Reader::fetchColumn() - Code Metrics - Inspection of "Introducing the TabularDataReader interface" - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#379)

by ignace nyamagana

created 2020-02-21 00:00 UTC

Reader::fetchColumn() A

↳ Parent: Reader

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	1

Importance

Changes

Metric	Value
cc	1
nc	1
nop	1
dl	0
loc	6
ccs	3
cts	3
cp	1
crap	1
rs	10
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 CallbackFilterIterator;
use Iterator;
use IteratorAggregate;
use JsonSerializable;
use League\Csv\Polyfill\EmptyEscapeParser;
use SplFileObject;
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 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.
 */
class Reader extends AbstractCsv implements TabularDataReader, 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_included = false;

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

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

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

    /**
     * {@inheritDoc}
     */
    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 (in_array($header, [[], [null]], true)) {
            throw new SyntaxError(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.
     */
    protected function seekRow(int $offset): array
    {
        foreach ($this->getDocument() as $index => $record) {
            if ($offset === $index) {
                return $record;
            }
        }

        return [];
    }

    /**
     * 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 fetchColumn($index = 0): Iterator
    {
        $tabular_data = new ResultSet($this->getRecords(), $this->getHeader());

        return $tabular_data->fetchColumn($index);
    }

    /**
     * {@inheritdoc}
     */
    public function fetchOne(int $nth_record = 0): array
    {
        $tabular_data = new ResultSet($this->getRecords(), $this->getHeader());

        return $tabular_data->fetchOne($nth_record);
    }

    /**
     * {@inheritdoc}
     */
    public function fetchPairs($offset_index = 0, $value_index = 1): Iterator
    {
        $tabular_data = new ResultSet($this->getRecords(), $this->getHeader());

        return $tabular_data->fetchPairs($offset_index, $value_index);
    }

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

        $bom = '';
        if (!$this->is_input_bom_included) {
            $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_included) {
            $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 SyntaxError('The header record must be an 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);
            }

            /** @var array<string|null> $assocRecord */
            $assocRecord = array_combine($header, $record);

            return $assocRecord;
        };

        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(?int $offset): self
    {
        if ($offset === $this->header_offset) {
            return $this;
        }

        if (null !== $offset && 0 > $offset) {
            throw new InvalidArgument(__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 skipEmptyRecords(): self
    {
        if ($this->is_empty_records_included) {
            $this->is_empty_records_included = false;
            $this->nb_records = -1;
        }

        return $this;
    }

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

        return $this;
    }

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


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 CallbackFilterIterator;
17		use Iterator;
18		use IteratorAggregate;
19		use JsonSerializable;
20		use League\Csv\Polyfill\EmptyEscapeParser;
21		use SplFileObject;
22		use function array_combine;
23		use function array_filter;
24		use function array_pad;
25		use function array_slice;
26		use function array_unique;
27		use function count;
28		use function is_array;
29		use function iterator_count;
30		use function iterator_to_array;
31		use function mb_strlen;
32		use function mb_substr;
33		use function sprintf;
34		use function strlen;
35		use function substr;
36		use const PHP_VERSION_ID;
37		use const STREAM_FILTER_READ;
38
39		/**
40		* A class to parse and read records from a CSV document.
41		*/
42		class Reader extends AbstractCsv implements TabularDataReader, IteratorAggregate, JsonSerializable
43		{
44		/**
45		* header offset.
46		*
47		* @var int\|null
48		*/
49		protected $header_offset;
50
51		/**
52		* header record.
53		*
54		* @var string[]
55		*/
56		protected $header = [];
57
58		/**
59		* records count.
60		*
61		* @var int
62		*/
63		protected $nb_records = -1;
64
65		/**
66		* {@inheritdoc}
67		*/
68		protected $stream_filter_mode = STREAM_FILTER_READ;
69
70		/**
71		* @var bool
72		*/
73		protected $is_empty_records_included = false;
74
75		/**
76		* {@inheritdoc}
77		*/
78		public static function createFromPath(string $path, string $open_mode = 'r', $context = null)
79		{
80		return parent::createFromPath($path, $open_mode, $context);
81		}
82
83		/**
84		* {@inheritdoc}
85	3	*/
86		protected function resetProperties(): void
87	3	{
88		parent::resetProperties();
89		$this->nb_records = -1;
90		$this->header = [];
91		}
92
93	33	/**
94		* Returns the header offset.
95	33	*
96	33	* If no CSV header offset is set this method MUST return null
97	33	*
98	33	*/
99		public function getHeaderOffset(): ?int
100		{
101		return $this->header_offset;
102		}
103
104		/**
105		* {@inheritDoc}
106	21	*/
107		public function getHeader(): array
108	21	{
109		if (null === $this->header_offset) {
110		return $this->header;
111		}
112
113		if ([] !== $this->header) {
114		return $this->header;
115		}
116
117		$this->header = $this->setHeader($this->header_offset);
118	24
119		return $this->header;
120	24	}
121	18
122		/**
123		* Determine the CSV record header.
124	9	*
125	3	* @throws Exception If the header offset is set and no record is found or is the empty array
126		*
127		* @return string[]
128	9	*/
129		protected function setHeader(int $offset): array
130	6	{
131		$header = $this->seekRow($offset);
132		if (in_array($header, [[], [null]], true)) {
133		throw new SyntaxError(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
134		}
135
136		if (0 === $offset) {
137		return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
138		}
139
140	12	return $header;
141		}
142	12
143	12	/**
144	6	* Returns the row at a given offset.
145		*/
146		protected function seekRow(int $offset): array
147	6	{
148	3	foreach ($this->getDocument() as $index => $record) {
149		if ($offset === $index) {
150		return $record;
151	3	}
152		}
153
154		return [];
155		}
156
157		/**
158	12	* Returns the document as an Iterator.
159		*/
160	12	protected function getDocument(): Iterator
161	12	{
162	6	if (70400 > PHP_VERSION_ID && '' === $this->escape) {
163		$this->document->setCsvControl($this->delimiter, $this->enclosure);
164
165		return EmptyEscapeParser::parse($this->document);
166	6	}
167
168		$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD);
169		$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
170		$this->document->rewind();
171
172	21	return $this->document;
173		}
174	21
175	4	/**
176		* Strip the BOM sequence from a record.
177	4	*
178		* @param string[] $record
179		*
180	17	* @return string[]
181	17	*/
182	17	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
183		{
184	17	if (0 === $bom_length) {
185		return $record;
186		}
187
188		$record[0] = mb_substr($record[0], $bom_length);
189		if ($enclosure.$enclosure != substr($record[0].$record[0], strlen($record[0]) - 1, 2)) {
190		return $record;
191		}
192
193		$record[0] = substr($record[0], 1, -1);
194	12
195		return $record;
196	12	}
197	3
198		/**
199		* {@inheritdoc}
200	9	*/
201	9	public function fetchColumn($index = 0): Iterator
202	6	{
203		$tabular_data = new ResultSet($this->getRecords(), $this->getHeader());
204
205	3	return $tabular_data->fetchColumn($index);
206		}
207	3
208		/**
209		* {@inheritdoc}
210		*/
211		public function fetchOne(int $nth_record = 0): array
212		{
213	9	$tabular_data = new ResultSet($this->getRecords(), $this->getHeader());
214
215	9	return $tabular_data->fetchOne($nth_record);
216	9	}
217	3
218		/**
219		* {@inheritdoc}
220	6	*/
221		public function fetchPairs($offset_index = 0, $value_index = 1): Iterator
222		{
223		$tabular_data = new ResultSet($this->getRecords(), $this->getHeader());
224
225		return $tabular_data->fetchPairs($offset_index, $value_index);
226	3	}
227
228	3	/**
229	3	* {@inheritdoc}
230		*/
231		public function count(): int
232	3	{
233		if (-1 === $this->nb_records) {
234		$this->nb_records = iterator_count($this->getRecords());
235		}
236
237		return $this->nb_records;
238	6	}
239
240	6	/**
241		* {@inheritdoc}
242		*/
243		public function getIterator(): Iterator
244		{
245		return $this->getRecords();
246	3	}
247
248	3	/**
249		* {@inheritdoc}
250		*/
251		public function jsonSerialize(): array
252		{
253		return iterator_to_array($this->getRecords(), false);
254		}
255
256		/**
257		* Returns the CSV records as an iterator object.
258		*
259		* Each CSV record is represented as a simple array containing strings or null values.
260		*
261		* If the CSV document has a header record then each record is combined
262		* to the header record and the header record is removed from the iterator.
263		*
264		* If the CSV document is inconsistent. Missing record fields are
265	36	* filled with null values while extra record fields are strip from
266		* the returned object.
267	36	*
268		* @param string[] $header an optional header to use instead of the CSV document header
269	33	*/
270	33	public function getRecords(array $header = []): Iterator
271		{
272	33	$header = $this->computeHeader($header);
273	33	$normalized = function ($record): bool {
274	30	return is_array($record) && ($this->is_empty_records_included \|\| $record != [null]);
275		};
276
277	33	$bom = '';
278	33	if (!$this->is_input_bom_included) {
279	33	$bom = $this->getInputBOM();
280		}
281	18
282	18	$document = $this->getDocument();
283		$records = $this->stripBOM(new CallbackFilterIterator($document, $normalized), $bom);
284		if (null !== $this->header_offset) {
285	33	$records = new CallbackFilterIterator($records, function (array $record, int $offset): bool {
286		return $offset !== $this->header_offset;
287	12	});
288	12	}
289
290		if ($this->is_empty_records_included) {
291	12	$normalized_empty_records = static function (array $record): array {
292	12	if ([null] === $record) {
293		return [];
294	12	}
295
296		return $record;
297	33	};
298
299		return $this->combineHeader(new MapIterator($records, $normalized_empty_records), $header);
300		}
301
302		return $this->combineHeader($records, $header);
303		}
304
305		/**
306		* Returns the header to be used for iteration.
307		*
308		* @param string[] $header
309	30	*
310		* @throws Exception If the header contains non unique column name
311	30	*
312	27	* @return string[]
313		*/
314		protected function computeHeader(array $header)
315	30	{
316	27	if ([] === $header) {
317		$header = $this->getHeader();
318		}
319	3
320		if ($header === array_unique(array_filter($header, 'is_string'))) {
321		return $header;
322		}
323
324		throw new SyntaxError('The header record must be an empty or a flat array with unique string values.');
325		}
326
327	36	/**
328		* Combine the CSV header to each record if present.
329	36	*
330	27	* @param string[] $header
331		*/
332		protected function combineHeader(Iterator $iterator, array $header): Iterator
333	12	{
334		if ([] === $header) {
335	12	return $iterator;
336	6	}
337
338		$field_count = count($header);
339		$mapper = static function (array $record) use ($header, $field_count): array {
340	12	if (count($record) != $field_count) {
341		$record = array_slice(array_pad($record, $field_count, null), 0, $field_count);
342	12	}
343	12
344		/** @var array<string\|null> $assocRecord */
345	12	$assocRecord = array_combine($header, $record);
346
347		return $assocRecord;
348		};
349
350		return new MapIterator($iterator, $mapper);
351	30	}
352
353	30	/**
354	21	* Strip the BOM sequence from the returned records if necessary.
355		*/
356		protected function stripBOM(Iterator $iterator, string $bom): Iterator
357	9	{
358		if ('' === $bom) {
359	9	return $iterator;
360	3	}
361
362		$bom_length = mb_strlen($bom);
363	9	$mapper = function (array $record, int $index) use ($bom_length): array {
364	9	if (0 !== $index) {
365		return $record;
366	9	}
367
368		return $this->removeBOM($record, $bom_length, $this->enclosure);
369		};
370
371		return new MapIterator($iterator, $mapper);
372		}
373
374		/**
375		* Selects the record to be used as the CSV header.
376		*
377		* Because the header is represented as an array, to be valid
378		* a header MUST contain only unique string value.
379		*
380		* @param int\|null $offset the header record offset
381	27	*
382		* @throws Exception if the offset is a negative integer
383	27	*
384	18	* @return static
385		*/
386		public function setHeaderOffset(?int $offset): self
387	9	{
388	3	if ($offset === $this->header_offset) {
389		return $this;
390		}
391	6
392	6	if (null !== $offset && 0 > $offset) {
393		throw new InvalidArgument(__METHOD__.'() expects 1 Argument to be greater or equal to 0');
394	6	}
395
396		$this->header_offset = $offset;
397		$this->resetProperties();
398
399		return $this;
400	12	}
401
402	12	/**
403	12	* Enable skipping empty records.
404	12	*/
405		public function skipEmptyRecords(): self
406		{
407	12	if ($this->is_empty_records_included) {
408		$this->is_empty_records_included = false;
409		$this->nb_records = -1;
410		}
411
412		return $this;
413	12	}
414
415	12	/**
416	12	* Disable skipping empty records.
417	12	*/
418		public function includeEmptyRecords(): self
419		{
420	12	if (!$this->is_empty_records_included) {
421		$this->is_empty_records_included = true;
422		$this->nb_records = -1;
423		}
424
425		return $this;
426	12	}
427
428	12	/**
429		* Tells whether empty records are skipped by the instance.
430		*/
431		public function isEmptyRecordsIncluded(): bool
432		{
433		return $this->is_empty_records_included;
434		}
435		}
436

thephpleague / csv

Pull Request — master (#379)

Reader::fetchColumn() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like