Reader::removeBOM() - Code Metrics - Inspection of "adding Reader::__call to reduce BC break (#230)" - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( a25e8e...8b5e7d )

by ignace nyamagana

created 2017-05-29 10:56 UTC

Reader::removeBOM() A

↳ Parent: Reader

Complexity

Conditions	4
Paths	3

Size

Total Lines	13
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	7
CRAP Score	4

Importance

Changes

Metric	Value
dl	0
loc	13
ccs	7
cts	7
cp	1
rs	9.2
c	0
b	0
f	0
cc	4
eloc	7
nc	3
nop	3
crap	4

<?php
/**
* This file is part of the League.csv library
*
* @license http://opensource.org/licenses/MIT
* @link https://github.com/thephpleague/csv/
* @version 9.0.0
* @package League.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 Iterator;
use IteratorAggregate;
use League\Csv\Exception\RuntimeException;
use LimitIterator;
use SplFileObject;

/**
 * A class to manage records selection from a CSV document
 *
 * @package League.csv
 * @since  3.0.0
 *
 * @method array fetchAll() Returns a sequential array of all CSV records
 * @method array fetchOne(int $offset = 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, string|int $value_index) Fetches the next key-value pairs from the CSV document
 */
class Reader extends AbstractCsv implements IteratorAggregate
{
    /**
     * @inheritdoc
     */
    protected $stream_filter_mode = STREAM_FILTER_READ;

    /**
     * CSV Document header offset
     *
     * @var int|null
     */
    protected $header_offset;

    /**
     * CSV Document Header record
     *
     * @var string[]
     */
    protected $header = [];

    /**
     * Tell whether the header needs to be re-generated
     *
     * @var bool
     */
    protected $is_header_loaded = false;

    /**
     * The value to pad if the record is less than header size.
     *
     * @var mixed
     */
    protected $record_padding_value;

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

    /**
     * Returns wether the selected header can be combine to each record
     *
     * A valid header must be empty or contains unique string field names
     *
     * @return bool
     */
    public function supportsHeaderAsRecordKeys(): bool
    {
        $header = $this->getHeader();

        return empty($header) || $header === array_unique(array_filter($header, 'is_string'));
    }

    /**
     * Returns the CSV record header
     *
     * The returned header is represented as an array of string values
     *
     * @return string[]
     */
    public function getHeader(): array
    {
        if ($this->is_header_loaded) {
            return $this->header;
        }

        $this->is_header_loaded = true;
        $this->header = [];
        if (null !== $this->header_offset) {
            $this->header = $this->setHeader($this->header_offset);
        }

        return $this->header;
    }

    /**
     * Determine the CSV record header
     *
     * @param int $offset
     *
     * @throws RuntimeException If the header offset is an integer
     *                          and the corresponding record is missing
     *                          or is an empty array
     *
     * @return string[]
     */
    protected function setHeader(int $offset): array
    {
        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
        $this->document->seek($offset);
        $header = $this->document->current();
        if (empty($header)) {
            throw new RuntimeException(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;
    }

    /**
     * Strip the BOM sequence from a record
     *
     * @param string[] $record
     * @param int      $bom_length
     * @param string   $enclosure
     *
     * @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 == mb_substr($record[0], 0, 1) && $enclosure == mb_substr($record[0], -1, 1)) {
            $record[0] = mb_substr($record[0], 1, -1);
        }

        return $record;
    }

    /**
     * Returns the record padding value
     *
     * @return mixed
     */
    public function getRecordPaddingValue()
    {
        return $this->record_padding_value;
    }

    /**
     * Returns a CSV records collection
     *
     * @param Statement $stmt
     *
     * @return ResultSet
     */
    public function select(Statement $stmt): ResultSet
    {
        return $stmt->process($this);
    }

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

        throw new BadMethodCallException(sprintf('Reader::%s does not exists', $method));
    }

    /**
     * Detect Delimiters occurences in the CSV
     *
     * Returns a associative array where each key represents
     * a valid delimiter and each value the number of occurences
     *
     * @param string[] $delimiters the delimiters to consider
     * @param int      $nb_records Detection is made using $nb_records of the CSV
     *
     * @return array
     */
    public function fetchDelimitersOccurrence(array $delimiters, int $nb_records = 1): array
    {
        $filter = function ($value): bool {
            return 1 == strlen($value);
        };

        $nb_records = $this->filterMinRange($nb_records, 1, 'The number of records to consider must be a valid positive integer');
        $delimiters = array_unique(array_filter($delimiters, $filter));
        $reducer = function (array $res, string $delimiter) use ($nb_records): array {
            $res[$delimiter] = $this->getCellCount($delimiter, $nb_records);

            return $res;
        };

        $res = array_reduce($delimiters, $reducer, []);
        arsort($res, SORT_NUMERIC);

        return $res;
    }

    /**
     * Returns the cell count for a specified delimiter
     * and a specified number of records
     *
     * @param string $delimiter  CSV delimiter
     * @param int    $nb_records CSV records to consider
     *
     * @return int
     */
    protected function getCellCount(string $delimiter, int $nb_records): int
    {
        $filter = function ($record): bool {
            return is_array($record) && count($record) > 1;
        };

        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($delimiter, $this->enclosure, $this->escape);
        $iterator = new CallbackFilterIterator(new LimitIterator($this->document, 0, $nb_records), $filter);

        return count(iterator_to_array($iterator, false), COUNT_RECURSIVE);
    }

    /**
     * @inheritdoc
     */
    public function getIterator(): Iterator
    {
        return $this->getRecords();
    }

    /**
     * Returns the CSV records in an iterator object.
     *
     * Each CSV record is represented as a simple array of string or null values.
     *
     * If the CSV document has a header record then each record is combined
     * to each 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.
     *
     * @throws RuntimeException If the header contains non unique column name
     *
     * @return Iterator
     */
    public function getRecords(): Iterator
    {
        if (!$this->supportsHeaderAsRecordKeys()) {
            throw new RuntimeException('The header record must be empty or a flat array with unique string values');
        }

        $normalized = function ($record): bool {
            return is_array($record) && $record != [null];
        };
        $bom = $this->getInputBOM();
        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);

        return $this->combineHeader($this->stripBOM(new CallbackFilterIterator($this->document, $normalized), $bom));
    }

    /**
     * Add the CSV header if present and valid
     *
     * @param Iterator $iterator
     *
     * @return Iterator
     */
    protected function combineHeader(Iterator $iterator): Iterator
    {
        if (null === $this->header_offset) {
            return $iterator;
        }

        $iterator = new CallbackFilterIterator($iterator, function (array $record, int $offset): bool {
            return $offset != $this->header_offset;
        });

        $header = $this->getHeader();
        $field_count = count($header);
        $mapper = function (array $record) use ($header, $field_count): array {
            if (count($record) != $field_count) {
                $record = array_slice(array_pad($record, $field_count, $this->record_padding_value), 0, $field_count);
            }

            return array_combine($header, $record);
        };

        return new MapIterator($iterator, $mapper);
    }

    /**
     * Strip the BOM sequence if present
     *
     * @param Iterator $iterator
     * @param string   $bom
     *
     * @return Iterator
     */
    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 of the header is represented as an array, to be valid
     * a header MUST contain only unique string value.
     *
     * @param int|null $offset the header record offset
     *
     * @return static
     */
    public function setHeaderOffset($offset): self
    {
        if (null !== $offset) {
            $offset = $this->filterMinRange($offset, 0, 'The header offset index must be a positive integer or 0');
        }

        if ($offset !== $this->header_offset) {
            $this->header_offset = $offset;
            $this->resetProperties();
        }

        return $this;
    }

    /**
     * Set the record padding value
     *
     * @param mixed $record_padding_value
     *
     * @return static
     */
    public function setRecordPaddingValue($record_padding_value): self
    {
        $this->record_padding_value = $record_padding_value;

        return $this;
    }

    /**
     * @inheritdoc
     */
    protected function resetProperties()
    {
        return $this->is_header_loaded = false;
    }
}


1		<?php
2		/**
3		* This file is part of the League.csv library
4		*
5		* @license http://opensource.org/licenses/MIT
6		* @link https://github.com/thephpleague/csv/
7		* @version 9.0.0
8		* @package League.csv
9		*
10		* For the full copyright and license information, please view the LICENSE
11		* file that was distributed with this source code.
12		*/
13		declare(strict_types=1);
14
15		namespace League\Csv;
16
17		use BadMethodCallException;
18		use CallbackFilterIterator;
19		use Iterator;
20		use IteratorAggregate;
21		use League\Csv\Exception\RuntimeException;
22		use LimitIterator;
23		use SplFileObject;
24
25		/**
26		* A class to manage records selection from a CSV document
27		*
28		* @package League.csv
29		* @since 3.0.0
30		*
31		* @method array fetchAll() Returns a sequential array of all CSV records
32		* @method array fetchOne(int $offset = 0) Returns a single record from the CSV
33		* @method Generator fetchColumn(string\|int $column_index) Returns the next value from a single CSV record field
34		* @method Generator fetchPairs(string\|int $offset_index, string\|int $value_index) Fetches the next key-value pairs from the CSV document
35		*/
36		class Reader extends AbstractCsv implements IteratorAggregate
37		{
38		/**
39		* @inheritdoc
40		*/
41		protected $stream_filter_mode = STREAM_FILTER_READ;
42
43		/**
44		* CSV Document header offset
45		*
46		* @var int\|null
47		*/
48		protected $header_offset;
49
50		/**
51		* CSV Document Header record
52		*
53		* @var string[]
54		*/
55		protected $header = [];
56
57		/**
58		* Tell whether the header needs to be re-generated
59		*
60		* @var bool
61		*/
62		protected $is_header_loaded = false;
63
64		/**
65		* The value to pad if the record is less than header size.
66		*
67		* @var mixed
68		*/
69		protected $record_padding_value;
70
71		/**
72		* Returns the record offset used as header
73		*
74		* If no CSV record is used this method MUST return null
75		*
76		* @return int\|null
77		*/
78	2	public function getHeaderOffset()
79		{
80	2	return $this->header_offset;
81		}
82
83		/**
84		* Returns wether the selected header can be combine to each record
85		*
86		* A valid header must be empty or contains unique string field names
87		*
88		* @return bool
89		*/
90	2	public function supportsHeaderAsRecordKeys(): bool
91		{
92	2	$header = $this->getHeader();
93
94	2	return empty($header) \|\| $header === array_unique(array_filter($header, 'is_string'));
95		}
96
97		/**
98		* Returns the CSV record header
99		*
100		* The returned header is represented as an array of string values
101		*
102		* @return string[]
103		*/
104	4	public function getHeader(): array
105		{
106	4	if ($this->is_header_loaded) {
107	2	return $this->header;
108		}
109
110	4	$this->is_header_loaded = true;
111	4	$this->header = [];
112	4	if (null !== $this->header_offset) {
113	4	$this->header = $this->setHeader($this->header_offset);
114		}
115
116	4	return $this->header;
117		}
118
119		/**
120		* Determine the CSV record header
121		*
122		* @param int $offset
123		*
124		* @throws RuntimeException If the header offset is an integer
125		* and the corresponding record is missing
126		* or is an empty array
127		*
128		* @return string[]
129		*/
130	6	protected function setHeader(int $offset): array
131		{
132	6	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
133	6	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
134	6	$this->document->seek($offset);
135	6	$header = $this->document->current();
136	6	if (empty($header)) {
137	2	throw new RuntimeException(sprintf('The header record does not exist or is empty at offset: `%s`', $offset));
138		}
139
140	4	if (0 === $offset) {
141	2	return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
142		}
143
144	2	return $header;
145		}
146
147		/**
148		* Strip the BOM sequence from a record
149		*
150		* @param string[] $record
151		* @param int $bom_length
152		* @param string $enclosure
153		*
154		* @return string[]
155		*/
156	8	protected function removeBOM(array $record, int $bom_length, string $enclosure): array
157		{
158	8	if (0 == $bom_length) {
159	2	return $record;
160		}
161
162	6	$record[0] = mb_substr($record[0], $bom_length);
163	6	if ($enclosure == mb_substr($record[0], 0, 1) && $enclosure == mb_substr($record[0], -1, 1)) {
164	2	$record[0] = mb_substr($record[0], 1, -1);
165		}
166
167	6	return $record;
168		}
169
170		/**
171		* Returns the record padding value
172		*
173		* @return mixed
174		*/
175	2	public function getRecordPaddingValue()
176		{
177	2	return $this->record_padding_value;
178		}
179
180		/**
181		* Returns a CSV records collection
182		*
183		* @param Statement $stmt
184		*
185		* @return ResultSet
186		*/
187	2	public function select(Statement $stmt): ResultSet
188		{
189	2	return $stmt->process($this);
190		}
191
192		/**
193		* @inheritdoc
194		*/
195	6	public function __call($method, array $arguments)
196		{
197	6	$whitelisted = ['fetchColumn' => 1, 'fetchPairs' => 1, 'fetchOne' => 1, 'fetchAll' => 1];
198	6	if (isset($whitelisted[$method])) {
199	2	return (new ResultSet($this->getRecords(), $this->getHeader()))
200	2	->$method(...$arguments)
201		;
202		}
203
204	4	throw new BadMethodCallException(sprintf('Reader::%s does not exists', $method));
205		}
206
207		/**
208		* Detect Delimiters occurences in the CSV
209		*
210		* Returns a associative array where each key represents
211		* a valid delimiter and each value the number of occurences
212		*
213		* @param string[] $delimiters the delimiters to consider
214		* @param int $nb_records Detection is made using $nb_records of the CSV
215		*
216		* @return array
217		*/
218	6	public function fetchDelimitersOccurrence(array $delimiters, int $nb_records = 1): array
219		{
220		$filter = function ($value): bool {
221	4	return 1 == strlen($value);
222	3	};
223
224	6	$nb_records = $this->filterMinRange($nb_records, 1, 'The number of records to consider must be a valid positive integer');
225	4	$delimiters = array_unique(array_filter($delimiters, $filter));
226		$reducer = function (array $res, string $delimiter) use ($nb_records): array {
227	4	$res[$delimiter] = $this->getCellCount($delimiter, $nb_records);
228
229	4	return $res;
230	4	};
231
232	4	$res = array_reduce($delimiters, $reducer, []);
233	4	arsort($res, SORT_NUMERIC);
234
235	4	return $res;
236		}
237
238		/**
239		* Returns the cell count for a specified delimiter
240		* and a specified number of records
241		*
242		* @param string $delimiter CSV delimiter
243		* @param int $nb_records CSV records to consider
244		*
245		* @return int
246		*/
247	2	protected function getCellCount(string $delimiter, int $nb_records): int
248		{
249		$filter = function ($record): bool {
250	2	return is_array($record) && count($record) > 1;
251	1	};
252
253	2	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
254	2	$this->document->setCsvControl($delimiter, $this->enclosure, $this->escape);
255	2	$iterator = new CallbackFilterIterator(new LimitIterator($this->document, 0, $nb_records), $filter);
256
257	2	return count(iterator_to_array($iterator, false), COUNT_RECURSIVE);
258		}
259
260		/**
261		* @inheritdoc
262		*/
263	18	public function getIterator(): Iterator
264		{
265	18	return $this->getRecords();
266		}
267
268		/**
269		* Returns the CSV records in an iterator object.
270		*
271		* Each CSV record is represented as a simple array of string or null values.
272		*
273		* If the CSV document has a header record then each record is combined
274		* to each header record and the header record is removed from the iterator.
275		*
276		* If the CSV document is inconsistent. Missing record fields are
277		* filled with null values while extra record fields are strip from
278		* the returned object.
279		*
280		* @throws RuntimeException If the header contains non unique column name
281		*
282		* @return Iterator
283		*/
284	4	public function getRecords(): Iterator
285		{
286	4	if (!$this->supportsHeaderAsRecordKeys()) {
287	2	throw new RuntimeException('The header record must be empty or a flat array with unique string values');
288		}
289
290		$normalized = function ($record): bool {
291	2	return is_array($record) && $record != [null];
292	1	};
293	2	$bom = $this->getInputBOM();
294	2	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
295	2	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
296
297	2	return $this->combineHeader($this->stripBOM(new CallbackFilterIterator($this->document, $normalized), $bom));
298		}
299
300		/**
301		* Add the CSV header if present and valid
302		*
303		* @param Iterator $iterator
304		*
305		* @return Iterator
306		*/
307	14	protected function combineHeader(Iterator $iterator): Iterator
308		{
309	14	if (null === $this->header_offset) {
310	8	return $iterator;
311		}
312
313		$iterator = new CallbackFilterIterator($iterator, function (array $record, int $offset): bool {
314	8	return $offset != $this->header_offset;
315	8	});
316
317	8	$header = $this->getHeader();
318	8	$field_count = count($header);
319		$mapper = function (array $record) use ($header, $field_count): array {
320	8	if (count($record) != $field_count) {
321	4	$record = array_slice(array_pad($record, $field_count, $this->record_padding_value), 0, $field_count);
322		}
323
324	8	return array_combine($header, $record);
325	8	};
326
327	8	return new MapIterator($iterator, $mapper);
328		}
329
330		/**
331		* Strip the BOM sequence if present
332		*
333		* @param Iterator $iterator
334		* @param string $bom
335		*
336		* @return Iterator
337		*/
338	10	protected function stripBOM(Iterator $iterator, string $bom): Iterator
339		{
340	10	if ('' === $bom) {
341	4	return $iterator;
342		}
343
344	6	$bom_length = mb_strlen($bom);
345	6	$mapper = function (array $record, int $index) use ($bom_length): array {
346	6	if (0 != $index) {
347	2	return $record;
348		}
349
350	6	return $this->removeBOM($record, $bom_length, $this->enclosure);
351	6	};
352
353	6	return new MapIterator($iterator, $mapper);
354		}
355
356
357		/**
358		* Selects the record to be used as the CSV header
359		*
360		* Because of the header is represented as an array, to be valid
361		* a header MUST contain only unique string value.
362		*
363		* @param int\|null $offset the header record offset
364		*
365		* @return static
366		*/
367	2	public function setHeaderOffset($offset): self
368		{
369	2	if (null !== $offset) {
370	2	$offset = $this->filterMinRange($offset, 0, 'The header offset index must be a positive integer or 0');
371		}
372
373	2	if ($offset !== $this->header_offset) {
374	2	$this->header_offset = $offset;
375	2	$this->resetProperties();
376		}
377
378	2	return $this;
379		}
380
381		/**
382		* Set the record padding value
383		*
384		* @param mixed $record_padding_value
385		*
386		* @return static
387		*/
388	2	public function setRecordPaddingValue($record_padding_value): self
389		{
390	2	$this->record_padding_value = $record_padding_value;
391
392	2	return $this;
393		}
394
395		/**
396		* @inheritdoc
397		*/
398	2	protected function resetProperties()
399		{
400	2	return $this->is_header_loaded = false;
401		}
402		}
403

thephpleague / csv

Push — master ( a25e8e...8b5e7d )

Reader::removeBOM() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like