|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of Railt package. |
|
4
|
|
|
* |
|
5
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
6
|
|
|
* file that was distributed with this source code. |
|
7
|
|
|
*/ |
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace Railt\Lexer\Iterator; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* An iterator which returns a list of regex groups |
|
14
|
|
|
*/ |
|
15
|
|
|
class RegexIterator implements \IteratorAggregate |
|
16
|
|
|
{ |
|
17
|
|
|
public const PREG_PARSING_ERROR = 'The error occurs while compiling PCRE'; |
|
18
|
|
|
public const PREG_INTERNAL_ERROR = 'There was an internal PCRE error'; |
|
19
|
|
|
public const PREG_BACKTRACK_LIMIT_ERROR = 'Backtrack limit was exhausted'; |
|
20
|
|
|
public const PREG_RECURSION_LIMIT_ERROR = 'Recursion limit was exhausted'; |
|
21
|
|
|
public const PREG_BAD_UTF8_ERROR = 'The offset didn\'t correspond to the begin of a valid UTF-8 code point'; |
|
22
|
|
|
public const PREG_BAD_UTF8_OFFSET_ERROR = 'Malformed UTF-8 data'; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var string |
|
26
|
|
|
*/ |
|
27
|
|
|
private $pattern; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @var string |
|
31
|
|
|
*/ |
|
32
|
|
|
private $subject; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* RegexIterator constructor. |
|
36
|
|
|
* @param string $pattern |
|
37
|
|
|
* @param string $subject |
|
38
|
|
|
*/ |
|
39
|
6 |
|
public function __construct(string $pattern, string $subject) |
|
40
|
|
|
{ |
|
41
|
6 |
|
$this->pattern = $pattern; |
|
42
|
6 |
|
$this->subject = $subject; |
|
43
|
6 |
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* @return \Traversable|array[] |
|
47
|
|
|
* @throws \InvalidArgumentException |
|
48
|
|
|
* @throws \RuntimeException |
|
49
|
|
|
*/ |
|
50
|
6 |
|
public function getIterator(): \Traversable |
|
51
|
|
|
{ |
|
52
|
6 |
|
$result = new \SplQueue(); |
|
53
|
|
|
|
|
54
|
6 |
|
$status = @\preg_replace_callback($this->pattern, function (array $matches) use ($result): void { |
|
55
|
6 |
|
$result->push($matches); |
|
56
|
6 |
|
}, $this->subject); |
|
57
|
|
|
|
|
58
|
6 |
|
$this->validate($status); |
|
59
|
|
|
|
|
60
|
6 |
|
return $result; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* @param $status |
|
65
|
|
|
* @return void |
|
66
|
|
|
* @throws \InvalidArgumentException |
|
67
|
|
|
* @throws \RuntimeException |
|
68
|
|
|
*/ |
|
69
|
6 |
|
private function validate($status): void |
|
70
|
|
|
{ |
|
71
|
6 |
|
$code = \preg_last_error(); |
|
72
|
|
|
|
|
73
|
6 |
|
if ($code !== \PREG_NO_ERROR) { |
|
74
|
|
|
throw new \RuntimeException($this->getErrorMessage($code), $code); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
6 |
|
if ($status === null) { |
|
78
|
|
|
$parts = \explode(':', \error_get_last()['message'] ?? ''); |
|
79
|
|
|
$error = \sprintf('%s, %s', self::PREG_PARSING_ERROR, \trim(\end($parts))); |
|
80
|
|
|
throw new \InvalidArgumentException($error); |
|
81
|
|
|
} |
|
82
|
6 |
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* @param int $code |
|
86
|
|
|
* @return string |
|
87
|
|
|
*/ |
|
88
|
|
|
private function getErrorMessage(int $code): string |
|
89
|
|
|
{ |
|
90
|
|
|
switch ($code) { |
|
91
|
|
|
case \PREG_INTERNAL_ERROR: |
|
92
|
|
|
return self::PREG_INTERNAL_ERROR; |
|
93
|
|
|
|
|
94
|
|
|
case \PREG_BACKTRACK_LIMIT_ERROR: |
|
95
|
|
|
return self::PREG_BACKTRACK_LIMIT_ERROR; |
|
96
|
|
|
|
|
97
|
|
|
case \PREG_RECURSION_LIMIT_ERROR: |
|
98
|
|
|
return self::PREG_RECURSION_LIMIT_ERROR; |
|
99
|
|
|
|
|
100
|
|
|
case \PREG_BAD_UTF8_ERROR: |
|
101
|
|
|
return self::PREG_BAD_UTF8_ERROR; |
|
102
|
|
|
|
|
103
|
|
|
case \PREG_BAD_UTF8_OFFSET_ERROR: |
|
104
|
|
|
return self::PREG_BAD_UTF8_OFFSET_ERROR; |
|
105
|
|
|
} |
|
106
|
|
|
return 'Unexpected PCRE error (Code ' . $code . ')'; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
/** |
|
110
|
|
|
* Destroy current body |
|
111
|
|
|
*/ |
|
112
|
6 |
|
public function __destruct() |
|
113
|
|
|
{ |
|
114
|
6 |
|
unset($this->pattern, $this->subject); |
|
115
|
6 |
|
} |
|
116
|
|
|
} |
|
117
|
|
|
|
railt /
lexer
