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