Reader::setHeader() - 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 00:09 UTC

Reader::setHeader() A

↳ Parent: Reader

Complexity

Conditions	3
Paths	3

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	7
CRAP Score	3

Importance

Changes

Metric	Value
cc	3
nc	3
nop	1
dl	0
loc	13
ccs	7
cts	7
cp	1
crap	3
rs	9.8333
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 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, 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 JsonSerializable;
19		use League\Csv\Polyfill\EmptyEscapeParser;
20		use SplFileObject;
21		use function array_combine;
22		use function array_filter;
23		use function array_pad;
24		use function array_slice;
25		use function array_unique;
26		use function count;
27		use function is_array;
28		use function iterator_count;
29		use function iterator_to_array;
30		use function mb_strlen;
31		use function mb_substr;
32		use function sprintf;
33		use function strlen;
34		use function substr;
35		use const PHP_VERSION_ID;
36		use const STREAM_FILTER_READ;
37
38		/**
39		* A class to parse and read records from a CSV document.
40		*/
41		class Reader extends AbstractCsv implements TabularDataReader, JsonSerializable
42		{
43		/**
44		* header offset.
45		*
46		* @var int\|null
47		*/
48		protected $header_offset;
49
50		/**
51		* header record.
52		*
53		* @var string[]
54		*/
55		protected $header = [];
56
57		/**
58		* records count.
59		*
60		* @var int
61		*/
62		protected $nb_records = -1;
63
64		/**
65		* {@inheritdoc}
66		*/
67		protected $stream_filter_mode = STREAM_FILTER_READ;
68
69		/**
70		* @var bool
71		*/
72		protected $is_empty_records_included = false;
73
74		/**
75		* {@inheritdoc}
76		*/
77	3	public static function createFromPath(string $path, string $open_mode = 'r', $context = null)
78		{
79	3	return parent::createFromPath($path, $open_mode, $context);
80		}
81
82		/**
83		* {@inheritdoc}
84		*/
85	30	protected function resetProperties(): void
86		{
87	30	parent::resetProperties();
88	30	$this->nb_records = -1;
89	30	$this->header = [];
90	30	}
91
92		/**
93		* Returns the header offset.
94		*
95		* If no CSV header offset is set this method MUST return null
96		*
97		*/
98	21	public function getHeaderOffset(): ?int
99		{
100	21	return $this->header_offset;
101		}
102
103		/**
104		* {@inheritDoc}
105		*/
106	24	public function getHeader(): array
107		{
108	24	if (null === $this->header_offset) {
109	18	return $this->header;
110		}
111
112	9	if ([] !== $this->header) {
113	3	return $this->header;
114		}
115
116	9	$this->header = $this->setHeader($this->header_offset);
117
118	6	return $this->header;
119		}
120
121		/**
122		* Determine the CSV record header.
123		*
124		* @throws Exception If the header offset is set and no record is found or is the empty array
125		*
126		* @return string[]
127		*/
128	12	protected function setHeader(int $offset): array
129		{
130	12	$header = $this->seekRow($offset);
131	12	if (in_array($header, [[], [null]], true)) {
132	6	throw new SyntaxError(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
133		}
134
135	6	if (0 === $offset) {
136	3	return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
137		}
138
139	3	return $header;
140		}
141
142		/**
143		* Returns the row at a given offset.
144		*/
145	12	protected function seekRow(int $offset): array
146		{
147	12	foreach ($this->getDocument() as $index => $record) {
148	12	if ($offset === $index) {
149	6	return $record;
150		}
151		}
152
153	6	return [];
154		}
155
156		/**
157		* Returns the document as an Iterator.
158		*/
159	21	protected function getDocument(): Iterator
160		{
161	21	if (70400 > PHP_VERSION_ID && '' === $this->escape) {
162	6	$this->document->setCsvControl($this->delimiter, $this->enclosure);
163
164	6	return EmptyEscapeParser::parse($this->document);
165		}
166
167	15	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD);
168	15	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
169	15	$this->document->rewind();
170
171	15	return $this->document;
172		}
173
174		/**
175		* Strip the BOM sequence from a record.
176		*
177		* @param string[] $record
178		*
179		* @return string[]
180		*/
181	12	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
182		{
183	12	if (0 === $bom_length) {
184	3	return $record;
185		}
186
187	9	$record[0] = mb_substr($record[0], $bom_length);
188	9	if ($enclosure.$enclosure != substr($record[0].$record[0], strlen($record[0]) - 1, 2)) {
189	6	return $record;
190		}
191
192	3	$record[0] = substr($record[0], 1, -1);
193
194	3	return $record;
195		}
196
197		/**
198		* {@inheritdoc}
199		*/
200	3	public function fetchColumn($index = 0): Iterator
201		{
202	3	$tabular_data = new ResultSet($this->getRecords(), $this->getHeader());
203
204	3	return $tabular_data->fetchColumn($index);
205		}
206
207		/**
208		* {@inheritdoc}
209		*/
210	3	public function fetchOne(int $nth_record = 0): array
211		{
212	3	$tabular_data = new ResultSet($this->getRecords(), $this->getHeader());
213
214	3	return $tabular_data->fetchOne($nth_record);
215		}
216
217		/**
218		* {@inheritdoc}
219		*/
220	3	public function fetchPairs($offset_index = 0, $value_index = 1): Iterator
221		{
222	3	$tabular_data = new ResultSet($this->getRecords(), $this->getHeader());
223
224	3	return $tabular_data->fetchPairs($offset_index, $value_index);
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	6	public function getIterator(): Iterator
243		{
244	6	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	36	public function getRecords(array $header = []): Iterator
270		{
271	36	$header = $this->computeHeader($header);
272		$normalized = function ($record): bool {
273	33	return is_array($record) && ($this->is_empty_records_included \|\| $record != [null]);
274	33	};
275
276	33	$bom = '';
277	33	if (!$this->is_input_bom_included) {
278	30	$bom = $this->getInputBOM();
279		}
280
281	33	$document = $this->getDocument();
282	33	$records = $this->stripBOM(new CallbackFilterIterator($document, $normalized), $bom);
283	33	if (null !== $this->header_offset) {
284		$records = new CallbackFilterIterator($records, function (array $record, int $offset): bool {
285	18	return $offset !== $this->header_offset;
286	18	});
287		}
288
289	33	if ($this->is_empty_records_included) {
290		$normalized_empty_records = static function (array $record): array {
291	12	if ([null] === $record) {
292	12	return [];
293		}
294
295	12	return $record;
296	12	};
297
298	12	return $this->combineHeader(new MapIterator($records, $normalized_empty_records), $header);
299		}
300
301	33	return $this->combineHeader($records, $header);
302		}
303
304		/**
305		* Returns the header to be used for iteration.
306		*
307		* @param string[] $header
308		*
309		* @throws Exception If the header contains non unique column name
310		*
311		* @return string[]
312		*/
313	30	protected function computeHeader(array $header)
314		{
315	30	if ([] === $header) {
316	27	$header = $this->getHeader();
317		}
318
319	30	if ($header === array_unique(array_filter($header, 'is_string'))) {
320	27	return $header;
321		}
322
323	3	throw new SyntaxError('The header record must be an empty or a flat array with unique string values.');
324		}
325
326		/**
327		* Combine the CSV header to each record if present.
328		*
329		* @param string[] $header
330		*/
331	36	protected function combineHeader(Iterator $iterator, array $header): Iterator
332		{
333	36	if ([] === $header) {
334	27	return $iterator;
335		}
336
337	12	$field_count = count($header);
338		$mapper = static function (array $record) use ($header, $field_count): array {
339	12	if (count($record) != $field_count) {
340	6	$record = array_slice(array_pad($record, $field_count, null), 0, $field_count);
341		}
342
343		/** @var array<string\|null> $assocRecord */
344	12	$assocRecord = array_combine($header, $record);
345
346	12	return $assocRecord;
347	12	};
348
349	12	return new MapIterator($iterator, $mapper);
350		}
351
352		/**
353		* Strip the BOM sequence from the returned records if necessary.
354		*/
355	30	protected function stripBOM(Iterator $iterator, string $bom): Iterator
356		{
357	30	if ('' === $bom) {
358	21	return $iterator;
359		}
360
361	9	$bom_length = mb_strlen($bom);
362		$mapper = function (array $record, int $index) use ($bom_length): array {
363	9	if (0 !== $index) {
364	3	return $record;
365		}
366
367	9	return $this->removeBOM($record, $bom_length, $this->enclosure);
368	9	};
369
370	9	return new MapIterator($iterator, $mapper);
371		}
372
373		/**
374		* Selects the record to be used as the CSV header.
375		*
376		* Because the header is represented as an array, to be valid
377		* a header MUST contain only unique string value.
378		*
379		* @param int\|null $offset the header record offset
380		*
381		* @throws Exception if the offset is a negative integer
382		*
383		* @return static
384		*/
385	27	public function setHeaderOffset(?int $offset): self
386		{
387	27	if ($offset === $this->header_offset) {
388	18	return $this;
389		}
390
391	9	if (null !== $offset && 0 > $offset) {
392	3	throw new InvalidArgument(__METHOD__.'() expects 1 Argument to be greater or equal to 0');
393		}
394
395	6	$this->header_offset = $offset;
396	6	$this->resetProperties();
397
398	6	return $this;
399		}
400
401		/**
402		* Enable skipping empty records.
403		*/
404	12	public function skipEmptyRecords(): self
405		{
406	12	if ($this->is_empty_records_included) {
407	12	$this->is_empty_records_included = false;
408	12	$this->nb_records = -1;
409		}
410
411	12	return $this;
412		}
413
414		/**
415		* Disable skipping empty records.
416		*/
417	12	public function includeEmptyRecords(): self
418		{
419	12	if (!$this->is_empty_records_included) {
420	12	$this->is_empty_records_included = true;
421	12	$this->nb_records = -1;
422		}
423
424	12	return $this;
425		}
426
427		/**
428		* Tells whether empty records are skipped by the instance.
429		*/
430	12	public function isEmptyRecordsIncluded(): bool
431		{
432	12	return $this->is_empty_records_included;
433		}
434		}
435

thephpleague / csv

Pull Request — master (#379)

Reader::setHeader() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like