Reader::getFieldIndex() - Code Metrics - Inspection of "Implementing 9.0 Public API" - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#210)

by ignace nyamagana

created 2017-02-23 08:20 UTC

Reader::getFieldIndex() A

↳ Parent: Reader

Complexity

Conditions	4
Paths	4

Size

Total Lines	17
Code Lines	9

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	8
CRAP Score	4

Importance

Changes

Metric	Value
dl	0
loc	17
rs	9.2
c	0
b	0
f	0
ccs	8
cts	8
cp	1
cc	4
eloc	9
nc	4
nop	2
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 CallbackFilterIterator;
use Iterator;
use IteratorAggregate;
use League\Csv\Exception\InvalidArgumentException;
use LimitIterator;
use SplFileObject;

/**
 *  A class to manage extracting and filtering a CSV
 *
 * @package League.csv
 * @since  3.0.0
 *
 */
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;

    /**
     * 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;
    }

    /**
     * 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 $offset the header row offset
     *
     * @return static
     */
    public function setHeaderOffset($offset): self
    {
        $this->header_offset = null;
        if (null !== $offset) {
            $this->header_offset = $this->filterInteger(
                $offset,
                0,
                'the header offset index must be a positive integer or 0'
            );
        }
        $this->resetDynamicProperties();

        return $this;
    }

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

    /**
     * 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_rows    Detection is made using $nb_rows of the CSV
     *
     * @return array
     */
    public function fetchDelimitersOccurrence(array $delimiters, int $nb_rows = 1): array
    {
        $nb_rows = $this->filterInteger($nb_rows, 1, 'The number of rows to consider must be a valid positive integer');
        $filter_row = function ($row) {
            return is_array($row) && count($row) > 1;
        };
        $delimiters = array_unique(array_filter($delimiters, function ($value) {
            return 1 == strlen($value);
        }));
        $this->document->setFlags(SplFileObject::READ_CSV);
        $res = [];
        foreach ($delimiters as $delim) {
            $this->document->setCsvControl($delim, $this->enclosure, $this->escape);
            $iterator = new CallbackFilterIterator(new LimitIterator($this->document, 0, $nb_rows), $filter_row);
            $res[$delim] = count(iterator_to_array($iterator, false), COUNT_RECURSIVE);
        }
        arsort($res, SORT_NUMERIC);

        return $res;
    }

    /**
     * Returns a collection of selected records
     *
     * @param Statement|null $stmt
     *
     * @return RecordSet
     */
    public function select(Statement $stmt = null): RecordSet
    {
        $stmt = $stmt ?? new Statement();

        return $stmt->process($this);
    }

    /**
     * @inheritdoc
     */
    public function getIterator(): Iterator
    {
        $bom = $this->getInputBOM();
        $header = $this->getHeader();
        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
        $normalized = function ($row) {
            return is_array($row) && $row != [null];
        };
        $iterator = new CallbackFilterIterator($this->document, $normalized);
        $iterator = $this->combineHeader($iterator, $header);

        return $this->stripBOM($iterator, $bom);
    }

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

        $header = $this->filterColumnNames($header);
        $header_count = count($header);
        $iterator = new CallbackFilterIterator($iterator, function (array $row, int $offset) {
            return $offset != $this->header_offset;
        });

        return new MapIterator($iterator, function (array $row) use ($header_count, $header) {
            if ($header_count != count($row)) {
                $row = array_slice(array_pad($row, $header_count, null), 0, $header_count);
            }

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

    /**
     * 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);
        return new MapIterator($iterator, function (array $row, $index) use ($bom_length) {
            if (0 != $index) {
                return $row;
            }

            return $this->removeBOM($row, $bom_length, $this->enclosure);
        });
    }

    /**
     * Strip the BOM sequence from a record
     *
     * @param string[] $row
     * @param int      $bom_length
     * @param string   $enclosure
     *
     * @return string[]
     */
    protected function removeBOM(array $row, int $bom_length, string $enclosure): array
    {
        if (0 == $bom_length) {
            return $row;
        }

        $row[0] = mb_substr($row[0], $bom_length);
        if ($enclosure == mb_substr($row[0], 0, 1) && $enclosure == mb_substr($row[0], -1, 1)) {
            $row[0] = mb_substr($row[0], 1, -1);
        }

        return $row;
    }

    /**
     * Returns the column header associate with the RecordSet
     *
     * @throws InvalidArgumentException If no header is found
     *
     * @return string[]
     */
    public function getHeader(): array
    {
        if ($this->is_header_loaded) {
            return $this->header;
        }

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

            return $this->header;
        }

        $this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY);
        $this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
        $this->document->seek($this->header_offset);
        $header = $this->document->current();
        if (empty($header)) {
            throw new InvalidArgumentException('The header record specified by `Reader::setHeaderOffset` does not exist or is empty');
        }

        if (0 !== $this->header_offset) {
            $this->header = $header;

            return $this->header;
        }

        $this->header = $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);

        return $this->header;
    }
}


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 CallbackFilterIterator;
18		use Iterator;
19		use IteratorAggregate;
20		use League\Csv\Exception\InvalidArgumentException;
21		use LimitIterator;
22		use SplFileObject;
23
24		/**
25		* A class to manage extracting and filtering a CSV
26		*
27		* @package League.csv
28		* @since 3.0.0
29		*
30		*/
31		class Reader extends AbstractCsv implements IteratorAggregate
32		{
33		/**
34		* @inheritdoc
35		*/
36		protected $stream_filter_mode = STREAM_FILTER_READ;
37
38		/**
39		* CSV Document header offset
40		*
41		* @var int\|null
42		*/
43		protected $header_offset;
44
45		/**
46		* CSV Document Header record
47		*
48		* @var string[]
49		*/
50		protected $header = [];
51
52		/**
53		* Tell whether the header needs to be re-generated
54		*
55		* @var bool
56		*/
57		protected $is_header_loaded = false;
58
59		/**
60		* Returns the record offset used as header
61		*
62		* If no CSV record is used this method MUST return null
63		*
64		* @return int\|null
65		*/
66	2	public function getHeaderOffset()
67		{
68	2	return $this->header_offset;
69		}
70
71		/**
72		* Selects the record to be used as the CSV header
73		*
74		* Because of the header is represented as an array, to be valid
75		* a header MUST contain only unique string value.
76		*
77		* @param int\|null $offset the header row offset
78		*
79		* @return static
80		*/
81	28	public function setHeaderOffset($offset): self
82		{
83	28	$this->header_offset = null;
84	28	if (null !== $offset) {
85	24	$this->header_offset = $this->filterInteger(
86		$offset,
87	24	0,
88	24	'the header offset index must be a positive integer or 0'
89		);
90		}
91	26	$this->resetDynamicProperties();
92
93	26	return $this;
94		}
95
96		/**
97		* @inheritdoc
98		*/
99	38	protected function resetDynamicProperties()
100		{
101	38	return $this->is_header_loaded = false;
102		}
103
104		/**
105		* Detect Delimiters occurences in the CSV
106		*
107		* Returns a associative array where each key represents
108		* a valid delimiter and each value the number of occurences
109		*
110		* @param string[] $delimiters the delimiters to consider
111		* @param int $nb_rows Detection is made using $nb_rows of the CSV
112		*
113		* @return array
114		*/
115	8	public function fetchDelimitersOccurrence(array $delimiters, int $nb_rows = 1): array
116		{
117	8	$nb_rows = $this->filterInteger($nb_rows, 1, 'The number of rows to consider must be a valid positive integer');
118		$filter_row = function ($row) {
119	6	return is_array($row) && count($row) > 1;
120	6	};
121		$delimiters = array_unique(array_filter($delimiters, function ($value) {
122	6	return 1 == strlen($value);
123	6	}));
124	6	$this->document->setFlags(SplFileObject::READ_CSV);
125	6	$res = [];
126	6	foreach ($delimiters as $delim) {
127	6	$this->document->setCsvControl($delim, $this->enclosure, $this->escape);
128	6	$iterator = new CallbackFilterIterator(new LimitIterator($this->document, 0, $nb_rows), $filter_row);
129	6	$res[$delim] = count(iterator_to_array($iterator, false), COUNT_RECURSIVE);
130		}
131	6	arsort($res, SORT_NUMERIC);
132
133	6	return $res;
134		}
135
136		/**
137		* Returns a collection of selected records
138		*
139		* @param Statement\|null $stmt
140		*
141		* @return RecordSet
142		*/
143	80	public function select(Statement $stmt = null): RecordSet
144		{
145	80	$stmt = $stmt ?? new Statement();
146
147	80	return $stmt->process($this);
148		}
149
150		/**
151		* @inheritdoc
152		*/
153	110	public function getIterator(): Iterator
154		{
155	110	$bom = $this->getInputBOM();
156	110	$header = $this->getHeader();
157	110	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
158	110	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
159		$normalized = function ($row) {
160	98	return is_array($row) && $row != [null];
161	110	};
162	110	$iterator = new CallbackFilterIterator($this->document, $normalized);
163	110	$iterator = $this->combineHeader($iterator, $header);
164
165	110	return $this->stripBOM($iterator, $bom);
166		}
167
168		/**
169		* Add the CSV header if present and valid
170		*
171		* @param Iterator $iterator
172		* @param string[] $header
173		*
174		* @return Iterator
175		*/
176	110	protected function combineHeader(Iterator $iterator, array $header): Iterator
177		{
178	110	if (null === $this->header_offset) {
179	94	return $iterator;
180		}
181
182	18	$header = $this->filterColumnNames($header);
183	18	$header_count = count($header);
184		$iterator = new CallbackFilterIterator($iterator, function (array $row, int $offset) {
185	14	return $offset != $this->header_offset;
186	18	});
187
188		return new MapIterator($iterator, function (array $row) use ($header_count, $header) {
189	14	if ($header_count != count($row)) {
190		$row = array_slice(array_pad($row, $header_count, null), 0, $header_count);
191		}
192
193	14	return array_combine($header, $row);
194	18	});
195		}
196
197		/**
198		* Strip the BOM sequence if present
199		*
200		* @param Iterator $iterator
201		* @param string $bom
202		*
203		* @return Iterator
204		*/
205	110	protected function stripBOM(Iterator $iterator, string $bom): Iterator
206		{
207	110	if ('' === $bom) {
208	90	return $iterator;
209		}
210
211	20	$bom_length = mb_strlen($bom);
212	20	return new MapIterator($iterator, function (array $row, $index) use ($bom_length) {
213	20	if (0 != $index) {
214	16	return $row;
215		}
216
217	14	return $this->removeBOM($row, $bom_length, $this->enclosure);
218	20	});
219		}
220
221		/**
222		* Strip the BOM sequence from a record
223		*
224		* @param string[] $row
225		* @param int $bom_length
226		* @param string $enclosure
227		*
228		* @return string[]
229		*/
230	30	protected function removeBOM(array $row, int $bom_length, string $enclosure): array
231		{
232	30	if (0 == $bom_length) {
233	10	return $row;
234		}
235
236	20	$row[0] = mb_substr($row[0], $bom_length);
237	20	if ($enclosure == mb_substr($row[0], 0, 1) && $enclosure == mb_substr($row[0], -1, 1)) {
238	10	$row[0] = mb_substr($row[0], 1, -1);
239		}
240
241	20	return $row;
242		}
243
244		/**
245		* Returns the column header associate with the RecordSet
246		*
247		* @throws InvalidArgumentException If no header is found
248		*
249		* @return string[]
250		*/
251	116	public function getHeader(): array
252		{
253	116	if ($this->is_header_loaded) {
254	108	return $this->header;
255		}
256
257	116	$this->is_header_loaded = true;
258	116	if (null === $this->header_offset) {
259	98	$this->header = [];
260
261	98	return $this->header;
262		}
263
264	22	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
265	22	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
266	22	$this->document->seek($this->header_offset);
267	22	$header = $this->document->current();
268	22	if (empty($header)) {
269	2	throw new InvalidArgumentException('The header record specified by `Reader::setHeaderOffset` does not exist or is empty');
270		}
271
272	20	if (0 !== $this->header_offset) {
273	4	$this->header = $header;
274
275	4	return $this->header;
276		}
277
278	16	$this->header = $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure);
279
280	16	return $this->header;
281		}
282		}
283

thephpleague / csv

Pull Request — master (#210)

Reader::getFieldIndex() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like