1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
|
4
|
|
|
namespace CondParse; |
5
|
|
|
|
6
|
|
|
|
7
|
|
|
use CondParse\Exception\LexerException; |
8
|
|
|
|
9
|
|
|
class Lexer implements LexerInterface |
10
|
|
|
{ |
11
|
|
|
/** @var callable[] */ |
12
|
|
|
private $postFunctions = []; |
13
|
|
|
|
14
|
|
|
/** @var LexerTokenFactory */ |
15
|
|
|
private $lexerTokenFactory; |
16
|
|
|
|
17
|
18 |
|
public function __construct() |
18
|
|
|
{ |
19
|
18 |
|
$this->lexerTokenFactory = new LexerTokenFactory; |
20
|
18 |
|
} |
21
|
|
|
|
22
|
|
|
/** |
23
|
|
|
* @param callable $postFunction |
24
|
|
|
* @return $this |
25
|
|
|
*/ |
26
|
9 |
|
public function registerPostFunction(callable $postFunction) |
27
|
|
|
{ |
28
|
9 |
|
$this->postFunctions[] = $postFunction; |
29
|
9 |
|
return $this; |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @param TokenMap $tokenMap |
34
|
|
|
* @return string |
35
|
|
|
*/ |
36
|
18 |
|
protected function buildRegex(TokenMap $tokenMap) |
37
|
|
|
{ |
38
|
18 |
|
return sprintf( |
39
|
18 |
|
'/%s/Si', |
40
|
18 |
|
join('|', |
41
|
|
|
array_map(function ($key, $value) { |
42
|
18 |
|
return sprintf('(?P<%s>%s)', $key, $value); |
43
|
18 |
|
}, array_keys($tokenMap->getTokens()), $tokenMap->getTokens()) |
44
|
18 |
|
) |
45
|
18 |
|
); |
46
|
|
|
} |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* @param string $token |
50
|
|
|
* @param string $match |
51
|
|
|
* @return string |
52
|
|
|
*/ |
53
|
18 |
|
protected function applyPostFunctions($token, $match) |
54
|
|
|
{ |
55
|
18 |
|
foreach ($this->postFunctions as $postFunction) { |
56
|
9 |
|
$match = call_user_func($postFunction, $token, $match); |
57
|
18 |
|
} |
58
|
|
|
|
59
|
18 |
|
return $match; |
60
|
|
|
} |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* @param string $conditionString |
64
|
|
|
* @param string $regex |
65
|
|
|
* @return \Traversable <LexerToken> |
66
|
|
|
* @throws LexerException |
67
|
|
|
*/ |
68
|
18 |
|
protected function getTokenStreamWithRegex($conditionString, $regex) |
69
|
|
|
{ |
70
|
18 |
|
$offSet = 0; |
71
|
|
|
|
72
|
18 |
|
$matches = []; |
73
|
18 |
|
while (preg_match($regex, $conditionString, $matches, 0, $offSet) !== 0) { |
74
|
18 |
|
list($match, $token) = $this->extractMatch($matches); |
75
|
|
|
|
76
|
18 |
|
$this->checkMatchOffset($conditionString, $match, $offSet); |
77
|
|
|
|
78
|
18 |
|
$offSet += strlen($match); |
79
|
|
|
|
80
|
18 |
|
yield $this->lexerTokenFactory->buildLexerToken($token, $this->applyPostFunctions($token, $match)); |
81
|
18 |
|
} |
82
|
17 |
|
} |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* @param string $conditionString |
86
|
|
|
* @param TokenMap $tokenMap |
87
|
|
|
* @return \Traversable <string> |
88
|
|
|
*/ |
89
|
18 |
|
public function getTokenStream($conditionString, TokenMap $tokenMap) { |
90
|
18 |
|
return $this->getTokenStreamWithRegex($conditionString, $this->buildRegex($tokenMap)); |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* @param $matches |
95
|
|
|
* @return array |
96
|
|
|
*/ |
97
|
|
|
protected function extractMatch($matches) |
98
|
|
|
{ |
99
|
18 |
|
$matches = array_filter($matches, function ($value, $key) { |
100
|
18 |
|
return is_string($key) && !empty($value); |
101
|
18 |
|
}, ARRAY_FILTER_USE_BOTH); |
102
|
|
|
|
103
|
18 |
|
$match = each($matches); |
104
|
18 |
|
$token = $match['key']; |
105
|
18 |
|
$match = $match['value']; |
106
|
18 |
|
return [$match, $token]; |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
/** |
110
|
|
|
* @param $conditionString |
111
|
|
|
* @param $match |
112
|
|
|
* @param $offSet |
113
|
|
|
* @throws LexerException |
114
|
|
|
*/ |
115
|
18 |
|
protected function checkMatchOffset($conditionString, $match, $offSet) |
116
|
|
|
{ |
117
|
18 |
|
if (($matchOffset = strpos($conditionString, $match, $offSet)) !== $offSet) { |
118
|
1 |
|
throw new LexerException(sprintf( |
119
|
1 |
|
'Found unrecognized token <%s> at offset %d', |
120
|
1 |
|
substr($conditionString, $offSet, ($matchOffset - $offSet)), |
121
|
|
|
$offSet |
122
|
1 |
|
)); |
123
|
|
|
} |
124
|
18 |
|
} |
125
|
|
|
} |
126
|
|
|
|