|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* @license http://opensource.org/licenses/mit-license.php MIT |
|
7
|
|
|
* @link https://github.com/nicoSWD |
|
8
|
|
|
* @author Nicolas Oelgart <[email protected]> |
|
9
|
|
|
*/ |
|
10
|
|
|
namespace nicoSWD\Rule\Tokenizer; |
|
11
|
|
|
|
|
12
|
|
|
use nicoSWD\Rule\Grammar\Grammar; |
|
13
|
|
|
use nicoSWD\Rule\TokenStream\Token\TokenFactory; |
|
14
|
|
|
use SplPriorityQueue; |
|
15
|
|
|
use stdClass; |
|
16
|
|
|
|
|
17
|
|
|
final class Tokenizer implements TokenizerInterface |
|
18
|
|
|
{ |
|
19
|
|
|
/** @var TokenFactory */ |
|
20
|
|
|
private $tokenFactory; |
|
21
|
|
|
/** @var Grammar */ |
|
22
|
|
|
private $grammar; |
|
23
|
|
|
/** @var stdClass[] */ |
|
24
|
|
|
private $internalTokens = []; |
|
25
|
|
|
/** @var string */ |
|
26
|
|
|
private $regex = ''; |
|
27
|
|
|
|
|
28
|
234 |
|
public function __construct(Grammar $grammar, TokenFactory $tokenFactory) |
|
29
|
|
|
{ |
|
30
|
234 |
|
$this->tokenFactory = $tokenFactory; |
|
31
|
234 |
|
$this->grammar = $grammar; |
|
32
|
|
|
|
|
33
|
234 |
|
foreach ($grammar->getDefinition() as list($class, $regex, $priority)) { |
|
34
|
234 |
|
$this->registerToken($class, $regex, $priority); |
|
35
|
|
|
} |
|
36
|
234 |
|
} |
|
37
|
|
|
|
|
38
|
228 |
|
public function tokenize(string $string): TokenStack |
|
39
|
|
|
{ |
|
40
|
228 |
|
$stack = new TokenStack(); |
|
41
|
228 |
|
$regex = $this->getRegex(); |
|
42
|
228 |
|
$offset = 0; |
|
43
|
|
|
|
|
44
|
228 |
|
while (preg_match($regex, $string, $matches, 0, $offset)) { |
|
45
|
226 |
|
$token = $this->getMatchedToken($matches); |
|
46
|
226 |
|
$className = $this->tokenFactory->createFromTokenName($token); |
|
47
|
|
|
|
|
48
|
226 |
|
$stack->attach(new $className( |
|
49
|
226 |
|
$matches[$token], |
|
50
|
226 |
|
$offset, |
|
51
|
226 |
|
$stack |
|
52
|
|
|
)); |
|
53
|
|
|
|
|
54
|
226 |
|
$offset += strlen($matches[0]); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
228 |
|
return $stack; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
134 |
|
public function getGrammar(): Grammar |
|
61
|
|
|
{ |
|
62
|
134 |
|
return $this->grammar; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
234 |
|
private function registerToken(string $class, string $regex, int $priority) |
|
66
|
|
|
{ |
|
67
|
234 |
|
$token = new stdClass(); |
|
68
|
234 |
|
$token->class = $class; |
|
69
|
234 |
|
$token->regex = $regex; |
|
70
|
234 |
|
$token->priority = $priority; |
|
71
|
|
|
|
|
72
|
234 |
|
$this->internalTokens[$class] = $token; |
|
73
|
234 |
|
} |
|
74
|
|
|
|
|
75
|
228 |
|
private function getMatchedToken(array $matches): string |
|
76
|
|
|
{ |
|
77
|
228 |
|
foreach ($matches as $key => $value) { |
|
78
|
226 |
|
if ($value !== '' && !is_int($key)) { |
|
79
|
226 |
|
return $key; |
|
80
|
|
|
} |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
2 |
|
return 'Unknown'; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
228 |
|
private function getRegex(): string |
|
87
|
|
|
{ |
|
88
|
228 |
|
if (!$this->regex) { |
|
89
|
170 |
|
$regex = []; |
|
90
|
|
|
|
|
91
|
170 |
|
foreach ($this->getQueue() as $token) { |
|
92
|
170 |
|
$regex[] = "(?<$token->class>$token->regex)"; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
170 |
|
$this->regex = '~(' . implode('|', $regex) . ')~As'; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
228 |
|
return $this->regex; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
170 |
|
private function getQueue(): SplPriorityQueue |
|
102
|
|
|
{ |
|
103
|
170 |
|
$queue = new SplPriorityQueue(); |
|
104
|
|
|
|
|
105
|
170 |
|
foreach ($this->internalTokens as $class) { |
|
106
|
170 |
|
$queue->insert($class, $class->priority); |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
170 |
|
return $queue; |
|
110
|
|
|
} |
|
111
|
|
|
} |
|
112
|
|
|
|
nicoSWD /
php-rule-parser
