|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Remorhaz\UniLex\RegExp\FSM; |
|
4
|
|
|
|
|
5
|
|
|
use Remorhaz\UniLex\Exception as UniLexException; |
|
6
|
|
|
|
|
7
|
|
|
class LanguageBuilder |
|
8
|
|
|
{ |
|
9
|
|
|
|
|
10
|
|
|
private $symbolTable; |
|
11
|
|
|
|
|
12
|
|
|
private $transitionMap; |
|
13
|
|
|
|
|
14
|
|
|
public function __construct(SymbolTable $symbolTable, TransitionMap $transitionMap) |
|
15
|
|
|
{ |
|
16
|
|
|
$this->symbolTable = $symbolTable; |
|
17
|
|
|
$this->transitionMap = $transitionMap; |
|
18
|
|
|
} |
|
19
|
|
|
|
|
20
|
|
|
public static function forNfa(Nfa $nfa): self |
|
21
|
|
|
{ |
|
22
|
|
|
return new self($nfa->getSymbolTable(), $nfa->getSymbolTransitionMap()); |
|
23
|
|
|
} |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @param int $stateIn |
|
27
|
|
|
* @param int $stateOut |
|
28
|
|
|
* @param Range ...$ranges |
|
29
|
|
|
* @throws UniLexException |
|
30
|
|
|
*/ |
|
31
|
|
|
public function addTransition(int $stateIn, int $stateOut, Range ...$ranges): void |
|
32
|
|
|
{ |
|
33
|
|
|
$this->transitionMap->addTransition($stateIn, $stateOut, $this->getSymbolList(...$ranges)); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @param Range ...$ranges |
|
38
|
|
|
* @return array |
|
39
|
|
|
* @throws UniLexException |
|
40
|
|
|
*/ |
|
41
|
|
|
public function getSymbolList(Range ...$ranges): array |
|
42
|
|
|
{ |
|
43
|
|
|
$rangeSetCalc = new RangeSetCalc(); |
|
44
|
|
|
$newRangeSet = new RangeSet(...$ranges); |
|
45
|
|
|
$symbolList = []; |
|
46
|
|
|
$shouldAddNewSymbol = true; |
|
47
|
|
|
foreach ($this->symbolTable->getRangeSetList() as $symbolId => $oldRangeSet) { |
|
48
|
|
|
if ($rangeSetCalc->equals($oldRangeSet, $newRangeSet)) { |
|
49
|
|
|
$symbolList[] = $symbolId; |
|
50
|
|
|
$shouldAddNewSymbol = false; |
|
51
|
|
|
break; |
|
52
|
|
|
} |
|
53
|
|
|
$rangeSetDiff = $rangeSetCalc->xor($oldRangeSet, $newRangeSet); |
|
54
|
|
|
$onlyInOldRangeSet = $rangeSetCalc->and($oldRangeSet, $rangeSetDiff); |
|
55
|
|
|
if ($onlyInOldRangeSet->isEmpty()) { |
|
56
|
|
|
$symbolList[] = $symbolId; |
|
57
|
|
|
$newRangeSet = $rangeSetDiff; |
|
58
|
|
|
continue; |
|
59
|
|
|
} |
|
60
|
|
|
if ($rangeSetCalc->equals($onlyInOldRangeSet, $oldRangeSet)) { |
|
61
|
|
|
continue; |
|
62
|
|
|
} |
|
63
|
|
|
$splitSymbolId = $this |
|
64
|
|
|
->symbolTable |
|
65
|
|
|
->replaceSymbol($symbolId, $onlyInOldRangeSet) |
|
66
|
|
|
->addSymbol($and = $rangeSetCalc->and($oldRangeSet, $newRangeSet)); |
|
67
|
|
|
$this->splitSymbolInTransitions($symbolId, $splitSymbolId); |
|
68
|
|
|
$symbolList[] = $splitSymbolId; |
|
69
|
|
|
$newRangeSet = $rangeSetCalc->and($newRangeSet, $rangeSetDiff); |
|
70
|
|
|
} |
|
71
|
|
|
if ($shouldAddNewSymbol && !$newRangeSet->isEmpty()) { |
|
72
|
|
|
$newSymbolId = $this->symbolTable->addSymbol($newRangeSet); |
|
73
|
|
|
$symbolList[] = $newSymbolId; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return $symbolList; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
private function splitSymbolInTransitions(int $symbolId, int $symbolToAdd): void |
|
80
|
|
|
{ |
|
81
|
|
|
$addSymbol = function (array $symbolList) use ($symbolId, $symbolToAdd) { |
|
82
|
|
|
if (in_array($symbolId, $symbolList)) { |
|
83
|
|
|
$symbolList[] = $symbolToAdd; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
return $symbolList; |
|
87
|
|
|
}; |
|
88
|
|
|
$this->transitionMap->replaceEachTransition($addSymbol); |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
|