|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Remorhaz\UniLex\Parser\LL1; |
|
6
|
|
|
|
|
7
|
|
|
use Remorhaz\UniLex\Exception; |
|
8
|
|
|
use Remorhaz\UniLex\Grammar\ContextFree\GrammarInterface; |
|
9
|
|
|
use Remorhaz\UniLex\Parser\LL1\Lookup\Table; |
|
10
|
|
|
use Remorhaz\UniLex\Parser\Production; |
|
11
|
|
|
use Remorhaz\UniLex\Parser\Symbol; |
|
12
|
|
|
use Remorhaz\UniLex\Stack\SymbolStack; |
|
13
|
|
|
use Remorhaz\UniLex\Stack\StackableSymbolInterface; |
|
14
|
|
|
use Remorhaz\UniLex\Lexer\Token; |
|
15
|
|
|
use Remorhaz\UniLex\Lexer\TokenReaderInterface; |
|
16
|
|
|
use Remorhaz\UniLex\Parser\LL1\Lookup\TableInterface; |
|
17
|
|
|
use Remorhaz\UniLex\Parser\LL1\Lookup\TableBuilder; |
|
18
|
|
|
use Throwable; |
|
19
|
|
|
|
|
20
|
|
|
class Parser |
|
21
|
|
|
{ |
|
22
|
|
|
private $grammar; |
|
23
|
|
|
|
|
24
|
|
|
private $lookupTable; |
|
25
|
|
|
|
|
26
|
|
|
private $symbolStack; |
|
27
|
|
|
|
|
28
|
|
|
private $token; |
|
29
|
|
|
|
|
30
|
|
|
private $tokenReader; |
|
31
|
|
|
|
|
32
|
|
|
private $listener; |
|
33
|
|
|
|
|
34
|
|
|
private $nextSymbolIndex = 0; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var Production[] |
|
38
|
|
|
*/ |
|
39
|
|
|
private $productionMap = []; |
|
40
|
|
|
|
|
41
|
|
|
public function __construct( |
|
42
|
|
|
GrammarInterface $grammar, |
|
43
|
|
|
TokenReaderInterface $tokenReader, |
|
44
|
|
|
ParserListenerInterface $listener |
|
45
|
|
|
) { |
|
46
|
|
|
$this->grammar = $grammar; |
|
47
|
|
|
$this->tokenReader = $tokenReader; |
|
48
|
|
|
$this->listener = $listener; |
|
49
|
|
|
$this->symbolStack = new SymbolStack(); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* @param string $fileName |
|
54
|
|
|
* @throws Exception |
|
55
|
|
|
*/ |
|
56
|
|
|
public function loadLookupTable(string $fileName): void |
|
57
|
|
|
{ |
|
58
|
|
|
/** @noinspection PhpIncludeInspection */ |
|
59
|
|
|
$data = @include $fileName; |
|
60
|
|
|
if (false === $data) { |
|
61
|
|
|
throw new Exception("Failed to load lookup table from file {$fileName}"); |
|
62
|
|
|
} |
|
63
|
|
|
$table = new Table(); |
|
64
|
|
|
$table->importMap($data); |
|
65
|
|
|
$this->lookupTable = $table; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* @throws Exception |
|
70
|
|
|
* @throws UnexpectedTokenException |
|
71
|
|
|
*/ |
|
72
|
|
|
public function run(): void |
|
73
|
|
|
{ |
|
74
|
|
|
$this->initRun(); |
|
75
|
|
|
while ($this->hasSymbolInStack()) { |
|
76
|
|
|
$symbol = $this->popSymbol(); |
|
77
|
|
|
if ($symbol instanceof Symbol) { |
|
78
|
|
|
$this->isTerminalSymbol($symbol) |
|
79
|
|
|
? $this->readSymbolToken($symbol) |
|
80
|
|
|
: $this->pushMatchingProduction($symbol); |
|
81
|
|
|
continue; |
|
82
|
|
|
} |
|
83
|
|
|
if ($symbol instanceof Production) { |
|
84
|
|
|
$this->listener->onFinishProduction($symbol); |
|
85
|
|
|
} |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
private function initRun(): void |
|
90
|
|
|
{ |
|
91
|
|
|
$this->nextSymbolIndex = 0; |
|
92
|
|
|
$this->productionMap = []; |
|
93
|
|
|
$this->symbolStack->reset(); |
|
94
|
|
|
$this->listener->onStart(); |
|
95
|
|
|
$this->pushProduction($this->createRootProduction()); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
private function createRootProduction(): Production |
|
99
|
|
|
{ |
|
100
|
|
|
$rootSymbol = new Symbol($this->getNextSymbolIndex(), $this->grammar->getRootSymbol()); |
|
101
|
|
|
$this->listener->onRootSymbol($rootSymbol); |
|
102
|
|
|
return $this->createParsedProduction($rootSymbol, 0); |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
private function getNextSymbolIndex(): int |
|
106
|
|
|
{ |
|
107
|
|
|
return $this->nextSymbolIndex++; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
private function hasSymbolInStack(): bool |
|
111
|
|
|
{ |
|
112
|
|
|
return !$this->symbolStack->isEmpty(); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
private function isTerminalSymbol(Symbol $symbol): bool |
|
116
|
|
|
{ |
|
117
|
|
|
return $this->grammar->isTerminal($symbol->getSymbolId()); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
private function previewToken(): Token |
|
121
|
|
|
{ |
|
122
|
|
|
if (!isset($this->token)) { |
|
123
|
|
|
$this->token = $this->tokenReader->read(); |
|
124
|
|
|
} |
|
125
|
|
|
return $this->token; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* @return StackableSymbolInterface |
|
130
|
|
|
* @throws Exception |
|
131
|
|
|
*/ |
|
132
|
|
|
private function popSymbol(): StackableSymbolInterface |
|
133
|
|
|
{ |
|
134
|
|
|
return $this->symbolStack->pop(); |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
/** |
|
138
|
|
|
* @param Symbol $symbol |
|
139
|
|
|
* @throws Exception |
|
140
|
|
|
*/ |
|
141
|
|
|
private function onSymbol(Symbol $symbol): void |
|
142
|
|
|
{ |
|
143
|
|
|
if (!isset($this->productionMap[$symbol->getIndex()])) { |
|
144
|
|
|
throw new Exception("No production in map for symbol {$symbol->getIndex()}"); |
|
145
|
|
|
} |
|
146
|
|
|
[$symbolIndex, $production] = $this->productionMap[$symbol->getIndex()]; |
|
147
|
|
|
$this->listener->onSymbol($symbolIndex, $production); |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
/** |
|
151
|
|
|
* @param Symbol $symbol |
|
152
|
|
|
* @throws Exception |
|
153
|
|
|
*/ |
|
154
|
|
|
private function readSymbolToken(Symbol $symbol): void |
|
155
|
|
|
{ |
|
156
|
|
|
$token = $this->previewToken(); |
|
157
|
|
|
if (!$this->grammar->tokenMatchesTerminal($symbol->getSymbolId(), $token->getType())) { |
|
158
|
|
|
throw new Exception("Unexpected token {$token->getType()} for symbol {$symbol->getSymbolId()}"); |
|
159
|
|
|
} |
|
160
|
|
|
$token->isEoi() |
|
161
|
|
|
? $this->listener->onEoi($symbol, $token) |
|
162
|
|
|
: $this->listener->onToken($symbol, $token); |
|
163
|
|
|
$this->onSymbol($symbol); |
|
164
|
|
|
unset($this->token); |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
/** |
|
168
|
|
|
* @param Symbol $symbol |
|
169
|
|
|
* @throws UnexpectedTokenException |
|
170
|
|
|
* @throws Exception |
|
171
|
|
|
*/ |
|
172
|
|
|
private function pushMatchingProduction(Symbol $symbol): void |
|
173
|
|
|
{ |
|
174
|
|
|
$this->onSymbol($symbol); |
|
175
|
|
|
$production = $this->getMatchingProduction($symbol, $this->previewToken()); |
|
176
|
|
|
$this->pushProduction($production); |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
private function pushProduction(Production $production): void |
|
180
|
|
|
{ |
|
181
|
|
|
$this->symbolStack->push($production); |
|
182
|
|
|
foreach ($production->getSymbolList() as $symbolIndexInProduction => $symbol) { |
|
183
|
|
|
$this->productionMap[$symbol->getIndex()] = [$symbolIndexInProduction, $production]; |
|
184
|
|
|
} |
|
185
|
|
|
$this->symbolStack->push(...$production->getSymbolList()); |
|
186
|
|
|
$this->listener->onBeginProduction($production); |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
/** |
|
190
|
|
|
* @param Symbol $symbol |
|
191
|
|
|
* @param Token $token |
|
192
|
|
|
* @return Production |
|
193
|
|
|
* @throws UnexpectedTokenException |
|
194
|
|
|
* @throws Exception |
|
195
|
|
|
*/ |
|
196
|
|
|
private function getMatchingProduction(Symbol $symbol, Token $token): Production |
|
197
|
|
|
{ |
|
198
|
|
|
$lookupTable = $this->getLookupTable(); |
|
199
|
|
|
try { |
|
200
|
|
|
$productionIndex = $lookupTable->getProductionIndex($symbol->getSymbolId(), $token->getType()); |
|
201
|
|
|
} catch (Throwable $e) { |
|
202
|
|
|
$expectedTokenList = $lookupTable->getExpectedTokenList($symbol->getSymbolId()); |
|
203
|
|
|
$error = new UnexpectedTokenError($token, $symbol, ...$expectedTokenList); |
|
204
|
|
|
throw new UnexpectedTokenException($error, 0, $e); |
|
205
|
|
|
} |
|
206
|
|
|
return $this->createParsedProduction($symbol, $productionIndex); |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
private function createParsedProduction(Symbol $symbol, $productionIndex): Production |
|
210
|
|
|
{ |
|
211
|
|
|
$grammarProduction = $this |
|
212
|
|
|
->grammar |
|
213
|
|
|
->getProduction($symbol->getSymbolId(), $productionIndex); |
|
214
|
|
|
$symbolList = []; |
|
215
|
|
|
foreach ($grammarProduction->getSymbolList() as $symbolId) { |
|
216
|
|
|
$symbolList[] = new Symbol($this->getNextSymbolIndex(), $symbolId); |
|
217
|
|
|
} |
|
218
|
|
|
return new Production($symbol, $grammarProduction->getIndex(), ...$symbolList); |
|
219
|
|
|
} |
|
220
|
|
|
|
|
221
|
|
|
/** |
|
222
|
|
|
* @return TableInterface |
|
223
|
|
|
* @throws Exception |
|
224
|
|
|
*/ |
|
225
|
|
|
private function getLookupTable(): TableInterface |
|
226
|
|
|
{ |
|
227
|
|
|
if (!isset($this->lookupTable)) { |
|
228
|
|
|
$builder = new TableBuilder($this->grammar); |
|
229
|
|
|
$this->lookupTable = $builder->getTable(); |
|
230
|
|
|
} |
|
231
|
|
|
return $this->lookupTable; |
|
232
|
|
|
} |
|
233
|
|
|
} |
|
234
|
|
|
|