RFC4180Iterator::extractFieldEnclosed() - Code Metrics - Inspection of "Adding support for the escape char with an empty s..." - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#309)

by ignace nyamagana

created 2018-09-21 10:41 UTC

RFC4180Iterator::extractFieldEnclosed() B

↳ Parent: RFC4180Iterator

Complexity

Conditions	7
Paths	24

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	20
CRAP Score	7

Importance

Changes

Metric	Value
cc	7
nc	24
nop	1
dl	0
loc	39
ccs	20
cts	20
cp	1
crap	7
rs	8.3626
c	0
b	0
f	0

<?php

/**
 * League.Csv (https://csv.thephpleague.com).
 *
 * @author  Ignace Nyamagana Butera <[email protected]>
 * @license https://github.com/thephpleague/csv/blob/master/LICENSE (MIT License)
 * @version 9.1.5
 * @link    https://github.com/thephpleague/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 IteratorAggregate;
use SplFileObject;
use TypeError;
use function explode;
use function get_class;
use function gettype;
use function in_array;
use function is_object;
use function rtrim;
use function sprintf;
use function str_replace;
use function substr;
use function trim;

/**
 * A RFC4180 Compliant Parser in Pure PHP.
 *
 * @see https://php.net/manual/en/function.fgetcsv.php
 * @see https://php.net/manual/en/function.fgets.php
 * @see https://tools.ietf.org/html/rfc4180
 * @see http://edoceo.com/utilitas/csv-file-format
 *
 * @package League.csv
 * @since   9.2.0
 * @author  Ignace Nyamagana Butera <[email protected]>
 * @internal used internally to produce RFC4180 compliant records
 */
final class RFC4180Iterator implements IteratorAggregate
{
    /**
     * @internal
     */
    const FIELD_BREAKS = [false, '', "\r\n", "\n", "\r"];

    /**
     * @var SplFileObject|Stream
     */
    private $document;

    /**
     * @var string
     */
    private $delimiter;

    /**
     * @var string
     */
    private $enclosure;

    /**
     * @var string
     */
    private $trim_mask;

    /**
     * New instance.
     *
     * @param SplFileObject|Stream $document
     */
    public function __construct($document)
    {
        if (!$document instanceof Stream && !$document instanceof SplFileObject) {
            throw new TypeError(sprintf(
                'Expected a %s or an SplFileObject object, % given',
                Stream::class,
                is_object($document) ? get_class($document) : gettype($document)
            ));
        }

        $this->document = $document;
    }

    /**
     * @inheritdoc
     *
     * Converts the stream into a CSV record iterator by extracting records one by one
     *
     * The returned record array is similar to the returned value of fgetcsv
     *
     * - If the line is empty the record will be an array with a single value equals to null
     * - Otherwise the array contains strings.
     */
    public function getIterator()
    {
        //initialisation
        list($this->delimiter, $this->enclosure, ) = $this->document->getCsvControl();
        $this->trim_mask = str_replace([$this->delimiter, $this->enclosure], '', " \t\0\x0B");
        $this->document->setFlags(0);
        $this->document->rewind();
        do {
            $record = [];
            $line = $this->document->fgets();
            do {
                $method = 'extractField';
                if (($line[0] ?? '') === $this->enclosure) {
                    $method = 'extractFieldEnclosed';
                }
                $record[] = $this->$method($line);
            } while (false !== $line);

            yield $record;
        } while ($this->document->valid());
    }

    /**
     * Extract field without enclosure as per RFC4180.
     *
     * - Leading and trailing whitespaces must be removed.
     * - trailing line-breaks must be removed.
     *
     * @param bool|string $line
     *
     * @return null|string
     */
    private function extractField(&$line)
    {
        if (in_array($line, self::FIELD_BREAKS, true)) {
            $line = false;

            return null;
        }

        list($content, $line) = explode($this->delimiter, $line, 2) + [1 => false];
        if (false === $line) {
            return trim(rtrim($content, "\r\n"), $this->trim_mask);
        }

        return trim($content, $this->trim_mask);
    }

    /**
     * Extract field with enclosure as per RFC4180.
     *
     * - Field content can spread on multiple document lines.
     * - Content inside enclosure must be preserved.
     * - Double enclosure sequence must be replaced by single enclosure character.
     * - Trailing line break must be removed if they are not part of the field content.
     * - Invalid field do not throw as per fgetcsv behavior.
     *
     * @param bool|string $line
     *
     * @return null|string
     */
    private function extractFieldEnclosed(&$line)
    {
        //remove the starting enclosure character if present
        if (($line[0] ?? '') === $this->enclosure) {
            $line = substr($line, 1);
        }

        $content = '';
        while (false !== $line) {
            list($buffer, $line) = explode($this->enclosure, $line, 2) + [1 => false];
            $content .= $buffer;
            if (false !== $line) {
                break;
            }
            $line = $this->document->fgets();
        }

        if (in_array($line, self::FIELD_BREAKS, true)) {
            $line = false;

            return rtrim($content, "\r\n");
        }

        $char = $line[0] ?? '';
        //handles end of content by delimiter
        if ($char === $this->delimiter) {
            $line = substr($line, 1);

            return $content;
        }

        //handles double quoted data
        if ($char === $this->enclosure) {
            return $content.$char.$this->extractFieldEnclosed($line);
        }

        //handles malformed CSV like fgetcsv
        return $content.$this->extractFieldEnclosed($line);
    }
}


1		<?php
2
3		/**
4		* League.Csv (https://csv.thephpleague.com).
5		*
6		* @author Ignace Nyamagana Butera <[email protected]>
7		* @license https://github.com/thephpleague/csv/blob/master/LICENSE (MIT License)
8		* @version 9.1.5
9		* @link https://github.com/thephpleague/csv
10		*
11		* For the full copyright and license information, please view the LICENSE
12		* file that was distributed with this source code.
13		*/
14
15		declare(strict_types=1);
16
17		namespace League\Csv;
18
19		use IteratorAggregate;
20		use SplFileObject;
21		use TypeError;
22		use function explode;
23		use function get_class;
24		use function gettype;
25		use function in_array;
26		use function is_object;
27		use function rtrim;
28		use function sprintf;
29		use function str_replace;
30		use function substr;
31		use function trim;
32
33		/**
34		* A RFC4180 Compliant Parser in Pure PHP.
35		*
36		* @see https://php.net/manual/en/function.fgetcsv.php
37		* @see https://php.net/manual/en/function.fgets.php
38		* @see https://tools.ietf.org/html/rfc4180
39		* @see http://edoceo.com/utilitas/csv-file-format
40		*
41		* @package League.csv
42		* @since 9.2.0
43		* @author Ignace Nyamagana Butera <[email protected]>
44		* @internal used internally to produce RFC4180 compliant records
45		*/
46		final class RFC4180Iterator implements IteratorAggregate
47		{
48		/**
49		* @internal
50		*/
51		const FIELD_BREAKS = [false, '', "\r\n", "\n", "\r"];
52
53		/**
54		* @var SplFileObject\|Stream
55		*/
56		private $document;
57
58		/**
59		* @var string
60		*/
61		private $delimiter;
62
63		/**
64		* @var string
65		*/
66		private $enclosure;
67
68		/**
69		* @var string
70		*/
71		private $trim_mask;
72
73		/**
74		* New instance.
75		*
76		* @param SplFileObject\|Stream $document
77		*/
78	6	public function __construct($document)
79		{
80	6	if (!$document instanceof Stream && !$document instanceof SplFileObject) {
81	3	throw new TypeError(sprintf(
82	3	'Expected a %s or an SplFileObject object, % given',
83	3	Stream::class,
84	3	is_object($document) ? get_class($document) : gettype($document)
85		));
86		}
87
88	3	$this->document = $document;
89	3	}
90
91		/**
92		* @inheritdoc
93		*
94		* Converts the stream into a CSV record iterator by extracting records one by one
95		*
96		* The returned record array is similar to the returned value of fgetcsv
97		*
98		* - If the line is empty the record will be an array with a single value equals to null
99		* - Otherwise the array contains strings.
100		*/
101	27	public function getIterator()
102		{
103		//initialisation
104	27	list($this->delimiter, $this->enclosure, ) = $this->document->getCsvControl();
105	27	$this->trim_mask = str_replace([$this->delimiter, $this->enclosure], '', " \t\0\x0B");
106	27	$this->document->setFlags(0);
107	27	$this->document->rewind();
108		do {
109	27	$record = [];
110	27	$line = $this->document->fgets();
111		do {
112	27	$method = 'extractField';
113	27	if (($line[0] ?? '') === $this->enclosure) {
114	21	$method = 'extractFieldEnclosed';
115		}
116	27	$record[] = $this->$method($line);
117	27	} while (false !== $line);
118
119	27	yield $record;
120	27	} while ($this->document->valid());
121	27	}
122
123		/**
124		* Extract field without enclosure as per RFC4180.
125		*
126		* - Leading and trailing whitespaces must be removed.
127		* - trailing line-breaks must be removed.
128		*
129		* @param bool\|string $line
130		*
131		* @return null\|string
132		*/
133	27	private function extractField(&$line)
134		{
135	27	if (in_array($line, self::FIELD_BREAKS, true)) {
136	3	$line = false;
137
138	3	return null;
139		}
140
141	24	list($content, $line) = explode($this->delimiter, $line, 2) + [1 => false];
142	24	if (false === $line) {
143	18	return trim(rtrim($content, "\r\n"), $this->trim_mask);
144		}
145
146	24	return trim($content, $this->trim_mask);
147		}
148
149		/**
150		* Extract field with enclosure as per RFC4180.
151		*
152		* - Field content can spread on multiple document lines.
153		* - Content inside enclosure must be preserved.
154		* - Double enclosure sequence must be replaced by single enclosure character.
155		* - Trailing line break must be removed if they are not part of the field content.
156		* - Invalid field do not throw as per fgetcsv behavior.
157		*
158		* @param bool\|string $line
159		*
160		* @return null\|string
161		*/
162	21	private function extractFieldEnclosed(&$line)
163		{
164		//remove the starting enclosure character if present
165	21	if (($line[0] ?? '') === $this->enclosure) {
166	21	$line = substr($line, 1);
167		}
168
169	21	$content = '';
170	21	while (false !== $line) {
171	21	list($buffer, $line) = explode($this->enclosure, $line, 2) + [1 => false];
172	21	$content .= $buffer;
173	21	if (false !== $line) {
174	18	break;
175		}
176	12	$line = $this->document->fgets();
177		}
178
179	21	if (in_array($line, self::FIELD_BREAKS, true)) {
180	9	$line = false;
181
182	9	return rtrim($content, "\r\n");
183		}
184
185	18	$char = $line[0] ?? '';
186		//handles end of content by delimiter
187	18	if ($char === $this->delimiter) {
188	15	$line = substr($line, 1);
189
190	15	return $content;
191		}
192
193		//handles double quoted data
194	12	if ($char === $this->enclosure) {
195	9	return $content.$char.$this->extractFieldEnclosed($line);
196		}
197
198		//handles malformed CSV like fgetcsv
199	3	return $content.$this->extractFieldEnclosed($line);
200		}
201		}
202

thephpleague / csv

Pull Request — master (#309)

RFC4180Iterator::extractFieldEnclosed() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like