RFC4180Iterator::extractField() - 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 2018-09-19 11:18 UTC

RFC4180Iterator::extractField() B

↳ Parent: RFC4180Iterator

Complexity

Conditions	7
Paths	3

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	8
CRAP Score	7

Importance

Changes

Metric	Value
cc	7
nc	3
nop	1
dl	0
loc	20
ccs	8
cts	8
cp	1
crap	7
rs	8.6666
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 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
{
    /**
     * @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
     */
    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 {
            $line = $this->document->fgets();
            yield $this->extractRecord($line);
        } while ($this->document->valid());
    }

    /**
     * Extract a record from the Stream document.
     *
     * @param string|bool $line
     */
    private function extractRecord($line): array
    {
        $record = [];
        do {
            $method = ($line[0] ?? '') === $this->enclosure ? 'extractEnclosedField' : 'extractField';
            $record[] = $this->$method($line);
        } while (false !== $line);

        return $record;
    }

    /**
     * Extract field without enclosure.
     *
     * @param bool|string $line
     *
     * @return null|string
     */
    private function extractField(& $line)
    {
        //process the line if it is only a line-break or the empty string
        if ($line === false || $line === "\r" || $line === "\r\n" || $line === "\n" || $line === '') {
            $line = false;

            return null;
        }

        //explode the line on the next delimiter character
        list($content, $line) = explode($this->delimiter, $line, 2) + [1 => false];

        //if this is the end of line remove line breaks
        if (false === $line) {
            $content = rtrim($content, "\r\n");
        }

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

    /**
     * Extract field with enclosure.
     *
     * @param bool|string $line
     *
     * @return null|string
     */
    private function extractEnclosedField(& $line)
    {
        //remove the first enclosure from the line if present to easily use explode
        if ($line[0] ?? '' === $this->enclosure) {
            $line = substr($line, 1);
        }

        //covers multiline fields
        $content = '';
        do {
            //explode the line on the next enclosure character found
            list($buffer, $line) = explode($this->enclosure, $line, 2) + [1 => false];
            $content .= $buffer;
        } while (false === $line && $this->document->valid() && false !== ($line = $this->document->fgets()));

        //format the field content by removing double quoting if present
        $content = str_replace($this->enclosure.$this->enclosure, $this->enclosure, $content);

        //process the line if it is only a line-break or the empty string
        if ($line === false || $line === "\r" || $line === "\r\n" || $line === "\n" || $line === '') {
            $line = false;

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

        //the field data is extracted since we have a delimiter
        if (($line[0] ?? '') === $this->delimiter) {
            $line = substr($line, 1);

            return $content;
        }

        //double quote content found
        if (($line[0] ?? '') === $this->enclosure) {
            $content .= '"'.$this->extractEnclosedField($line);
        }

        return $content;
    }
}


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 is_object;
26		use function rtrim;
27		use function sprintf;
28		use function str_replace;
29		use function substr;
30		use function trim;
31
32		/**
33		* A RFC4180 Compliant Parser in Pure PHP.
34		*
35		* @see https://php.net/manual/en/function.fgetcsv.php
36		* @see https://php.net/manual/en/function.fgets.php
37		* @see https://tools.ietf.org/html/rfc4180
38		* @see http://edoceo.com/utilitas/csv-file-format
39		*
40		* @package League.csv
41		* @since 9.2.0
42		* @author Ignace Nyamagana Butera <[email protected]>
43		* @internal used internally to produce RFC4180 compliant records
44		*/
45		final class RFC4180Iterator implements IteratorAggregate
46		{
47		/**
48		* @var SplFileObject\|Stream
49		*/
50		private $document;
51
52		/**
53		* @var string
54		*/
55		private $delimiter;
56		/**
57		* @var string
58		*/
59		private $enclosure;
60
61		/**
62		* @var string
63		*/
64		private $trim_mask;
65
66		/**
67		* New instance.
68		*
69		* @param SplFileObject\|Stream $document
70		*/
71	21	public function __construct($document)
72		{
73	21	if (!$document instanceof Stream && !$document instanceof SplFileObject) {
74	3	throw new TypeError(sprintf(
75	3	'Expected a %s or an SplFileObject object, % given',
76	3	Stream::class,
77	3	is_object($document) ? get_class($document) : gettype($document)
78		));
79		}
80
81	18	$this->document = $document;
82	18	}
83
84		/**
85		* @inheritdoc
86		*
87		* Converts the stream into a CSV record iterator
88		*/
89	21	public function getIterator()
90		{
91		//initialisation
92	21	list($this->delimiter, $this->enclosure, ) = $this->document->getCsvControl();
93	21	$this->trim_mask = str_replace([$this->delimiter, $this->enclosure], '', " \t\0\x0B");
94	21	$this->document->setFlags(0);
95	21	$this->document->rewind();
96		do {
97	21	$line = $this->document->fgets();
98	21	yield $this->extractRecord($line);
99	21	} while ($this->document->valid());
100	21	}
101
102		/**
103		* Extract a record from the Stream document.
104		*
105		* @param string\|bool $line
106		*/
107	18	private function extractRecord($line): array
108		{
109	18	$record = [];
110		do {
111	18	$method = ($line[0] ?? '') === $this->enclosure ? 'extractEnclosedField' : 'extractField';
112	18	$record[] = $this->$method($line);
113	18	} while (false !== $line);
114
115	18	return $record;
116		}
117
118		/**
119		* Extract field without enclosure.
120		*
121		* @param bool\|string $line
122		*
123		* @return null\|string
124		*/
125	18	private function extractField(& $line)
126		{
127		//process the line if it is only a line-break or the empty string
128	18	if ($line === false \|\| $line === "\r" \|\| $line === "\r\n" \|\| $line === "\n" \|\| $line === '') {
129	3	$line = false;
130
131	3	return null;
132		}
133
134		//explode the line on the next delimiter character
135	15	list($content, $line) = explode($this->delimiter, $line, 2) + [1 => false];
136
137		//if this is the end of line remove line breaks
138	15	if (false === $line) {
139	12	$content = rtrim($content, "\r\n");
140		}
141
142		//remove whitespaces
143	15	return trim($content, $this->trim_mask);
144		}
145
146		/**
147		* Extract field with enclosure.
148		*
149		* @param bool\|string $line
150		*
151		* @return null\|string
152		*/
153	12	private function extractEnclosedField(& $line)
154		{
155		//remove the first enclosure from the line if present to easily use explode
156	12	if ($line[0] ?? '' === $this->enclosure) {
157	12	$line = substr($line, 1);
158		}
159
160		//covers multiline fields
161	12	$content = '';
162		do {
163		//explode the line on the next enclosure character found
164	12	list($buffer, $line) = explode($this->enclosure, $line, 2) + [1 => false];
165	12	$content .= $buffer;
166	12	} while (false === $line && $this->document->valid() && false !== ($line = $this->document->fgets()));
167
168		//format the field content by removing double quoting if present
169	12	$content = str_replace($this->enclosure.$this->enclosure, $this->enclosure, $content);
170
171		//process the line if it is only a line-break or the empty string
172	12	if ($line === false \|\| $line === "\r" \|\| $line === "\r\n" \|\| $line === "\n" \|\| $line === '') {
173	6	$line = false;
174
175	6	return rtrim($content, "\r\n");
176		}
177
178		//the field data is extracted since we have a delimiter
179	9	if (($line[0] ?? '') === $this->delimiter) {
180	9	$line = substr($line, 1);
181
182	9	return $content;
183		}
184
185		//double quote content found
186	3	if (($line[0] ?? '') === $this->enclosure) {
187	3	$content .= '"'.$this->extractEnclosedField($line);
188		}
189
190	3	return $content;
191		}
192		}
193

thephpleague / csv

Pull Request — master (#309)

RFC4180Iterator::extractField() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like