|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Remorhaz\UniLex\RegExp\Grammar; |
|
6
|
|
|
|
|
7
|
|
|
use Remorhaz\UniLex\Exception; |
|
8
|
|
|
use Remorhaz\UniLex\Parser\Symbol; |
|
9
|
|
|
use Remorhaz\UniLex\Lexer\Token; |
|
10
|
|
|
use Throwable; |
|
11
|
|
|
|
|
12
|
|
|
class TokenTranslationScheme |
|
13
|
|
|
{ |
|
14
|
|
|
private $symbol; |
|
15
|
|
|
|
|
16
|
|
|
private $token; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* @param Symbol $symbol |
|
20
|
|
|
* @param Token $token |
|
21
|
|
|
* @throws Exception |
|
22
|
|
|
*/ |
|
23
|
|
|
public function applyActions(Symbol $symbol, Token $token): void |
|
24
|
|
|
{ |
|
25
|
|
|
$this->setContext($symbol, $token); |
|
26
|
|
|
|
|
27
|
|
|
switch ($symbol->getSymbolId()) { |
|
28
|
|
|
case SymbolType::T_DIGIT_ZERO: |
|
29
|
|
|
case SymbolType::T_DIGIT_OCT: |
|
30
|
|
|
$this |
|
31
|
|
|
->copyTokenAttribute('s.code', TokenAttribute::CODE) |
|
32
|
|
|
->copyTokenAttribute('s.oct_digit', TokenAttribute::DIGIT) |
|
33
|
|
|
->copyTokenAttribute('s.dec_digit', TokenAttribute::DIGIT) |
|
34
|
|
|
->copyTokenAttribute('s.hex_digit', TokenAttribute::DIGIT); |
|
35
|
|
|
break; |
|
36
|
|
|
|
|
37
|
|
|
case SymbolType::T_DIGIT_DEC: |
|
38
|
|
|
$this |
|
39
|
|
|
->copyTokenAttribute('s.code', TokenAttribute::CODE) |
|
40
|
|
|
->copyTokenAttribute('s.dec_digit', TokenAttribute::DIGIT) |
|
41
|
|
|
->copyTokenAttribute('s.hex_digit', TokenAttribute::DIGIT); |
|
42
|
|
|
break; |
|
43
|
|
|
|
|
44
|
|
|
case SymbolType::T_SMALL_C: |
|
45
|
|
|
case SymbolType::T_OTHER_HEX_LETTER: |
|
46
|
|
|
$this |
|
47
|
|
|
->copyTokenAttribute('s.code', TokenAttribute::CODE) |
|
48
|
|
|
->copyTokenAttribute('s.hex_digit', TokenAttribute::DIGIT); |
|
49
|
|
|
break; |
|
50
|
|
|
|
|
51
|
|
|
case SymbolType::T_COMMA: |
|
52
|
|
|
case SymbolType::T_HYPHEN: |
|
53
|
|
|
case SymbolType::T_CAPITAL_P: |
|
54
|
|
|
case SymbolType::T_RIGHT_SQUARE_BRACKET: |
|
55
|
|
|
case SymbolType::T_SMALL_O: |
|
56
|
|
|
case SymbolType::T_SMALL_P: |
|
57
|
|
|
case SymbolType::T_SMALL_U: |
|
58
|
|
|
case SymbolType::T_SMALL_X: |
|
59
|
|
|
case SymbolType::T_RIGHT_CURLY_BRACKET: |
|
60
|
|
|
case SymbolType::T_CTL_ASCII: |
|
61
|
|
|
case SymbolType::T_PRINTABLE_ASCII_OTHER: |
|
62
|
|
|
case SymbolType::T_NOT_ASCII: |
|
63
|
|
|
case SymbolType::T_DOLLAR: |
|
64
|
|
|
case SymbolType::T_LEFT_BRACKET: |
|
65
|
|
|
case SymbolType::T_RIGHT_BRACKET: |
|
66
|
|
|
case SymbolType::T_STAR: |
|
67
|
|
|
case SymbolType::T_PLUS: |
|
68
|
|
|
case SymbolType::T_QUESTION: |
|
69
|
|
|
case SymbolType::T_LEFT_SQUARE_BRACKET: |
|
70
|
|
|
case SymbolType::T_BACKSLASH: |
|
71
|
|
|
case SymbolType::T_CIRCUMFLEX: |
|
72
|
|
|
case SymbolType::T_LEFT_CURLY_BRACKET: |
|
73
|
|
|
case SymbolType::T_VERTICAL_LINE: |
|
74
|
|
|
case SymbolType::T_DOT: |
|
75
|
|
|
case SymbolType::T_OTHER_ASCII_LETTER: |
|
76
|
|
|
$this |
|
77
|
|
|
->copyTokenAttribute('s.code', TokenAttribute::CODE); |
|
78
|
|
|
break; |
|
79
|
|
|
} |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* @param string $symbolAttr |
|
84
|
|
|
* @param string $tokenAttr |
|
85
|
|
|
* @return $this |
|
86
|
|
|
* @throws Exception |
|
87
|
|
|
*/ |
|
88
|
|
|
private function copyTokenAttribute(string $symbolAttr, string $tokenAttr) |
|
89
|
|
|
{ |
|
90
|
|
|
try { |
|
91
|
|
|
$value = $this |
|
92
|
|
|
->getToken() |
|
93
|
|
|
->getAttribute($tokenAttr); |
|
94
|
|
|
$this |
|
95
|
|
|
->getSymbol() |
|
96
|
|
|
->setAttribute($symbolAttr, $value); |
|
97
|
|
|
} catch (Throwable $e) { |
|
98
|
|
|
$symbolText = "{$this->getSymbol()->getSymbolId()}.{$symbolAttr}"; |
|
99
|
|
|
$tokenText = "{$this->getToken()->getType()}.{$tokenAttr}"; |
|
100
|
|
|
throw new Exception("Failed to synthesize attribute {$symbolText} from token {$tokenText}", 0, $e); |
|
101
|
|
|
} |
|
102
|
|
|
return $this; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
|
|
106
|
|
|
private function setContext(Symbol $symbol, Token $token): void |
|
107
|
|
|
{ |
|
108
|
|
|
$this->symbol = $symbol; |
|
109
|
|
|
$this->token = $token; |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* @return Symbol |
|
114
|
|
|
* @throws Exception |
|
115
|
|
|
*/ |
|
116
|
|
|
private function getSymbol(): Symbol |
|
117
|
|
|
{ |
|
118
|
|
|
if (!isset($this->symbol)) { |
|
119
|
|
|
throw new Exception("No symbol defined in token translation scheme"); |
|
120
|
|
|
} |
|
121
|
|
|
return $this->symbol; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
/** |
|
125
|
|
|
* @return Token |
|
126
|
|
|
* @throws Exception |
|
127
|
|
|
*/ |
|
128
|
|
|
private function getToken(): Token |
|
129
|
|
|
{ |
|
130
|
|
|
if (!isset($this->token)) { |
|
131
|
|
|
throw new Exception("No token defined in token translation scheme"); |
|
132
|
|
|
} |
|
133
|
|
|
return $this->token; |
|
134
|
|
|
} |
|
135
|
|
|
} |
|
136
|
|
|
|