EmptyEscapeParser::extractEnclosedFieldContent() - Code Metrics - Inspection of "Merge branch 'master' of https://github.com/thephp..." - thephpleague/csv - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( e48390...c23027 )

by ignace nyamagana

created 2018-09-27 06:45 UTC

EmptyEscapeParser::extractEnclosedFieldContent() B

↳ Parent: EmptyEscapeParser

Complexity

Conditions	7
Paths	24

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	21
CRAP Score	7

Importance

Changes

Metric	Value
cc	7
nc	24
nop	0
dl	0
loc	36
ccs	21
cts	21
cp	1
crap	7
rs	8.4106
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 9.2.0
 * @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\Polyfill;

use Generator;
use League\Csv\Stream;
use SplFileObject;
use TypeError;
use function explode;
use function get_class;
use function gettype;
use function in_array;
use function is_object;
use function ltrim;
use function rtrim;
use function sprintf;
use function str_replace;
use function substr;

/**
 * A Polyfill to PHP's SplFileObject behavior when reading a CSV document
 * with the SplFileObject::READ_CSV and SplFileObject::SKIP_EMPTY flags on
 * and the empty string as the escape parameter.
 *
 * @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
 *
 * @internal used internally to parse a CSV document without using the escape character
 */
final class EmptyEscapeParser
{
    /**
     * @internal
     */
    const FIELD_BREAKS = [false, '', "\r\n", "\n", "\r"];

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

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

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

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

    /**
     * @var string|bool
     */
    private static $line;

    /**
     * Converts the document into a CSV record iterator.
     *
     * The returned record array is similar to the returned value of fgetcsv
     *
     * - If the line is empty the record is skipped
     * - Otherwise the array contains strings.
     *
     * @param SplFileObject|Stream $document
     */
    public static function parse($document): Generator
    {
        self::$document = self::filterDocument($document);
        list(self::$delimiter, self::$enclosure, ) = self::$document->getCsvControl();
        self::$trim_mask = str_replace([self::$delimiter, self::$enclosure], '', " \t\0\x0B");
        self::$document->setFlags(0);
        self::$document->rewind();
        while (self::$document->valid()) {
            $record = [];
            self::$line = self::$document->fgets();
            do {
                $method = 'extractFieldContent';
                $buffer = ltrim(self::$line, self::$trim_mask);
                if (($buffer[0] ?? '') === self::$enclosure) {
                    $method = 'extractEnclosedFieldContent';
                    self::$line = $buffer;
                }

                $record[] = self::$method();
            } while (false !== self::$line);

            if ([null] !== $record) {
                yield $record;
            }
        }
    }

    /**
     * Filter the submitted document.
     *
     * @param SplFileObject|Stream $document
     *
     * @return SplFileObject|Stream
     */
    private static function filterDocument($document)
    {
        if ($document instanceof Stream || $document instanceof SplFileObject) {
            return $document;
        }

        throw new TypeError(sprintf(
            'Expected a %s or an SplFileObject object, %s given',
            Stream::class,
            is_object($document) ? get_class($document) : gettype($document)
        ));
    }

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

            return null;
        }

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

        return $content;
    }

    /**
     * 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 fields content are treated as per fgetcsv behavior.
     */
    private static function extractEnclosedFieldContent(): string
    {
        if ((self::$line[0] ?? '') === self::$enclosure) {
            self::$line = substr(self::$line, 1);
        }

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

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

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

        $char = self::$line[0] ?? '';
        if (self::$delimiter === $char) {
            self::$line = substr(self::$line, 1);

            return $content;
        }

        if (self::$enclosure === $char) {
            return $content.self::$enclosure.self::extractEnclosedFieldContent();
        }

        return $content.self::extractFieldContent();
    }
}


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 9.2.0
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\Polyfill;
18
19		use Generator;
20		use League\Csv\Stream;
21		use SplFileObject;
22		use TypeError;
23		use function explode;
24		use function get_class;
25		use function gettype;
26		use function in_array;
27		use function is_object;
28		use function ltrim;
29		use function rtrim;
30		use function sprintf;
31		use function str_replace;
32		use function substr;
33
34		/**
35		* A Polyfill to PHP's SplFileObject behavior when reading a CSV document
36		* with the SplFileObject::READ_CSV and SplFileObject::SKIP_EMPTY flags on
37		* and the empty string as the escape parameter.
38		*
39		* @see https://php.net/manual/en/function.fgetcsv.php
40		* @see https://php.net/manual/en/function.fgets.php
41		* @see https://tools.ietf.org/html/rfc4180
42		* @see http://edoceo.com/utilitas/csv-file-format
43		*
44		* @internal used internally to parse a CSV document without using the escape character
45		*/
46		final class EmptyEscapeParser
47		{
48		/**
49		* @internal
50		*/
51		const FIELD_BREAKS = [false, '', "\r\n", "\n", "\r"];
52
53		/**
54		* @var SplFileObject\|Stream
55		*/
56		private static $document;
57
58		/**
59		* @var string
60		*/
61		private static $delimiter;
62
63		/**
64		* @var string
65		*/
66		private static $enclosure;
67
68		/**
69		* @var string
70		*/
71		private static $trim_mask;
72
73		/**
74		* @var string\|bool
75		*/
76		private static $line;
77
78		/**
79		* Converts the document into a CSV record iterator.
80		*
81		* The returned record array is similar to the returned value of fgetcsv
82		*
83		* - If the line is empty the record is skipped
84		* - Otherwise the array contains strings.
85		*
86		* @param SplFileObject\|Stream $document
87		*/
88	36	public static function parse($document): Generator
89		{
90	36	self::$document = self::filterDocument($document);
91	33	list(self::$delimiter, self::$enclosure, ) = self::$document->getCsvControl();
92	33	self::$trim_mask = str_replace([self::$delimiter, self::$enclosure], '', " \t\0\x0B");
93	33	self::$document->setFlags(0);
94	33	self::$document->rewind();
95	33	while (self::$document->valid()) {
96	33	$record = [];
97	33	self::$line = self::$document->fgets();
98		do {
99	33	$method = 'extractFieldContent';
100	33	$buffer = ltrim(self::$line, self::$trim_mask);
101	33	if (($buffer[0] ?? '') === self::$enclosure) {
102	24	$method = 'extractEnclosedFieldContent';
103	24	self::$line = $buffer;
104		}
105
106	33	$record[] = self::$method();
107	33	} while (false !== self::$line);
108
109	33	if ([null] !== $record) {
110	33	yield $record;
111		}
112		}
113	33	}
114
115		/**
116		* Filter the submitted document.
117		*
118		* @param SplFileObject\|Stream $document
119		*
120		* @return SplFileObject\|Stream
121		*/
122	9	private static function filterDocument($document)
123		{
124	9	if ($document instanceof Stream \|\| $document instanceof SplFileObject) {
125	6	return $document;
126		}
127
128	3	throw new TypeError(sprintf(
129	3	'Expected a %s or an SplFileObject object, %s given',
130	3	Stream::class,
131	3	is_object($document) ? get_class($document) : gettype($document)
132		));
133		}
134
135		/**
136		* Extract field without enclosure as per RFC4180.
137		*
138		* - Leading and trailing whitespaces must be removed.
139		* - trailing line-breaks must be removed.
140		*
141		* @return null\|string
142		*/
143	33	private static function extractFieldContent()
144		{
145	33	if (in_array(self::$line, self::FIELD_BREAKS, true)) {
146	3	self::$line = false;
147
148	3	return null;
149		}
150
151	30	list($content, self::$line) = explode(self::$delimiter, self::$line, 2) + [1 => false];
152	30	if (false === self::$line) {
153	24	return rtrim($content, "\r\n");
154		}
155
156	30	return $content;
157		}
158
159		/**
160		* Extract field with enclosure as per RFC4180.
161		*
162		* - Field content can spread on multiple document lines.
163		* - Content inside enclosure must be preserved.
164		* - Double enclosure sequence must be replaced by single enclosure character.
165		* - Trailing line break must be removed if they are not part of the field content.
166		* - Invalid fields content are treated as per fgetcsv behavior.
167		*/
168	24	private static function extractEnclosedFieldContent(): string
169		{
170	24	if ((self::$line[0] ?? '') === self::$enclosure) {
171	24	self::$line = substr(self::$line, 1);
172		}
173
174	24	$content = '';
175	24	while (false !== self::$line) {
176	24	list($buffer, $remainder) = explode(self::$enclosure, self::$line, 2) + [1 => false];
177	24	$content .= $buffer;
178	24	if (false !== $remainder) {
179	21	self::$line = $remainder;
180	21	break;
181		}
182	9	self::$line = self::$document->fgets();
183		}
184
185	24	if (in_array(self::$line, self::FIELD_BREAKS, true)) {
186	12	self::$line = false;
187
188	12	return rtrim($content, "\r\n");
189		}
190
191	21	$char = self::$line[0] ?? '';
192	21	if (self::$delimiter === $char) {
193	15	self::$line = substr(self::$line, 1);
194
195	15	return $content;
196		}
197
198	18	if (self::$enclosure === $char) {
199	9	return $content.self::$enclosure.self::extractEnclosedFieldContent();
200		}
201
202	9	return $content.self::extractFieldContent();
203		}
204		}
205

thephpleague / csv

Push — master ( e48390...c23027 )

EmptyEscapeParser::extractEnclosedFieldContent() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like