|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of PHP-Yacc package. |
|
4
|
|
|
* |
|
5
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
6
|
|
|
* file that was distributed with this source code. |
|
7
|
|
|
*/ |
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace PhpYacc\Yacc; |
|
11
|
|
|
|
|
12
|
|
|
use PhpYacc\Exception\ParseException; |
|
13
|
|
|
use PhpYacc\Grammar\Context; |
|
14
|
|
|
use PhpYacc\Grammar\Symbol; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Class Parser. |
|
18
|
|
|
*/ |
|
19
|
|
|
class Parser |
|
20
|
|
|
{ |
|
21
|
|
|
/** |
|
22
|
|
|
* @var Context |
|
23
|
|
|
*/ |
|
24
|
|
|
protected $context; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @var Lexer |
|
28
|
|
|
*/ |
|
29
|
|
|
protected $lexer; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var MacroSet |
|
33
|
|
|
*/ |
|
34
|
|
|
protected $macros; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var Symbol |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $eofToken; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @var Symbol |
|
43
|
|
|
*/ |
|
44
|
|
|
protected $errorToken; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @var Symbol |
|
48
|
|
|
*/ |
|
49
|
|
|
protected $startPrime; |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* @var int |
|
53
|
|
|
*/ |
|
54
|
|
|
protected $currentPrecedence = 0; |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* Parser constructor. |
|
58
|
|
|
* |
|
59
|
|
|
* @param Lexer $lexer |
|
60
|
|
|
* @param MacroSet $macros |
|
61
|
|
|
*/ |
|
62
|
|
|
public function __construct(Lexer $lexer, MacroSet $macros) |
|
63
|
|
|
{ |
|
64
|
|
|
$this->lexer = $lexer; |
|
65
|
|
|
$this->macros = $macros; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* @param string $code |
|
70
|
|
|
* @param Context|null $context |
|
71
|
|
|
* |
|
72
|
|
|
* @throws ParseException |
|
73
|
|
|
* @throws \PhpYacc\Exception\LexingException |
|
74
|
|
|
* |
|
75
|
|
|
* @return Context |
|
76
|
|
|
*/ |
|
77
|
|
|
public function parse(string $code, Context $context = null) |
|
78
|
|
|
{ |
|
79
|
|
|
$this->context = $context ?: new Context(); |
|
80
|
|
|
|
|
81
|
|
|
$this->lexer->startLexing($code, $this->context->filename); |
|
82
|
|
|
|
|
83
|
|
|
$this->doDeclaration(); |
|
84
|
|
|
$this->doGrammar(); |
|
85
|
|
|
|
|
86
|
|
|
$this->context->eofToken = $this->eofToken; |
|
87
|
|
|
$this->context->errorToken = $this->errorToken; |
|
88
|
|
|
$this->context->startPrime = $this->startPrime; |
|
89
|
|
|
|
|
90
|
|
|
$this->context->finish(); |
|
91
|
|
|
|
|
92
|
|
|
return $this->context; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
/** |
|
96
|
|
|
* @param array $symbols |
|
97
|
|
|
* @param int $n |
|
98
|
|
|
* @param $delm |
|
99
|
|
|
* @param array $attribute |
|
100
|
|
|
* |
|
101
|
|
|
* @throws ParseException |
|
102
|
|
|
* @throws \PhpYacc\Exception\LexingException |
|
103
|
|
|
* |
|
104
|
|
|
* @return string |
|
105
|
|
|
*/ |
|
106
|
|
|
protected function copyAction(array $symbols, int $n, $delm, array $attribute): string |
|
107
|
|
|
{ |
|
108
|
|
|
$tokens = []; |
|
109
|
|
|
$ct = 0; |
|
110
|
|
|
|
|
111
|
|
|
while (($token = $this->lexer->getRawToken())->getValue() !== $delm || $ct > 0) { |
|
112
|
|
|
switch ($token->getValue()) { |
|
113
|
|
|
case "\0": |
|
114
|
|
|
throw ParseException::unexpected($token, Token::decode($delm)); |
|
115
|
|
|
case '{': |
|
116
|
|
|
$ct++; |
|
117
|
|
|
break; |
|
118
|
|
|
case '}': |
|
119
|
|
|
$ct--; |
|
120
|
|
|
break; |
|
121
|
|
|
} |
|
122
|
|
|
$tokens[] = $token; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
$expanded = $this->macros->apply($this->context, $symbols, $tokens, $n, $attribute); |
|
126
|
|
|
|
|
127
|
|
|
$action = \implode('', \array_map(function (Token $token) { |
|
128
|
|
|
return $token->getValue(); |
|
129
|
|
|
}, $expanded)); |
|
130
|
|
|
|
|
131
|
|
|
return $action; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
/** |
|
135
|
|
|
* @throws ParseException |
|
136
|
|
|
* @throws \PhpYacc\Exception\LexingException |
|
137
|
|
|
*/ |
|
138
|
|
|
protected function doType() |
|
139
|
|
|
{ |
|
140
|
|
|
$type = $this->getType(); |
|
141
|
|
|
while (true) { |
|
142
|
|
|
if (($token = $this->lexer->getToken())->getValue() === ',') { |
|
143
|
|
|
continue; |
|
144
|
|
|
} |
|
145
|
|
|
if ($token->getType() !== Token::T_NAME && $token->getValue()[0] !== "'") { |
|
146
|
|
|
break; |
|
147
|
|
|
} |
|
148
|
|
|
$p = $this->context->internSymbol($token->getValue(), false); |
|
149
|
|
|
if ($type !== null) { |
|
150
|
|
|
$p->type = $type; |
|
151
|
|
|
} |
|
152
|
|
|
} |
|
153
|
|
|
$this->lexer->ungetToken(); |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
/** |
|
157
|
|
|
* @throws ParseException |
|
158
|
|
|
* @throws \PhpYacc\Exception\LexingException |
|
159
|
|
|
*/ |
|
160
|
|
|
protected function doGrammar() |
|
161
|
|
|
{ |
|
162
|
|
|
$attribute = []; |
|
163
|
|
|
$buffer = [null]; |
|
164
|
|
|
$production = new Production('', 0); |
|
165
|
|
|
|
|
166
|
|
|
$production->body = [$this->startPrime]; |
|
167
|
|
|
$this->context->addGram($production); |
|
168
|
|
|
|
|
169
|
|
|
$token = $this->lexer->getToken(); |
|
170
|
|
|
|
|
171
|
|
|
while ($token->getType() !== Token::T_MARK && $token->getType() !== Token::T_EOF) { |
|
172
|
|
|
if ($token->getType() === Token::T_NAME) { |
|
173
|
|
|
if ($this->lexer->peek()->getValue()[0] === '@') { |
|
174
|
|
|
$attribute[0] = $token->getValue(); |
|
175
|
|
|
$this->lexer->getToken(); |
|
176
|
|
|
$token = $this->lexer->getToken(); |
|
177
|
|
|
} else { |
|
178
|
|
|
$attribute[0] = null; |
|
179
|
|
|
} |
|
180
|
|
|
$buffer[0] = $this->context->internSymbol($token->getValue(), false); |
|
181
|
|
|
$attribute[1] = null; |
|
182
|
|
|
if ($buffer[0]->isTerminal) { |
|
183
|
|
|
throw new \RuntimeException("Non terminal symbol expected: $token"); |
|
184
|
|
|
} elseif (($tmp = $this->lexer->getToken())->getType() !== Token::T_COLON) { |
|
185
|
|
|
throw new \RuntimeException("':' expected, $tmp found"); |
|
186
|
|
|
} |
|
187
|
|
|
if ($this->context->startSymbol === null) { |
|
188
|
|
|
$this->context->startSymbol = $buffer[0]; |
|
189
|
|
|
} |
|
190
|
|
|
} elseif ($token->getValue()[0] === '|') { |
|
191
|
|
|
if (!$buffer[0]) { |
|
192
|
|
|
throw new \RuntimeException("Syntax Error, unexpected $token"); |
|
193
|
|
|
} |
|
194
|
|
|
$attribute[1] = null; |
|
195
|
|
|
} elseif ($token->getType() === Token::T_BEGIN_INC) { |
|
196
|
|
|
$this->doCopy(); |
|
197
|
|
|
$token = $this->lexer->getToken(); |
|
198
|
|
|
continue; |
|
199
|
|
|
} else { |
|
200
|
|
|
throw new \RuntimeException("Syntax Error, unexpected $token"); |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
|
|
$lastTerm = $this->startPrime; |
|
204
|
|
|
$action = null; |
|
205
|
|
|
$pos = 0; |
|
206
|
|
|
$i = 1; |
|
207
|
|
|
while (true) { |
|
208
|
|
|
$token = $this->lexer->getToken(); |
|
209
|
|
|
|
|
210
|
|
|
if ($token->getValue()[0] === '=') { |
|
211
|
|
|
$pos = $token->getLine(); |
|
212
|
|
|
if (($token = $this->lexer->getToken())->getValue()[0] === '{') { |
|
213
|
|
|
$pos = $token->getLine(); |
|
214
|
|
|
$action = $this->copyAction($buffer, $i - 1, '}', $attribute); |
|
215
|
|
|
} else { |
|
216
|
|
|
$this->lexer->ungetToken(); |
|
217
|
|
|
$action = $this->copyAction($buffer, $i - 1, ';', $attribute); |
|
218
|
|
|
} |
|
219
|
|
|
} elseif ($token->getValue()[0] === '{') { |
|
220
|
|
|
$pos = $token->getLine(); |
|
221
|
|
|
$action = $this->copyAction($buffer, $i - 1, '}', $attribute); |
|
222
|
|
|
} elseif ($token->getType() === Token::T_PRECTOK) { |
|
223
|
|
|
$lastTerm = $this->context->internSymbol($this->lexer->getToken()->getValue(), false); |
|
224
|
|
|
} elseif ($token->getType() === Token::T_NAME && $this->lexer->peek()->getType() === Token::T_COLON) { |
|
225
|
|
|
break; |
|
226
|
|
|
} elseif ($token->getType() === Token::T_NAME && $this->lexer->peek()->getValue()[0] === '@') { |
|
227
|
|
|
$attribute[$i] = $token->getValue(); |
|
228
|
|
|
$this->lexer->getToken(); |
|
229
|
|
|
} elseif ($token->getType() === Token::T_NAME || $token->getType() === Token::T_STRING) { |
|
230
|
|
|
if ($action) { |
|
|
|
|
|
|
231
|
|
|
$g = $this->context->genNonTerminal(); |
|
232
|
|
|
$production = new Production($action, $pos); |
|
233
|
|
|
$production->body = [$g]; |
|
234
|
|
|
$buffer[$i++] = $g; |
|
235
|
|
|
$attribute[$i] = null; |
|
236
|
|
|
$production->link = $production->body[0]->value; |
|
237
|
|
|
$g->value = $this->context->addGram($production); |
|
238
|
|
|
} |
|
239
|
|
|
$buffer[$i++] = $w = $this->context->internSymbol($token->getValue(), false); |
|
240
|
|
|
$attribute[$i] = null; |
|
241
|
|
|
if ($w->isTerminal) { |
|
242
|
|
|
$lastTerm = $w; |
|
243
|
|
|
} |
|
244
|
|
|
$action = null; |
|
245
|
|
|
} else { |
|
246
|
|
|
break; |
|
247
|
|
|
} |
|
248
|
|
|
} |
|
249
|
|
|
if (!$action) { |
|
|
|
|
|
|
250
|
|
|
if ($i > 1 && $buffer[0]->type !== null && $buffer[0]->type !== $buffer[1]->type) { |
|
251
|
|
|
throw new ParseException('Stack types are different'); |
|
252
|
|
|
} |
|
253
|
|
|
} |
|
254
|
|
|
$production = new Production($action, $pos); |
|
255
|
|
|
|
|
256
|
|
|
$production->body = \array_slice($buffer, 0, $i); |
|
257
|
|
|
$production->precedence = $lastTerm->precedence; |
|
258
|
|
|
$production->associativity = $lastTerm->associativity & Symbol::MASK; |
|
259
|
|
|
$production->link = $production->body[0]->value; |
|
260
|
|
|
$buffer[0]->value = $this->context->addGram($production); |
|
261
|
|
|
|
|
262
|
|
|
if ($token->getType() === Token::T_SEMICOLON) { |
|
263
|
|
|
$token = $this->lexer->getToken(); |
|
264
|
|
|
} |
|
265
|
|
|
} |
|
266
|
|
|
|
|
267
|
|
|
$this->context->gram(0)->body[] = $this->context->startSymbol; |
|
268
|
|
|
$this->startPrime->value = null; |
|
269
|
|
|
foreach ($this->context->nonterminals as $key => $symbol) { |
|
270
|
|
|
if ($symbol === $this->startPrime) { |
|
271
|
|
|
continue; |
|
272
|
|
|
} |
|
273
|
|
|
if (($j = $symbol->value) === null) { |
|
274
|
|
|
throw new ParseException("Non terminal {$symbol->name} used, but not defined"); |
|
275
|
|
|
} |
|
276
|
|
|
$k = null; |
|
277
|
|
|
while ($j) { |
|
278
|
|
|
$w = $j->link; |
|
279
|
|
|
$j->link = $k; |
|
280
|
|
|
$k = $j; |
|
281
|
|
|
$j = $w; |
|
282
|
|
|
} |
|
283
|
|
|
$symbol->value = $k; |
|
284
|
|
|
} |
|
285
|
|
|
} |
|
286
|
|
|
|
|
287
|
|
|
/** |
|
288
|
|
|
* @throws ParseException |
|
289
|
|
|
* @throws \PhpYacc\Exception\LexingException |
|
290
|
|
|
*/ |
|
291
|
|
|
protected function doDeclaration() |
|
292
|
|
|
{ |
|
293
|
|
|
$this->eofToken = $this->context->internSymbol('EOF', true); |
|
294
|
|
|
$this->eofToken->value = 0; |
|
295
|
|
|
$this->errorToken = $this->context->internSymbol('error', true); |
|
296
|
|
|
$this->startPrime = $this->context->internSymbol('$start', false); |
|
297
|
|
|
|
|
298
|
|
|
while (($token = $this->lexer->getToken())->getType() !== Token::T_MARK) { |
|
299
|
|
|
switch ($token->getType()) { |
|
300
|
|
|
case Token::T_TOKEN: |
|
301
|
|
|
case Token::T_RIGHT: |
|
302
|
|
|
case Token::T_LEFT: |
|
303
|
|
|
case Token::T_NON_ASSOC: |
|
304
|
|
|
$this->doToken($token); |
|
305
|
|
|
break; |
|
306
|
|
|
|
|
307
|
|
|
case Token::T_BEGIN_INC: |
|
308
|
|
|
$this->doCopy(); |
|
309
|
|
|
break; |
|
310
|
|
|
|
|
311
|
|
|
case Token::T_UNION: |
|
312
|
|
|
$this->doUnion(); |
|
313
|
|
|
$this->context->unioned = true; |
|
314
|
|
|
break; |
|
315
|
|
|
|
|
316
|
|
|
case Token::T_TYPE: |
|
317
|
|
|
$this->doType(); |
|
318
|
|
|
break; |
|
319
|
|
|
|
|
320
|
|
|
case Token::T_EXPECT: |
|
321
|
|
|
$token = $this->lexer->getToken(); |
|
322
|
|
|
if ($token->getType() === Token::T_NUMBER) { |
|
323
|
|
|
$this->context->expected = (int) $token->getValue(); |
|
324
|
|
|
} else { |
|
325
|
|
|
throw ParseException::unexpected($token, Token::T_NUMBER); |
|
326
|
|
|
} |
|
327
|
|
|
break; |
|
328
|
|
|
|
|
329
|
|
|
case Token::T_START: |
|
330
|
|
|
$token = $this->lexer->getToken(); |
|
331
|
|
|
$this->context->startSymbol = $this->context->internSymbol($token->getValue(), false); |
|
332
|
|
|
break; |
|
333
|
|
|
|
|
334
|
|
|
case Token::T_PURE_PARSER: |
|
335
|
|
|
$this->context->pureFlag = true; |
|
336
|
|
|
break; |
|
337
|
|
|
|
|
338
|
|
|
case Token::T_EOF: |
|
339
|
|
|
throw new ParseException('No grammar given'); |
|
340
|
|
|
default: |
|
341
|
|
|
throw new ParseException("Syntax error, unexpected {$token->getValue()}"); |
|
342
|
|
|
} |
|
343
|
|
|
} |
|
344
|
|
|
|
|
345
|
|
|
$base = 256; |
|
346
|
|
|
foreach ($this->context->terminals as $terminal) { |
|
347
|
|
|
if ($terminal === $this->context->eofToken) { |
|
348
|
|
|
continue; |
|
349
|
|
|
} |
|
350
|
|
|
if ($terminal->value < 0) { |
|
351
|
|
|
$terminal->value = $base++; |
|
352
|
|
|
} |
|
353
|
|
|
} |
|
354
|
|
|
} |
|
355
|
|
|
|
|
356
|
|
|
/** |
|
357
|
|
|
* @param Token $tag |
|
358
|
|
|
* |
|
359
|
|
|
* @throws ParseException |
|
360
|
|
|
* @throws \PhpYacc\Exception\LexingException |
|
361
|
|
|
*/ |
|
362
|
|
|
protected function doToken(Token $tag) |
|
363
|
|
|
{ |
|
364
|
|
|
$preIncr = 0; |
|
365
|
|
|
$type = $this->getType(); |
|
366
|
|
|
$token = $this->lexer->getToken(); |
|
367
|
|
|
|
|
368
|
|
|
while ($token->getType() === Token::T_NAME || $token->getType() === Token::T_STRING) { |
|
369
|
|
|
$p = $this->context->internSymbol($token->getValue(), true); |
|
370
|
|
|
|
|
371
|
|
|
if ($type) { |
|
372
|
|
|
$p->type = $type; |
|
373
|
|
|
} |
|
374
|
|
|
|
|
375
|
|
|
switch ($tag->getType()) { |
|
376
|
|
|
case Token::T_LEFT: |
|
377
|
|
|
$p->associativity |= Symbol::LEFT; |
|
378
|
|
|
break; |
|
379
|
|
|
case Token::T_RIGHT: |
|
380
|
|
|
$p->associativity |= Symbol::RIGHT; |
|
381
|
|
|
break; |
|
382
|
|
|
case Token::T_NON_ASSOC: |
|
383
|
|
|
$p->associativity |= Symbol::NON; |
|
384
|
|
|
break; |
|
385
|
|
|
} |
|
386
|
|
|
|
|
387
|
|
|
if ($tag->getType() !== Token::T_TOKEN) { |
|
388
|
|
|
$p->precedence = $this->currentPrecedence; |
|
389
|
|
|
$preIncr = 1; |
|
390
|
|
|
} |
|
391
|
|
|
|
|
392
|
|
|
$token = $this->lexer->getToken(); |
|
393
|
|
|
if ($token->getType() === Token::T_NUMBER) { |
|
394
|
|
|
if ($p->value === null) { |
|
395
|
|
|
$p->value = (int) $token->getValue(); |
|
396
|
|
|
} else { |
|
397
|
|
|
throw new ParseException( |
|
398
|
|
|
sprintf('Unexpected Token::NUMBER as %s already has a value', $p->name) |
|
399
|
|
|
); |
|
400
|
|
|
} |
|
401
|
|
|
$token = $this->lexer->getToken(); |
|
402
|
|
|
} |
|
403
|
|
|
|
|
404
|
|
|
if ($token->getType() === Token::T_COMMA) { |
|
405
|
|
|
$token = $this->lexer->getToken(); |
|
406
|
|
|
} |
|
407
|
|
|
} |
|
408
|
|
|
|
|
409
|
|
|
$this->lexer->ungetToken(); |
|
410
|
|
|
$this->currentPrecedence += $preIncr; |
|
411
|
|
|
} |
|
412
|
|
|
|
|
413
|
|
|
/** |
|
414
|
|
|
* @throws ParseException |
|
415
|
|
|
* @throws \PhpYacc\Exception\LexingException |
|
416
|
|
|
* |
|
417
|
|
|
* @return null|Symbol |
|
418
|
|
|
*/ |
|
419
|
|
|
protected function getType() |
|
420
|
|
|
{ |
|
421
|
|
|
$token = $this->lexer->getToken(); |
|
422
|
|
|
|
|
423
|
|
|
if ($token->getValue()[0] !== '<') { |
|
424
|
|
|
$this->lexer->ungetToken(); |
|
425
|
|
|
|
|
426
|
|
|
return; |
|
427
|
|
|
} |
|
428
|
|
|
|
|
429
|
|
|
$ct = 1; |
|
430
|
|
|
$p = ''; |
|
431
|
|
|
$token = $this->lexer->getToken(); |
|
432
|
|
|
|
|
433
|
|
|
while (true) { |
|
434
|
|
|
switch ($token->getValue()[0]) { |
|
435
|
|
|
case "\n": |
|
436
|
|
|
case "\0": |
|
437
|
|
|
throw ParseException::unexpected($token, '>'); |
|
438
|
|
|
case '<': |
|
439
|
|
|
$ct++; |
|
440
|
|
|
break; |
|
441
|
|
|
case '>': |
|
442
|
|
|
$ct--; |
|
443
|
|
|
break; |
|
444
|
|
|
} |
|
445
|
|
|
|
|
446
|
|
|
if ($ct === 0) { |
|
447
|
|
|
break; |
|
448
|
|
|
} |
|
449
|
|
|
|
|
450
|
|
|
$p .= $token->getValue(); |
|
451
|
|
|
$token = $this->lexer->getRawToken(); |
|
452
|
|
|
} |
|
453
|
|
|
$this->context->unioned = true; |
|
454
|
|
|
|
|
455
|
|
|
return $this->context->intern($p); |
|
456
|
|
|
} |
|
457
|
|
|
|
|
458
|
|
|
/** |
|
459
|
|
|
* @return void |
|
460
|
|
|
*/ |
|
461
|
|
|
protected function doCopy() |
|
462
|
|
|
{ |
|
463
|
|
|
// TODO |
|
464
|
|
|
} |
|
465
|
|
|
|
|
466
|
|
|
/** |
|
467
|
|
|
* @return void |
|
468
|
|
|
*/ |
|
469
|
|
|
protected function doUnion() |
|
470
|
|
|
{ |
|
471
|
|
|
// TODO |
|
472
|
|
|
} |
|
473
|
|
|
} |
|
474
|
|
|
|
In PHP, under loose comparison (like
==, or!=, orswitchconditions), values of different types might be equal.For
stringvalues, the empty string''is a special case, in particular the following results might be unexpected: