|
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 get_class; |
|
23
|
|
|
use function gettype; |
|
24
|
|
|
use function is_object; |
|
25
|
|
|
use function sprintf; |
|
26
|
|
|
use function substr; |
|
27
|
|
|
use function trim; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* A RFC4180 Compliant Parser in Pure PHP. |
|
31
|
|
|
* |
|
32
|
|
|
* @see https://php.net/manual/en/function.fgetcsv.php |
|
33
|
|
|
* @see https://php.net/manual/en/function.fgetc.php |
|
34
|
|
|
* @see https://tools.ietf.org/html/rfc4180 |
|
35
|
|
|
* @see http://edoceo.com/utilitas/csv-file-format |
|
36
|
|
|
* |
|
37
|
|
|
* @package League.csv |
|
38
|
|
|
* @since 9.2.0 |
|
39
|
|
|
* @author Ignace Nyamagana Butera <[email protected]> |
|
40
|
|
|
* @internal used internally to produce RFC4180 compliant records |
|
41
|
|
|
*/ |
|
42
|
|
|
final class RFC4180Iterator implements IteratorAggregate |
|
43
|
|
|
{ |
|
44
|
|
|
/** |
|
45
|
|
|
* The CSV document. |
|
46
|
|
|
* |
|
47
|
|
|
* @var SplFileObject|Stream |
|
48
|
|
|
*/ |
|
49
|
|
|
private $document; |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* New instance. |
|
53
|
|
|
* |
|
54
|
|
|
* @param SplFileObject|Stream $document |
|
55
|
|
|
*/ |
|
56
|
9 |
|
public function __construct($document) |
|
57
|
|
|
{ |
|
58
|
9 |
|
if (!$document instanceof Stream && !$document instanceof SplFileObject) { |
|
59
|
3 |
|
throw new TypeError(sprintf( |
|
60
|
3 |
|
'Expected a %s or an SplFileObject object, % given', |
|
61
|
3 |
|
Stream::class, |
|
62
|
3 |
|
is_object($document) ? get_class($document) : gettype($document) |
|
63
|
|
|
)); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
6 |
|
$this->document = $document; |
|
67
|
6 |
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* @inheritdoc |
|
71
|
|
|
* |
|
72
|
|
|
* Converts the stream into a CSV record iterator |
|
73
|
|
|
*/ |
|
74
|
12 |
|
public function getIterator() |
|
75
|
|
|
{ |
|
76
|
|
|
//initialisation |
|
77
|
12 |
|
$record = []; |
|
78
|
12 |
|
$buffer = null; |
|
79
|
12 |
|
$previous_char = ''; |
|
80
|
12 |
|
$enclosed_field = false; |
|
81
|
12 |
|
list($delimiter, $enclosure, ) = $this->document->getCsvControl(); |
|
82
|
12 |
|
$trim_mask = str_replace([$delimiter, $enclosure], '', " \t\0\x0B"); |
|
83
|
12 |
|
$this->document->rewind(); |
|
84
|
|
|
|
|
85
|
|
|
//let's walk through the stream char by char |
|
86
|
12 |
|
while (false !== ($char = $this->document->fgetc())) { |
|
87
|
|
|
switch ($char) { |
|
88
|
12 |
|
case $enclosure: |
|
89
|
12 |
|
if (!$enclosed_field) { |
|
90
|
|
|
//the enclosure is at the start of the record |
|
91
|
|
|
//so we have an enclosed field |
|
92
|
12 |
|
if (null === $buffer) { |
|
93
|
12 |
|
$enclosed_field = true; |
|
94
|
12 |
|
break; |
|
95
|
|
|
} |
|
96
|
|
|
//invalid CSV content let's deal with it like fgetcsv |
|
97
|
|
|
//we add the character to the buffer and we move on |
|
98
|
3 |
|
$previous_char = $char; |
|
99
|
3 |
|
$buffer .= $char; |
|
100
|
3 |
|
break; |
|
101
|
|
|
} |
|
102
|
|
|
//double quoted enclosure let's skip the character and move on |
|
103
|
12 |
|
if ($previous_char === $enclosure) { |
|
104
|
|
|
//we reset the previous character |
|
105
|
|
|
//to avoid stripping to much enclosure character |
|
106
|
3 |
|
$previous_char = ''; |
|
107
|
3 |
|
break; |
|
108
|
|
|
} |
|
109
|
12 |
|
$previous_char = $char; |
|
110
|
12 |
|
$buffer .= $char; |
|
111
|
12 |
|
break; |
|
112
|
12 |
|
case $delimiter: |
|
113
|
12 |
|
if ($enclosed_field) { |
|
114
|
|
|
//the delimiter is enclosed let's add it to the buffer and move on |
|
115
|
12 |
|
if ($previous_char !== $enclosure) { |
|
116
|
9 |
|
$buffer .= $char; |
|
117
|
9 |
|
$previous_char = $char; |
|
118
|
9 |
|
break; |
|
119
|
|
|
} |
|
120
|
|
|
//strip the enclosure character present at the |
|
121
|
|
|
//end of the buffer; this is the end of en enclosed field |
|
122
|
12 |
|
$buffer = substr($buffer, 0, -1); |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
//the buffer is the field content we add it to the record |
|
126
|
|
|
//and convert it into a string |
|
127
|
12 |
|
$buffer = ''.$buffer; |
|
128
|
|
|
//if the field is not enclose we trim white spaces |
|
129
|
12 |
|
if (!$enclosed_field) { |
|
130
|
9 |
|
$buffer = trim($buffer, $trim_mask); |
|
131
|
|
|
} |
|
132
|
12 |
|
$record[] = $buffer; |
|
133
|
|
|
|
|
134
|
|
|
//reset field parameters |
|
135
|
12 |
|
$buffer = null; |
|
136
|
12 |
|
$previous_char = ''; |
|
137
|
12 |
|
$enclosed_field = false; |
|
138
|
12 |
|
break; |
|
139
|
12 |
|
case "\n": |
|
140
|
12 |
|
case "\r": |
|
141
|
12 |
|
if ($enclosed_field) { |
|
142
|
|
|
//the line break is enclosed let's add it to the buffer and move on |
|
143
|
6 |
|
if ($previous_char !== $enclosure) { |
|
144
|
3 |
|
$previous_char = $char; |
|
145
|
3 |
|
$buffer .= $char; |
|
146
|
3 |
|
break; |
|
147
|
|
|
} |
|
148
|
|
|
//strip the enclosure character present at the |
|
149
|
|
|
//end of the buffer; this is the end of a record |
|
150
|
3 |
|
$buffer = substr($buffer, 0, -1); |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
//if the field is not enclose we trim white spaces |
|
154
|
12 |
|
if (null !== $buffer && !$enclosed_field) { |
|
155
|
9 |
|
$buffer = trim($buffer, $trim_mask); |
|
156
|
|
|
} |
|
157
|
|
|
//adding field content to the record |
|
158
|
12 |
|
$record[] = $buffer; |
|
159
|
|
|
//reset field parameters |
|
160
|
12 |
|
$buffer = null; |
|
161
|
12 |
|
$enclosed_field = false; |
|
162
|
12 |
|
$previous_char = ''; |
|
163
|
|
|
|
|
164
|
|
|
//yield the record |
|
165
|
12 |
|
yield $record; |
|
166
|
|
|
|
|
167
|
|
|
//reset record |
|
168
|
12 |
|
$record = []; |
|
169
|
12 |
|
break; |
|
170
|
|
|
default: |
|
171
|
12 |
|
$previous_char = $char; |
|
172
|
12 |
|
$buffer .= $char; |
|
173
|
12 |
|
break; |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
|
|
//yield the remaining buffer |
|
178
|
12 |
|
if ($enclosed_field && $enclosure === substr($buffer, -1, 1)) { |
|
179
|
|
|
//strip the enclosure character present at the |
|
180
|
|
|
//end of the buffer; this is the end of en enclosed field |
|
181
|
3 |
|
$buffer = substr($buffer, 0, -1); |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
//if the field is not enclose we trim white spaces |
|
185
|
12 |
|
if (!$enclosed_field && null !== $buffer) { |
|
186
|
9 |
|
$buffer = trim($buffer, $trim_mask); |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
12 |
|
$record[] = $buffer; |
|
190
|
|
|
|
|
191
|
12 |
|
yield $record; |
|
192
|
12 |
|
} |
|
193
|
|
|
} |
|
194
|
|
|
|
thephpleague /
csv
