|
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", "\r\n", "\n", '']; |
|
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 $double_enclosure; |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* @var string |
|
75
|
|
|
*/ |
|
76
|
|
|
private $trim_mask; |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* New instance. |
|
80
|
|
|
* |
|
81
|
|
|
* @param SplFileObject|Stream $document |
|
82
|
|
|
*/ |
|
83
|
24 |
|
public function __construct($document) |
|
84
|
|
|
{ |
|
85
|
24 |
|
if (!$document instanceof Stream && !$document instanceof SplFileObject) { |
|
86
|
3 |
|
throw new TypeError(sprintf( |
|
87
|
3 |
|
'Expected a %s or an SplFileObject object, % given', |
|
88
|
3 |
|
Stream::class, |
|
89
|
3 |
|
is_object($document) ? get_class($document) : gettype($document) |
|
90
|
|
|
)); |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
21 |
|
$this->document = $document; |
|
94
|
21 |
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* @inheritdoc |
|
98
|
|
|
* |
|
99
|
|
|
* Converts the stream into a CSV record iterator |
|
100
|
|
|
*/ |
|
101
|
24 |
|
public function getIterator() |
|
102
|
|
|
{ |
|
103
|
|
|
//initialisation |
|
104
|
24 |
|
list($this->delimiter, $this->enclosure, ) = $this->document->getCsvControl(); |
|
105
|
24 |
|
$this->double_enclosure = $this->enclosure.$this->enclosure; |
|
106
|
24 |
|
$this->trim_mask = str_replace([$this->delimiter, $this->enclosure], '', " \t\0\x0B"); |
|
107
|
24 |
|
$this->document->setFlags(0); |
|
108
|
24 |
|
$this->document->rewind(); |
|
109
|
|
|
do { |
|
110
|
24 |
|
yield $this->extractRecord($this->document->fgets()); |
|
111
|
24 |
|
} while ($this->document->valid()); |
|
112
|
24 |
|
} |
|
113
|
|
|
|
|
114
|
|
|
/** |
|
115
|
|
|
* Extract a record from the Stream document. |
|
116
|
|
|
* |
|
117
|
|
|
* The return array is similar as to the returned value of fgetcsv |
|
118
|
|
|
* If this the an empty line the record will be an array with a single value |
|
119
|
|
|
* equals to null otherwise the array contains string data. |
|
120
|
|
|
* |
|
121
|
|
|
* @param string|bool $line |
|
122
|
|
|
*/ |
|
123
|
21 |
|
private function extractRecord($line): array |
|
124
|
|
|
{ |
|
125
|
21 |
|
$record = []; |
|
126
|
|
|
do { |
|
127
|
21 |
|
$method = 'extractField'; |
|
128
|
21 |
|
if (($line[0] ?? '') === $this->enclosure) { |
|
129
|
15 |
|
$method = 'extractFieldEnclosed'; |
|
130
|
|
|
} |
|
131
|
21 |
|
$record[] = $this->$method($line); |
|
132
|
21 |
|
} while (false !== $line); |
|
133
|
|
|
|
|
134
|
21 |
|
return $record; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
/** |
|
138
|
|
|
* Extract field without enclosure as per RFC4180. |
|
139
|
|
|
* |
|
140
|
|
|
* Leading and trailing whitespaces are trimmed because the field |
|
141
|
|
|
* is not enclosed. trailing line-breaks are also removed. |
|
142
|
|
|
* |
|
143
|
|
|
* @param bool|string $line |
|
144
|
|
|
* |
|
145
|
|
|
* @return null|string |
|
146
|
|
|
*/ |
|
147
|
21 |
|
private function extractField(&$line) |
|
148
|
|
|
{ |
|
149
|
21 |
|
if (in_array($line, self::FIELD_BREAKS, true)) { |
|
150
|
3 |
|
$line = false; |
|
151
|
|
|
|
|
152
|
3 |
|
return null; |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
18 |
|
list($content, $line) = explode($this->delimiter, $line, 2) + [1 => false]; |
|
156
|
18 |
|
if (false === $line) { |
|
157
|
12 |
|
return trim(rtrim($content, "\r\n"), $this->trim_mask); |
|
158
|
|
|
} |
|
159
|
|
|
|
|
160
|
18 |
|
return trim($content, $this->trim_mask); |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
|
|
/** |
|
164
|
|
|
* Extract field with enclosure as per RFC4180. |
|
165
|
|
|
* |
|
166
|
|
|
* - Leading and trailing whitespaces are preserved because the field |
|
167
|
|
|
* is enclosed. |
|
168
|
|
|
* - The field content can spread on multiple document lines. |
|
169
|
|
|
* - Double enclosure character muse be replaced by single enclosure character. |
|
170
|
|
|
* - Trailing line break are remove if they are not part of the field content. |
|
171
|
|
|
* - Invalid field do not throw as per fgetcsv behavior. |
|
172
|
|
|
* |
|
173
|
|
|
* @param bool|string $line |
|
174
|
|
|
* |
|
175
|
|
|
* @return null|string |
|
176
|
|
|
*/ |
|
177
|
15 |
|
private function extractFieldEnclosed(&$line) |
|
178
|
|
|
{ |
|
179
|
|
|
//remove the starting enclosure character if present |
|
180
|
15 |
|
if ($line[0] ?? '' === $this->enclosure) { |
|
181
|
15 |
|
$line = substr($line, 1); |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
15 |
|
$content = ''; |
|
185
|
|
|
do { |
|
186
|
15 |
|
list($buffer, $line) = explode($this->enclosure, $line, 2) + [1 => false]; |
|
187
|
15 |
|
$content .= $buffer; |
|
188
|
15 |
|
} while (false === $line && $this->document->valid() && false !== ($line = $this->document->fgets())); |
|
189
|
|
|
|
|
190
|
15 |
|
$content = str_replace($this->double_enclosure, $this->enclosure, $content); |
|
191
|
15 |
|
if (in_array($line, self::FIELD_BREAKS, true)) { |
|
192
|
9 |
|
$line = false; |
|
193
|
|
|
|
|
194
|
9 |
|
return rtrim($content, "\r\n"); |
|
195
|
|
|
} |
|
196
|
|
|
|
|
197
|
12 |
|
$char = $line[0] ?? ''; |
|
198
|
12 |
|
if ($char === $this->delimiter) { |
|
199
|
9 |
|
$line = substr($line, 1); |
|
200
|
|
|
|
|
201
|
9 |
|
return $content; |
|
202
|
|
|
} |
|
203
|
|
|
|
|
204
|
|
|
//handles enclosure as per RFC4180 or malformed CSV like fgetcsv |
|
205
|
6 |
|
return $content.$char.$this->extractFieldEnclosed($line); |
|
206
|
|
|
} |
|
207
|
|
|
} |
|
208
|
|
|
|
thephpleague /
csv
