Reader::getHeaderOffset() - Code Metrics - Inspection of "Improve Package codebase" - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( babf40...27ada1 )

by ignace nyamagana

created 2017-03-10 15:42 UTC

Reader::getHeaderOffset() A

↳ Parent: Reader

Complexity

Conditions	1
Paths	1

Size

Total Lines	4
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	2
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	4
ccs	2
cts	2
cp	1
rs	10
c	0
b	0
f	0
cc	1
eloc	2
nc	1
nop	0
crap	1

<?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\RuntimeException;
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;
    }

    /**
     * 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, __METHOD__.': the number of rows to consider must be a valid positive integer');
        $filter_row = function ($row): bool {
            return is_array($row) && count($row) > 1;
        };
        $delimiters = array_unique(array_filter($delimiters, function ($value): bool {
            return 1 == strlen($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): bool {
            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): bool {
            return $offset != $this->header_offset;
        });

        $mapper = function (array $row) use ($header_count, $header): array {
            if ($header_count != count($row)) {
                $row = array_slice(array_pad($row, $header_count, null), 0, $header_count);
            }

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

        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 $row, int $index) use ($bom_length): array {
            if (0 != $index) {
                return $row;
            }

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

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

    /**
     * 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 RuntimeException 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);
        $this->header = $this->document->current();
        if (empty($this->header)) {
            throw new RuntimeException(sprintf('The header record does not exist or is empty at offset: `%s`', $this->header_offset));
        }

        if (0 === $this->header_offset) {
            $this->header = $this->removeBOM($this->header, mb_strlen($this->getInputBOM()), $this->enclosure);
        }

        return $this->header;
    }

    /**
     * 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
    {
        if (null !== $offset) {
            $offset = $this->filterInteger($offset, 0, __METHOD__.': the header offset index must be a positive integer or 0');
        }

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

        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 CallbackFilterIterator;
18		use Iterator;
19		use IteratorAggregate;
20		use League\Csv\Exception\RuntimeException;
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		* Detect Delimiters occurences in the CSV
73		*
74		* Returns a associative array where each key represents
75		* a valid delimiter and each value the number of occurences
76		*
77		* @param string[] $delimiters the delimiters to consider
78		* @param int $nb_rows Detection is made using $nb_rows of the CSV
79		*
80		* @return array
81		*/
82	8	public function fetchDelimitersOccurrence(array $delimiters, int $nb_rows = 1): array
83		{
84	8	$nb_rows = $this->filterInteger($nb_rows, 1, __METHOD__.': the number of rows to consider must be a valid positive integer');
85		$filter_row = function ($row): bool {
86	6	return is_array($row) && count($row) > 1;
87	3	};
88		$delimiters = array_unique(array_filter($delimiters, function ($value): bool {
89	6	return 1 == strlen($value);
90	6	}));
91	6	$this->document->setFlags(SplFileObject::READ_CSV);
92	6	$res = [];
93	6	foreach ($delimiters as $delim) {
94	6	$this->document->setCsvControl($delim, $this->enclosure, $this->escape);
95	6	$iterator = new CallbackFilterIterator(new LimitIterator($this->document, 0, $nb_rows), $filter_row);
96	6	$res[$delim] = count(iterator_to_array($iterator, false), COUNT_RECURSIVE);
97		}
98	6	arsort($res, SORT_NUMERIC);
99
100	6	return $res;
101		}
102
103		/**
104		* Returns a collection of selected records
105		*
106		* @param Statement\|null $stmt
107		*
108		* @return RecordSet
109		*/
110	80	public function select(Statement $stmt = null): RecordSet
111		{
112	80	$stmt = $stmt ?? new Statement();
113
114	80	return $stmt->process($this);
115		}
116
117		/**
118		* @inheritdoc
119		*/
120	114	public function getIterator(): Iterator
121		{
122	114	$bom = $this->getInputBOM();
123	114	$header = $this->getHeader();
124	114	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
125	114	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
126		$normalized = function ($row): bool {
127	98	return is_array($row) && $row != [null];
128	57	};
129	114	$iterator = new CallbackFilterIterator($this->document, $normalized);
130	114	$iterator = $this->combineHeader($iterator, $header);
131
132	114	return $this->stripBOM($iterator, $bom);
133		}
134
135		/**
136		* Add the CSV header if present and valid
137		*
138		* @param Iterator $iterator
139		* @param string[] $header
140		*
141		* @return Iterator
142		*/
143	114	protected function combineHeader(Iterator $iterator, array $header): Iterator
144		{
145	114	if (null === $this->header_offset) {
146	98	return $iterator;
147		}
148
149	18	$header = $this->filterColumnNames($header);
150	18	$header_count = count($header);
151		$iterator = new CallbackFilterIterator($iterator, function (array $row, int $offset): bool {
152	14	return $offset != $this->header_offset;
153	18	});
154
155		$mapper = function (array $row) use ($header_count, $header): array {
156	14	if ($header_count != count($row)) {
157		$row = array_slice(array_pad($row, $header_count, null), 0, $header_count);
158		}
159
160	14	return array_combine($header, $row);
161	18	};
162
163	18	return new MapIterator($iterator, $mapper);
164		}
165
166		/**
167		* Strip the BOM sequence if present
168		*
169		* @param Iterator $iterator
170		* @param string $bom
171		*
172		* @return Iterator
173		*/
174	114	protected function stripBOM(Iterator $iterator, string $bom): Iterator
175		{
176	114	if ('' === $bom) {
177	94	return $iterator;
178		}
179
180	20	$bom_length = mb_strlen($bom);
181	20	$mapper = function (array $row, int $index) use ($bom_length): array {
182	20	if (0 != $index) {
183	16	return $row;
184		}
185
186	14	return $this->removeBOM($row, $bom_length, $this->enclosure);
187	20	};
188
189	20	return new MapIterator($iterator, $mapper);
190		}
191
192		/**
193		* Strip the BOM sequence from a record
194		*
195		* @param string[] $row
196		* @param int $bom_length
197		* @param string $enclosure
198		*
199		* @return string[]
200		*/
201	30	protected function removeBOM(array $row, int $bom_length, string $enclosure): array
202		{
203	30	if (0 == $bom_length) {
204	10	return $row;
205		}
206
207	20	$row[0] = mb_substr($row[0], $bom_length);
208	20	if ($enclosure == mb_substr($row[0], 0, 1) && $enclosure == mb_substr($row[0], -1, 1)) {
209	10	$row[0] = mb_substr($row[0], 1, -1);
210		}
211
212	20	return $row;
213		}
214
215		/**
216		* Returns the column header associate with the RecordSet
217		*
218		* @throws RuntimeException If no header is found
219		*
220		* @return string[]
221		*/
222	120	public function getHeader(): array
223		{
224	120	if ($this->is_header_loaded) {
225	112	return $this->header;
226		}
227
228	120	$this->is_header_loaded = true;
229	120	if (null === $this->header_offset) {
230	102	$this->header = [];
231
232	102	return $this->header;
233		}
234
235	22	$this->document->setFlags(SplFileObject::READ_CSV \| SplFileObject::READ_AHEAD \| SplFileObject::SKIP_EMPTY);
236	22	$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape);
237	22	$this->document->seek($this->header_offset);
238	22	$this->header = $this->document->current();
239	22	if (empty($this->header)) {
240	2	throw new RuntimeException(sprintf('The header record does not exist or is empty at offset: `%s`', $this->header_offset));
241		}
242
243	20	if (0 === $this->header_offset) {
244	16	$this->header = $this->removeBOM($this->header, mb_strlen($this->getInputBOM()), $this->enclosure);
245		}
246
247	20	return $this->header;
248		}
249
250		/**
251		* Selects the record to be used as the CSV header
252		*
253		* Because of the header is represented as an array, to be valid
254		* a header MUST contain only unique string value.
255		*
256		* @param int\|null $offset the header row offset
257		*
258		* @return static
259		*/
260	26	public function setHeaderOffset($offset): self
261		{
262	26	if (null !== $offset) {
263	22	$offset = $this->filterInteger($offset, 0, __METHOD__.': the header offset index must be a positive integer or 0');
264		}
265
266	26	if ($offset !== $this->header_offset) {
267	22	$this->header_offset = $offset;
268	22	$this->resetProperties();
269		}
270
271	26	return $this;
272		}
273
274		/**
275		* @inheritdoc
276		*/
277	34	protected function resetProperties()
278		{
279	34	return $this->is_header_loaded = false;
280		}
281		}
282

thephpleague / csv

Push — master ( babf40...27ada1 )

Reader::getHeaderOffset() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like