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\Exception; |
21
|
|
|
use League\Csv\Stream; |
22
|
|
|
use SplFileObject; |
23
|
|
|
use TypeError; |
24
|
|
|
use function explode; |
25
|
|
|
use function get_class; |
26
|
|
|
use function gettype; |
27
|
|
|
use function in_array; |
28
|
|
|
use function is_object; |
29
|
|
|
use function ltrim; |
30
|
|
|
use function rtrim; |
31
|
|
|
use function sprintf; |
32
|
|
|
use function str_replace; |
33
|
|
|
use function strlen; |
34
|
|
|
use function substr; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* A Polyfill to PHP's fgetcsv behavior with 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 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 will be an array with a single value equals to null |
84
|
|
|
* - Otherwise the array contains strings. |
85
|
|
|
* |
86
|
|
|
* @param SplFileObject|Stream $document |
87
|
|
|
*/ |
88
|
42 |
|
public static function parse($document, string $delimiter = ',', string $enclosure = '"'): Generator |
89
|
|
|
{ |
90
|
42 |
|
self::$document = self::filterDocument($document); |
91
|
39 |
|
self::$delimiter = self::filterControl($delimiter, 'delimiter'); |
92
|
36 |
|
self::$enclosure = self::filterControl($enclosure, 'enclosure'); |
93
|
33 |
|
self::$trim_mask = str_replace([self::$delimiter, self::$enclosure], '', " \t\0\x0B"); |
94
|
33 |
|
self::$document->setFlags(0); |
95
|
33 |
|
self::$document->rewind(); |
96
|
33 |
|
while (self::$document->valid()) { |
97
|
33 |
|
$record = []; |
98
|
33 |
|
self::$line = self::$document->fgets(); |
99
|
|
|
do { |
100
|
33 |
|
$method = 'extractFieldContent'; |
101
|
33 |
|
$buffer = ltrim(self::$line, self::$trim_mask); |
102
|
33 |
|
if (($buffer[0] ?? '') === self::$enclosure) { |
103
|
24 |
|
$method = 'extractEnclosedFieldContent'; |
104
|
24 |
|
self::$line = $buffer; |
105
|
|
|
} |
106
|
|
|
|
107
|
33 |
|
$record[] = self::$method(); |
108
|
33 |
|
} while (false !== self::$line); |
109
|
|
|
|
110
|
33 |
|
yield $record; |
111
|
|
|
} |
112
|
33 |
|
} |
113
|
|
|
|
114
|
|
|
/** |
115
|
|
|
* Filter the submitted document. |
116
|
|
|
* |
117
|
|
|
* @param SplFileObject|Stream $document |
118
|
|
|
* |
119
|
|
|
* @return SplFileObject|Stream |
120
|
|
|
*/ |
121
|
9 |
|
private static function filterDocument($document) |
122
|
|
|
{ |
123
|
9 |
|
if ($document instanceof Stream || $document instanceof SplFileObject) { |
124
|
6 |
|
return $document; |
125
|
|
|
} |
126
|
|
|
|
127
|
3 |
|
throw new TypeError(sprintf( |
128
|
3 |
|
'Expected a %s or an SplFileObject object, %s given', |
129
|
3 |
|
Stream::class, |
130
|
3 |
|
is_object($document) ? get_class($document) : gettype($document) |
131
|
|
|
)); |
132
|
|
|
} |
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* Filter a control character. |
136
|
|
|
* |
137
|
|
|
* @throws Exception if the string is not a single byte character |
138
|
|
|
*/ |
139
|
9 |
|
private static function filterControl(string $value, string $name): string |
140
|
|
|
{ |
141
|
9 |
|
if (1 === strlen($value)) { |
142
|
6 |
|
return $value; |
143
|
|
|
} |
144
|
|
|
|
145
|
6 |
|
throw new Exception(sprintf('Expected %s to be a single character %s given', $name, $value)); |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
/** |
149
|
|
|
* Extract field without enclosure as per RFC4180. |
150
|
|
|
* |
151
|
|
|
* - Leading and trailing whitespaces must be removed. |
152
|
|
|
* - trailing line-breaks must be removed. |
153
|
|
|
* |
154
|
|
|
* @return null|string |
155
|
|
|
*/ |
156
|
33 |
|
private static function extractFieldContent() |
157
|
|
|
{ |
158
|
33 |
|
if (in_array(self::$line, self::FIELD_BREAKS, true)) { |
159
|
3 |
|
self::$line = false; |
160
|
|
|
|
161
|
3 |
|
return null; |
162
|
|
|
} |
163
|
|
|
|
164
|
30 |
|
list($content, self::$line) = explode(self::$delimiter, self::$line, 2) + [1 => false]; |
165
|
30 |
|
if (false === self::$line) { |
166
|
24 |
|
return rtrim($content, "\r\n"); |
167
|
|
|
} |
168
|
|
|
|
169
|
30 |
|
return $content; |
170
|
|
|
} |
171
|
|
|
|
172
|
|
|
/** |
173
|
|
|
* Extract field with enclosure as per RFC4180. |
174
|
|
|
* |
175
|
|
|
* - Field content can spread on multiple document lines. |
176
|
|
|
* - Content inside enclosure must be preserved. |
177
|
|
|
* - Double enclosure sequence must be replaced by single enclosure character. |
178
|
|
|
* - Trailing line break must be removed if they are not part of the field content. |
179
|
|
|
* - Invalid fields content are treated as per fgetcsv behavior. |
180
|
|
|
*/ |
181
|
24 |
|
private static function extractEnclosedFieldContent(): string |
182
|
|
|
{ |
183
|
24 |
|
if ((self::$line[0] ?? '') === self::$enclosure) { |
184
|
24 |
|
self::$line = substr(self::$line, 1); |
185
|
|
|
} |
186
|
|
|
|
187
|
24 |
|
$content = ''; |
188
|
24 |
|
while (false !== self::$line) { |
189
|
24 |
|
list($buffer, $remainder) = explode(self::$enclosure, self::$line, 2) + [1 => false]; |
190
|
24 |
|
$content .= $buffer; |
191
|
24 |
|
if (false !== $remainder) { |
192
|
21 |
|
self::$line = $remainder; |
193
|
21 |
|
break; |
194
|
|
|
} |
195
|
9 |
|
self::$line = self::$document->fgets(); |
196
|
|
|
} |
197
|
|
|
|
198
|
24 |
|
if (in_array(self::$line, self::FIELD_BREAKS, true)) { |
199
|
12 |
|
self::$line = false; |
200
|
|
|
|
201
|
12 |
|
return rtrim($content, "\r\n"); |
202
|
|
|
} |
203
|
|
|
|
204
|
21 |
|
$char = self::$line[0] ?? ''; |
205
|
21 |
|
if (self::$delimiter === $char) { |
206
|
15 |
|
self::$line = substr(self::$line, 1); |
207
|
|
|
|
208
|
15 |
|
return $content; |
209
|
|
|
} |
210
|
|
|
|
211
|
18 |
|
if (self::$enclosure === $char) { |
212
|
9 |
|
return $content.self::$enclosure.self::extractEnclosedFieldContent(); |
213
|
|
|
} |
214
|
|
|
|
215
|
9 |
|
return $content.self::extractFieldContent(); |
216
|
|
|
} |
217
|
|
|
} |
218
|
|
|
|