(array $header = []): Iterator
270		{
271	30	$header = $this->computeHeader($header);
272	27	$normalized = function ($record): bool {
273	27	return is_array($record) && (!$this->is_empty_records_skipped \|\| $record != [null]);
274	27	};
275	27	$bom = $this->getInputBOM();
276	27	$document = $this->getDocument();
277
278	27	$records = $this->stripBOM(new CallbackFilterIterator($document, $normalized), $bom);
279	27	if (null !== $this->header_offset) {
280	6	$records = new CallbackFilterIterator($records, function (array $record, int $offset): bool {
281	6	return $offset !== $this->header_offset;
282	6	});
283		}
284
285	27	if (!$this->is_empty_records_skipped) {
286	12	$normalized_empty_records = static function (array $record): array {
287	12	if ([null] === $record) {
288	12	return [];
289		}
290
291	12	return $record;
292	12	};
293
294	12	return $this->combineHeader(new MapIterator($records, $normalized_empty_records), $header);
295		}
296
297	27	return $this->combineHeader($records, $header);
298		}
299
300		/**
301		* Returns the header to be used for iteration.
302		*
303		* @param string[] $header
304		*
305		* @throws Exception If the header contains non unique column name
306		*
307		* @return string[]
308		*/
309	24	protected function computeHeader(array $header)
310		{
311	24	if ([] === $header) {
312	21	$header = $this->getHeader();
313		}
314
315	24	if ($header === array_unique(array_filter($header, 'is_string'))) {
316	21	return $header;
317		}
318
319	3	throw new Exception('The header record must be empty or a flat array with unique string values');
320		}
321
322		/**
323		* Combine the CSV header to each record if present.
324		*
325		* @param string[] $header
326		*/
327	30	protected function combineHeader(Iterator $iterator, array $header): Iterator
328		{
329	30	if ([] === $header) {
330	21	return $iterator;
331		}
332
333	12	$field_count = count($header);
334	12	$mapper = static function (array $record) use ($header, $field_count): array {
335	12	if (count($record) != $field_count) {
336	6	$record = array_slice(array_pad($record, $field_count, null), 0, $field_count);
337		}
338
339	12	return array_combine($header, $record);
340	12	};
341
342	12	return new MapIterator($iterator, $mapper);
343		}
344
345		/**
346		* Strip the BOM sequence from the returned records if necessary.
347		*/
348	24	protected function stripBOM(Iterator $iterator, string $bom): Iterator
349		{
350	24	if ('' === $bom) {
351	15	return $iterator;
352		}
353
354	9	$bom_length = mb_strlen($bom);
355	9	$mapper = function (array $record, int $index) use ($bom_length): array {
356	9	if (0 !== $index) {
357	3	return $record;
358		}
359
360	9	return $this->removeBOM($record, $bom_length, $this->enclosure);
361	9	};
362
363	9	return new MapIterator($iterator, $mapper);
364		}
365
366		/**
367		* Selects the record to be used as the CSV header.
368		*
369		* Because the header is represented as an array, to be valid
370		* a header MUST contain only unique string value.
371		*
372		* @param int\|null $offset the header record offset
373		*
374		* @throws Exception if the offset is a negative integer
375		*
376		* @return static
377		*/
378	24	public function setHeaderOffset($offset): self
379		{
380	24	if ($offset === $this->header_offset) {
381	12	return $this;
382		}
383
384	12	if (!is_nullable_int($offset)) {
385	3	throw new TypeError(sprintf(__METHOD__.'() expects 1 Argument to be null or an integer %s given', gettype($offset)));
386		}
387
388	9	if (null !== $offset && 0 > $offset) {
389	3	throw new Exception(__METHOD__.'() expects 1 Argument to be greater or equal to 0');
390		}
391
392	6	$this->header_offset = $offset;
393	6	$this->resetProperties();
394
395	6	return $this;
396		}
397
398		/**
399		* Enable skipping empty records.
400		*/
401	12	public function enableEmptyRecordsSkipping(): self
402		{
403	12	if (!$this->is_empty_records_skipped) {
404	12	$this->is_empty_records_skipped = true;
405	12	$this->nb_records = -1;
406		}
407
408	12	return $this;
409		}
410
411		/**
412		* Disable skipping empty records.
413		*/
414	12	public function disableEmptyRecordsSkipping(): self
415		{
416	12	if ($this->is_empty_records_skipped) {
417	12	$this->is_empty_records_skipped = false;
418	12	$this->nb_records = -1;
419		}
420
421	12	return $this;
422		}
423
424		/**
425		* Tells whether empty records are skipped by the instance.
426		*/
427	12	public function isEmptyRecordsSkipped(): bool
428		{
429	12	return $this->is_empty_records_skipped;
430		}
431		}
432

thephpleague / csv

Pull Request — master (#354)

Reader::stripBOM() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like