|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Remorhaz\UniLex\Parser\LL1\Lookup; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* Helper to calculate FIRST() sets. It's a part of LL(1) lookup table generation algorithm. FIRST(X) set |
|
9
|
|
|
* contains terminals (and, optionally, ε-production) that can occur as a starting token in production X. |
|
10
|
|
|
*/ |
|
11
|
|
|
class First extends Set implements FirstInterface |
|
12
|
|
|
{ |
|
13
|
|
|
private $epsilonMap = []; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Returns FIRST(X) set. |
|
17
|
|
|
* |
|
18
|
|
|
* @param int ...$symbolIdList |
|
19
|
|
|
* @return array |
|
20
|
|
|
*/ |
|
21
|
|
|
public function getProductionTokens(int ...$symbolIdList): array |
|
22
|
|
|
{ |
|
23
|
|
|
$first = []; |
|
24
|
|
|
foreach ($symbolIdList as $symbolId) { |
|
25
|
|
|
$first = array_merge($first, $this->getTokens($symbolId)); |
|
26
|
|
|
if (!$this->productionHasEpsilon($symbolId)) { |
|
27
|
|
|
break; |
|
28
|
|
|
} |
|
29
|
|
|
} |
|
30
|
|
|
return $first; |
|
31
|
|
|
} |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* Adds ε-production to FIRST(X) set. |
|
35
|
|
|
* |
|
36
|
|
|
* @param int $symbolId |
|
37
|
|
|
*/ |
|
38
|
|
|
public function addEpsilon(int $symbolId): void |
|
39
|
|
|
{ |
|
40
|
|
|
if ($this->hasEpsilon($symbolId)) { |
|
41
|
|
|
return; |
|
42
|
|
|
} |
|
43
|
|
|
$this->epsilonMap[$symbolId] = true; |
|
44
|
|
|
$this->increaseChangeCount(); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Reports presence of ε-production in FIRST(X) sets for all given X. |
|
49
|
|
|
* |
|
50
|
|
|
* @param int ...$symbolIdList |
|
51
|
|
|
* @return bool |
|
52
|
|
|
*/ |
|
53
|
|
|
public function productionHasEpsilon(int ...$symbolIdList): bool |
|
54
|
|
|
{ |
|
55
|
|
|
if (empty($symbolIdList)) { |
|
56
|
|
|
return true; |
|
57
|
|
|
} |
|
58
|
|
|
foreach ($symbolIdList as $symbolId) { |
|
59
|
|
|
if (!$this->hasEpsilon($symbolId)) { |
|
60
|
|
|
return false; |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
return true; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
public function hasEpsilon(int $symbolId) |
|
67
|
|
|
{ |
|
68
|
|
|
return $this->epsilonMap[$symbolId] ?? false; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Adds all tokens from source production's FIRST(X) to target production's FIRST(Y). |
|
73
|
|
|
* |
|
74
|
|
|
* @param int $targetSymbolId |
|
75
|
|
|
* @param int ...$sourceSymbolIdList |
|
76
|
|
|
*/ |
|
77
|
|
|
public function mergeProductionTokens(int $targetSymbolId, int ...$sourceSymbolIdList): void |
|
78
|
|
|
{ |
|
79
|
|
|
$this->addToken($targetSymbolId, ...$this->getProductionTokens(...$sourceSymbolIdList)); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
public function mergeProductionEpsilons(int $targetSymbolId, int ...$sourceSymbolIdList): void |
|
83
|
|
|
{ |
|
84
|
|
|
if ($this->productionHasEpsilon(...$sourceSymbolIdList)) { |
|
85
|
|
|
$this->addEpsilon($targetSymbolId); |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
|
|
|
remorhaz /
php-unilex
