1: <?php declare(strict_types=1);
2:
3: namespace PhpParser;
4:
5: /*
6: * This parser is based on a skeleton written by Moriyoshi Koizumi, which in
7: * turn is based on work by Masato Bito.
8: */
9: use PhpParser\Node\Expr;
10: use PhpParser\Node\Expr\Cast\Double;
11: use PhpParser\Node\Name;
12: use PhpParser\Node\Param;
13: use PhpParser\Node\Scalar\Encapsed;
14: use PhpParser\Node\Scalar\LNumber;
15: use PhpParser\Node\Scalar\String_;
16: use PhpParser\Node\Stmt\Class_;
17: use PhpParser\Node\Stmt\ClassConst;
18: use PhpParser\Node\Stmt\ClassMethod;
19: use PhpParser\Node\Stmt\Else_;
20: use PhpParser\Node\Stmt\ElseIf_;
21: use PhpParser\Node\Stmt\Enum_;
22: use PhpParser\Node\Stmt\Interface_;
23: use PhpParser\Node\Stmt\Namespace_;
24: use PhpParser\Node\Stmt\Nop;
25: use PhpParser\Node\Stmt\Property;
26: use PhpParser\Node\Stmt\TryCatch;
27: use PhpParser\Node\Stmt\UseUse;
28: use PhpParser\Node\VarLikeIdentifier;
29:
30: abstract class ParserAbstract implements Parser
31: {
32: const SYMBOL_NONE = -1;
33:
34: /*
35: * The following members will be filled with generated parsing data:
36: */
37:
38: /** @var int Size of $tokenToSymbol map */
39: protected $tokenToSymbolMapSize;
40: /** @var int Size of $action table */
41: protected $actionTableSize;
42: /** @var int Size of $goto table */
43: protected $gotoTableSize;
44:
45: /** @var int Symbol number signifying an invalid token */
46: protected $invalidSymbol;
47: /** @var int Symbol number of error recovery token */
48: protected $errorSymbol;
49: /** @var int Action number signifying default action */
50: protected $defaultAction;
51: /** @var int Rule number signifying that an unexpected token was encountered */
52: protected $unexpectedTokenRule;
53:
54: protected $YY2TBLSTATE;
55: /** @var int Number of non-leaf states */
56: protected $numNonLeafStates;
57:
58: /** @var int[] Map of lexer tokens to internal symbols */
59: protected $tokenToSymbol;
60: /** @var string[] Map of symbols to their names */
61: protected $symbolToName;
62: /** @var array Names of the production rules (only necessary for debugging) */
63: protected $productions;
64:
65: /** @var int[] Map of states to a displacement into the $action table. The corresponding action for this
66: * state/symbol pair is $action[$actionBase[$state] + $symbol]. If $actionBase[$state] is 0, the
67: * action is defaulted, i.e. $actionDefault[$state] should be used instead. */
68: protected $actionBase;
69: /** @var int[] Table of actions. Indexed according to $actionBase comment. */
70: protected $action;
71: /** @var int[] Table indexed analogously to $action. If $actionCheck[$actionBase[$state] + $symbol] != $symbol
72: * then the action is defaulted, i.e. $actionDefault[$state] should be used instead. */
73: protected $actionCheck;
74: /** @var int[] Map of states to their default action */
75: protected $actionDefault;
76: /** @var callable[] Semantic action callbacks */
77: protected $reduceCallbacks;
78:
79: /** @var int[] Map of non-terminals to a displacement into the $goto table. The corresponding goto state for this
80: * non-terminal/state pair is $goto[$gotoBase[$nonTerminal] + $state] (unless defaulted) */
81: protected $gotoBase;
82: /** @var int[] Table of states to goto after reduction. Indexed according to $gotoBase comment. */
83: protected $goto;
84: /** @var int[] Table indexed analogously to $goto. If $gotoCheck[$gotoBase[$nonTerminal] + $state] != $nonTerminal
85: * then the goto state is defaulted, i.e. $gotoDefault[$nonTerminal] should be used. */
86: protected $gotoCheck;
87: /** @var int[] Map of non-terminals to the default state to goto after their reduction */
88: protected $gotoDefault;
89:
90: /** @var int[] Map of rules to the non-terminal on their left-hand side, i.e. the non-terminal to use for
91: * determining the state to goto after reduction. */
92: protected $ruleToNonTerminal;
93: /** @var int[] Map of rules to the length of their right-hand side, which is the number of elements that have to
94: * be popped from the stack(s) on reduction. */
95: protected $ruleToLength;
96:
97: /*
98: * The following members are part of the parser state:
99: */
100:
101: /** @var Lexer Lexer that is used when parsing */
102: protected $lexer;
103: /** @var mixed Temporary value containing the result of last semantic action (reduction) */
104: protected $semValue;
105: /** @var array Semantic value stack (contains values of tokens and semantic action results) */
106: protected $semStack;
107: /** @var array[] Start attribute stack */
108: protected $startAttributeStack;
109: /** @var array[] End attribute stack */
110: protected $endAttributeStack;
111: /** @var array End attributes of last *shifted* token */
112: protected $endAttributes;
113: /** @var array Start attributes of last *read* token */
114: protected $lookaheadStartAttributes;
115:
116: /** @var ErrorHandler Error handler */
117: protected $errorHandler;
118: /** @var int Error state, used to avoid error floods */
119: protected $errorState;
120:
121: /**
122: * Initialize $reduceCallbacks map.
123: */
124: abstract protected function initReduceCallbacks();
125:
126: /**
127: * Creates a parser instance.
128: *
129: * Options: Currently none.
130: *
131: * @param Lexer $lexer A lexer
132: * @param array $options Options array.
133: */
134: public function __construct(Lexer $lexer, array $options = []) {
135: $this->lexer = $lexer;
136:
137: if (isset($options['throwOnError'])) {
138: throw new \LogicException(
139: '"throwOnError" is no longer supported, use "errorHandler" instead');
140: }
141:
142: $this->initReduceCallbacks();
143: }
144:
145: /**
146: * Parses PHP code into a node tree.
147: *
148: * If a non-throwing error handler is used, the parser will continue parsing after an error
149: * occurred and attempt to build a partial AST.
150: *
151: * @param string $code The source code to parse
152: * @param ErrorHandler|null $errorHandler Error handler to use for lexer/parser errors, defaults
153: * to ErrorHandler\Throwing.
154: *
155: * @return Node\Stmt[]|null Array of statements (or null non-throwing error handler is used and
156: * the parser was unable to recover from an error).
157: */
158: public function parse(string $code, ?ErrorHandler $errorHandler = null) {
159: $this->errorHandler = $errorHandler ?: new ErrorHandler\Throwing;
160:
161: $this->lexer->startLexing($code, $this->errorHandler);
162: $result = $this->doParse();
163:
164: // Clear out some of the interior state, so we don't hold onto unnecessary
165: // memory between uses of the parser
166: $this->startAttributeStack = [];
167: $this->endAttributeStack = [];
168: $this->semStack = [];
169: $this->semValue = null;
170:
171: return $result;
172: }
173:
174: protected function doParse() {
175: // We start off with no lookahead-token
176: $symbol = self::SYMBOL_NONE;
177:
178: // The attributes for a node are taken from the first and last token of the node.
179: // From the first token only the startAttributes are taken and from the last only
180: // the endAttributes. Both are merged using the array union operator (+).
181: $startAttributes = [];
182: $endAttributes = [];
183: $this->endAttributes = $endAttributes;
184:
185: // Keep stack of start and end attributes
186: $this->startAttributeStack = [];
187: $this->endAttributeStack = [$endAttributes];
188:
189: // Start off in the initial state and keep a stack of previous states
190: $state = 0;
191: $stateStack = [$state];
192:
193: // Semantic value stack (contains values of tokens and semantic action results)
194: $this->semStack = [];
195:
196: // Current position in the stack(s)
197: $stackPos = 0;
198:
199: $this->errorState = 0;
200:
201: for (;;) {
202: //$this->traceNewState($state, $symbol);
203:
204: if ($this->actionBase[$state] === 0) {
205: $rule = $this->actionDefault[$state];
206: } else {
207: if ($symbol === self::SYMBOL_NONE) {
208: // Fetch the next token id from the lexer and fetch additional info by-ref.
209: // The end attributes are fetched into a temporary variable and only set once the token is really
210: // shifted (not during read). Otherwise you would sometimes get off-by-one errors, when a rule is
211: // reduced after a token was read but not yet shifted.
212: $tokenId = $this->lexer->getNextToken($tokenValue, $startAttributes, $endAttributes);
213:
214: // map the lexer token id to the internally used symbols
215: $symbol = $tokenId >= 0 && $tokenId < $this->tokenToSymbolMapSize
216: ? $this->tokenToSymbol[$tokenId]
217: : $this->invalidSymbol;
218:
219: if ($symbol === $this->invalidSymbol) {
220: throw new \RangeException(sprintf(
221: 'The lexer returned an invalid token (id=%d, value=%s)',
222: $tokenId, $tokenValue
223: ));
224: }
225:
226: // Allow productions to access the start attributes of the lookahead token.
227: $this->lookaheadStartAttributes = $startAttributes;
228:
229: //$this->traceRead($symbol);
230: }
231:
232: $idx = $this->actionBase[$state] + $symbol;
233: if ((($idx >= 0 && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol)
234: || ($state < $this->YY2TBLSTATE
235: && ($idx = $this->actionBase[$state + $this->numNonLeafStates] + $symbol) >= 0
236: && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol))
237: && ($action = $this->action[$idx]) !== $this->defaultAction) {
238: /*
239: * >= numNonLeafStates: shift and reduce
240: * > 0: shift
241: * = 0: accept
242: * < 0: reduce
243: * = -YYUNEXPECTED: error
244: */
245: if ($action > 0) {
246: /* shift */
247: //$this->traceShift($symbol);
248:
249: ++$stackPos;
250: $stateStack[$stackPos] = $state = $action;
251: $this->semStack[$stackPos] = $tokenValue;
252: $this->startAttributeStack[$stackPos] = $startAttributes;
253: $this->endAttributeStack[$stackPos] = $endAttributes;
254: $this->endAttributes = $endAttributes;
255: $symbol = self::SYMBOL_NONE;
256:
257: if ($this->errorState) {
258: --$this->errorState;
259: }
260:
261: if ($action < $this->numNonLeafStates) {
262: continue;
263: }
264:
265: /* $yyn >= numNonLeafStates means shift-and-reduce */
266: $rule = $action - $this->numNonLeafStates;
267: } else {
268: $rule = -$action;
269: }
270: } else {
271: $rule = $this->actionDefault[$state];
272: }
273: }
274:
275: for (;;) {
276: if ($rule === 0) {
277: /* accept */
278: //$this->traceAccept();
279: return $this->semValue;
280: } elseif ($rule !== $this->unexpectedTokenRule) {
281: /* reduce */
282: //$this->traceReduce($rule);
283:
284: try {
285: $this->reduceCallbacks[$rule]($stackPos);
286: } catch (Error $e) {
287: if (-1 === $e->getStartLine() && isset($startAttributes['startLine'])) {
288: $e->setStartLine($startAttributes['startLine']);
289: }
290:
291: $this->emitError($e);
292: // Can't recover from this type of error
293: return null;
294: }
295:
296: /* Goto - shift nonterminal */
297: $lastEndAttributes = $this->endAttributeStack[$stackPos];
298: $ruleLength = $this->ruleToLength[$rule];
299: $stackPos -= $ruleLength;
300: $nonTerminal = $this->ruleToNonTerminal[$rule];
301: $idx = $this->gotoBase[$nonTerminal] + $stateStack[$stackPos];
302: if ($idx >= 0 && $idx < $this->gotoTableSize && $this->gotoCheck[$idx] === $nonTerminal) {
303: $state = $this->goto[$idx];
304: } else {
305: $state = $this->gotoDefault[$nonTerminal];
306: }
307:
308: ++$stackPos;
309: $stateStack[$stackPos] = $state;
310: $this->semStack[$stackPos] = $this->semValue;
311: $this->endAttributeStack[$stackPos] = $lastEndAttributes;
312: if ($ruleLength === 0) {
313: // Empty productions use the start attributes of the lookahead token.
314: $this->startAttributeStack[$stackPos] = $this->lookaheadStartAttributes;
315: }
316: } else {
317: /* error */
318: switch ($this->errorState) {
319: case 0:
320: $msg = $this->getErrorMessage($symbol, $state);
321: $this->emitError(new Error($msg, $startAttributes + $endAttributes));
322: // Break missing intentionally
323: case 1:
324: case 2:
325: $this->errorState = 3;
326:
327: // Pop until error-expecting state uncovered
328: while (!(
329: (($idx = $this->actionBase[$state] + $this->errorSymbol) >= 0
330: && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $this->errorSymbol)
331: || ($state < $this->YY2TBLSTATE
332: && ($idx = $this->actionBase[$state + $this->numNonLeafStates] + $this->errorSymbol) >= 0
333: && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $this->errorSymbol)
334: ) || ($action = $this->action[$idx]) === $this->defaultAction) { // Not totally sure about this
335: if ($stackPos <= 0) {
336: // Could not recover from error
337: return null;
338: }
339: $state = $stateStack[--$stackPos];
340: //$this->tracePop($state);
341: }
342:
343: //$this->traceShift($this->errorSymbol);
344: ++$stackPos;
345: $stateStack[$stackPos] = $state = $action;
346:
347: // We treat the error symbol as being empty, so we reset the end attributes
348: // to the end attributes of the last non-error symbol
349: $this->startAttributeStack[$stackPos] = $this->lookaheadStartAttributes;
350: $this->endAttributeStack[$stackPos] = $this->endAttributeStack[$stackPos - 1];
351: $this->endAttributes = $this->endAttributeStack[$stackPos - 1];
352: break;
353:
354: case 3:
355: if ($symbol === 0) {
356: // Reached EOF without recovering from error
357: return null;
358: }
359:
360: //$this->traceDiscard($symbol);
361: $symbol = self::SYMBOL_NONE;
362: break 2;
363: }
364: }
365:
366: if ($state < $this->numNonLeafStates) {
367: break;
368: }
369:
370: /* >= numNonLeafStates means shift-and-reduce */
371: $rule = $state - $this->numNonLeafStates;
372: }
373: }
374:
375: throw new \RuntimeException('Reached end of parser loop');
376: }
377:
378: protected function emitError(Error $error) {
379: $this->errorHandler->handleError($error);
380: }
381:
382: /**
383: * Format error message including expected tokens.
384: *
385: * @param int $symbol Unexpected symbol
386: * @param int $state State at time of error
387: *
388: * @return string Formatted error message
389: */
390: protected function getErrorMessage(int $symbol, int $state) : string {
391: $expectedString = '';
392: if ($expected = $this->getExpectedTokens($state)) {
393: $expectedString = ', expecting ' . implode(' or ', $expected);
394: }
395:
396: return 'Syntax error, unexpected ' . $this->symbolToName[$symbol] . $expectedString;
397: }
398:
399: /**
400: * Get limited number of expected tokens in given state.
401: *
402: * @param int $state State
403: *
404: * @return string[] Expected tokens. If too many, an empty array is returned.
405: */
406: protected function getExpectedTokens(int $state) : array {
407: $expected = [];
408:
409: $base = $this->actionBase[$state];
410: foreach ($this->symbolToName as $symbol => $name) {
411: $idx = $base + $symbol;
412: if ($idx >= 0 && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol
413: || $state < $this->YY2TBLSTATE
414: && ($idx = $this->actionBase[$state + $this->numNonLeafStates] + $symbol) >= 0
415: && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol
416: ) {
417: if ($this->action[$idx] !== $this->unexpectedTokenRule
418: && $this->action[$idx] !== $this->defaultAction
419: && $symbol !== $this->errorSymbol
420: ) {
421: if (count($expected) === 4) {
422: /* Too many expected tokens */
423: return [];
424: }
425:
426: $expected[] = $name;
427: }
428: }
429: }
430:
431: return $expected;
432: }
433:
434: /*
435: * Tracing functions used for debugging the parser.
436: */
437:
438: /*
439: protected function traceNewState($state, $symbol) {
440: echo '% State ' . $state
441: . ', Lookahead ' . ($symbol == self::SYMBOL_NONE ? '--none--' : $this->symbolToName[$symbol]) . "\n";
442: }
443:
444: protected function traceRead($symbol) {
445: echo '% Reading ' . $this->symbolToName[$symbol] . "\n";
446: }
447:
448: protected function traceShift($symbol) {
449: echo '% Shift ' . $this->symbolToName[$symbol] . "\n";
450: }
451:
452: protected function traceAccept() {
453: echo "% Accepted.\n";
454: }
455:
456: protected function traceReduce($n) {
457: echo '% Reduce by (' . $n . ') ' . $this->productions[$n] . "\n";
458: }
459:
460: protected function tracePop($state) {
461: echo '% Recovering, uncovered state ' . $state . "\n";
462: }
463:
464: protected function traceDiscard($symbol) {
465: echo '% Discard ' . $this->symbolToName[$symbol] . "\n";
466: }
467: */
468:
469: /*
470: * Helper functions invoked by semantic actions
471: */
472:
473: /**
474: * Moves statements of semicolon-style namespaces into $ns->stmts and checks various error conditions.
475: *
476: * @param Node\Stmt[] $stmts
477: * @return Node\Stmt[]
478: */
479: protected function handleNamespaces(array $stmts) : array {
480: $hasErrored = false;
481: $style = $this->getNamespacingStyle($stmts);
482: if (null === $style) {
483: // not namespaced, nothing to do
484: return $stmts;
485: } elseif ('brace' === $style) {
486: // For braced namespaces we only have to check that there are no invalid statements between the namespaces
487: $afterFirstNamespace = false;
488: foreach ($stmts as $stmt) {
489: if ($stmt instanceof Node\Stmt\Namespace_) {
490: $afterFirstNamespace = true;
491: } elseif (!$stmt instanceof Node\Stmt\HaltCompiler
492: && !$stmt instanceof Node\Stmt\Nop
493: && $afterFirstNamespace && !$hasErrored) {
494: $this->emitError(new Error(
495: 'No code may exist outside of namespace {}', $stmt->getAttributes()));
496: $hasErrored = true; // Avoid one error for every statement
497: }
498: }
499: return $stmts;
500: } else {
501: // For semicolon namespaces we have to move the statements after a namespace declaration into ->stmts
502: $resultStmts = [];
503: $targetStmts =& $resultStmts;
504: $lastNs = null;
505: foreach ($stmts as $stmt) {
506: if ($stmt instanceof Node\Stmt\Namespace_) {
507: if ($lastNs !== null) {
508: $this->fixupNamespaceAttributes($lastNs);
509: }
510: if ($stmt->stmts === null) {
511: $stmt->stmts = [];
512: $targetStmts =& $stmt->stmts;
513: $resultStmts[] = $stmt;
514: } else {
515: // This handles the invalid case of mixed style namespaces
516: $resultStmts[] = $stmt;
517: $targetStmts =& $resultStmts;
518: }
519: $lastNs = $stmt;
520: } elseif ($stmt instanceof Node\Stmt\HaltCompiler) {
521: // __halt_compiler() is not moved into the namespace
522: $resultStmts[] = $stmt;
523: } else {
524: $targetStmts[] = $stmt;
525: }
526: }
527: if ($lastNs !== null) {
528: $this->fixupNamespaceAttributes($lastNs);
529: }
530: return $resultStmts;
531: }
532: }
533:
534: private function fixupNamespaceAttributes(Node\Stmt\Namespace_ $stmt) {
535: // We moved the statements into the namespace node, as such the end of the namespace node
536: // needs to be extended to the end of the statements.
537: if (empty($stmt->stmts)) {
538: return;
539: }
540:
541: // We only move the builtin end attributes here. This is the best we can do with the
542: // knowledge we have.
543: $endAttributes = ['endLine', 'endFilePos', 'endTokenPos'];
544: $lastStmt = $stmt->stmts[count($stmt->stmts) - 1];
545: foreach ($endAttributes as $endAttribute) {
546: if ($lastStmt->hasAttribute($endAttribute)) {
547: $stmt->setAttribute($endAttribute, $lastStmt->getAttribute($endAttribute));
548: }
549: }
550: }
551:
552: /**
553: * Determine namespacing style (semicolon or brace)
554: *
555: * @param Node[] $stmts Top-level statements.
556: *
557: * @return null|string One of "semicolon", "brace" or null (no namespaces)
558: */
559: private function getNamespacingStyle(array $stmts) {
560: $style = null;
561: $hasNotAllowedStmts = false;
562: foreach ($stmts as $i => $stmt) {
563: if ($stmt instanceof Node\Stmt\Namespace_) {
564: $currentStyle = null === $stmt->stmts ? 'semicolon' : 'brace';
565: if (null === $style) {
566: $style = $currentStyle;
567: if ($hasNotAllowedStmts) {
568: $this->emitError(new Error(
569: 'Namespace declaration statement has to be the very first statement in the script',
570: $stmt->getLine() // Avoid marking the entire namespace as an error
571: ));
572: }
573: } elseif ($style !== $currentStyle) {
574: $this->emitError(new Error(
575: 'Cannot mix bracketed namespace declarations with unbracketed namespace declarations',
576: $stmt->getLine() // Avoid marking the entire namespace as an error
577: ));
578: // Treat like semicolon style for namespace normalization
579: return 'semicolon';
580: }
581: continue;
582: }
583:
584: /* declare(), __halt_compiler() and nops can be used before a namespace declaration */
585: if ($stmt instanceof Node\Stmt\Declare_
586: || $stmt instanceof Node\Stmt\HaltCompiler
587: || $stmt instanceof Node\Stmt\Nop) {
588: continue;
589: }
590:
591: /* There may be a hashbang line at the very start of the file */
592: if ($i === 0 && $stmt instanceof Node\Stmt\InlineHTML && preg_match('/\A#!.*\r?\n\z/', $stmt->value)) {
593: continue;
594: }
595:
596: /* Everything else if forbidden before namespace declarations */
597: $hasNotAllowedStmts = true;
598: }
599: return $style;
600: }
601:
602: /**
603: * Fix up parsing of static property calls in PHP 5.
604: *
605: * In PHP 5 A::$b[c][d] and A::$b[c][d]() have very different interpretation. The former is
606: * interpreted as (A::$b)[c][d], while the latter is the same as A::{$b[c][d]}(). We parse the
607: * latter as the former initially and this method fixes the AST into the correct form when we
608: * encounter the "()".
609: *
610: * @param Node\Expr\StaticPropertyFetch|Node\Expr\ArrayDimFetch $prop
611: * @param Node\Arg[] $args
612: * @param array $attributes
613: *
614: * @return Expr\StaticCall
615: */
616: protected function fixupPhp5StaticPropCall($prop, array $args, array $attributes) : Expr\StaticCall {
617: if ($prop instanceof Node\Expr\StaticPropertyFetch) {
618: $name = $prop->name instanceof VarLikeIdentifier
619: ? $prop->name->toString() : $prop->name;
620: $var = new Expr\Variable($name, $prop->name->getAttributes());
621: return new Expr\StaticCall($prop->class, $var, $args, $attributes);
622: } elseif ($prop instanceof Node\Expr\ArrayDimFetch) {
623: $tmp = $prop;
624: while ($tmp->var instanceof Node\Expr\ArrayDimFetch) {
625: $tmp = $tmp->var;
626: }
627:
628: /** @var Expr\StaticPropertyFetch $staticProp */
629: $staticProp = $tmp->var;
630:
631: // Set start attributes to attributes of innermost node
632: $tmp = $prop;
633: $this->fixupStartAttributes($tmp, $staticProp->name);
634: while ($tmp->var instanceof Node\Expr\ArrayDimFetch) {
635: $tmp = $tmp->var;
636: $this->fixupStartAttributes($tmp, $staticProp->name);
637: }
638:
639: $name = $staticProp->name instanceof VarLikeIdentifier
640: ? $staticProp->name->toString() : $staticProp->name;
641: $tmp->var = new Expr\Variable($name, $staticProp->name->getAttributes());
642: return new Expr\StaticCall($staticProp->class, $prop, $args, $attributes);
643: } else {
644: throw new \Exception;
645: }
646: }
647:
648: protected function fixupStartAttributes(Node $to, Node $from) {
649: $startAttributes = ['startLine', 'startFilePos', 'startTokenPos'];
650: foreach ($startAttributes as $startAttribute) {
651: if ($from->hasAttribute($startAttribute)) {
652: $to->setAttribute($startAttribute, $from->getAttribute($startAttribute));
653: }
654: }
655: }
656:
657: protected function handleBuiltinTypes(Name $name) {
658: $builtinTypes = [
659: 'bool' => true,
660: 'int' => true,
661: 'float' => true,
662: 'string' => true,
663: 'iterable' => true,
664: 'void' => true,
665: 'object' => true,
666: 'null' => true,
667: 'false' => true,
668: 'mixed' => true,
669: 'never' => true,
670: 'true' => true,
671: ];
672:
673: if (!$name->isUnqualified()) {
674: return $name;
675: }
676:
677: $lowerName = $name->toLowerString();
678: if (!isset($builtinTypes[$lowerName])) {
679: return $name;
680: }
681:
682: return new Node\Identifier($lowerName, $name->getAttributes());
683: }
684:
685: /**
686: * Get combined start and end attributes at a stack location
687: *
688: * @param int $pos Stack location
689: *
690: * @return array Combined start and end attributes
691: */
692: protected function getAttributesAt(int $pos) : array {
693: return $this->startAttributeStack[$pos] + $this->endAttributeStack[$pos];
694: }
695:
696: protected function getFloatCastKind(string $cast): int
697: {
698: $cast = strtolower($cast);
699: if (strpos($cast, 'float') !== false) {
700: return Double::KIND_FLOAT;
701: }
702:
703: if (strpos($cast, 'real') !== false) {
704: return Double::KIND_REAL;
705: }
706:
707: return Double::KIND_DOUBLE;
708: }
709:
710: protected function parseLNumber($str, $attributes, $allowInvalidOctal = false) {
711: try {
712: return LNumber::fromString($str, $attributes, $allowInvalidOctal);
713: } catch (Error $error) {
714: $this->emitError($error);
715: // Use dummy value
716: return new LNumber(0, $attributes);
717: }
718: }
719:
720: /**
721: * Parse a T_NUM_STRING token into either an integer or string node.
722: *
723: * @param string $str Number string
724: * @param array $attributes Attributes
725: *
726: * @return LNumber|String_ Integer or string node.
727: */
728: protected function parseNumString(string $str, array $attributes) {
729: if (!preg_match('/^(?:0|-?[1-9][0-9]*)$/', $str)) {
730: return new String_($str, $attributes);
731: }
732:
733: $num = +$str;
734: if (!is_int($num)) {
735: return new String_($str, $attributes);
736: }
737:
738: return new LNumber($num, $attributes);
739: }
740:
741: protected function stripIndentation(
742: string $string, int $indentLen, string $indentChar,
743: bool $newlineAtStart, bool $newlineAtEnd, array $attributes
744: ) {
745: if ($indentLen === 0) {
746: return $string;
747: }
748:
749: $start = $newlineAtStart ? '(?:(?<=\n)|\A)' : '(?<=\n)';
750: $end = $newlineAtEnd ? '(?:(?=[\r\n])|\z)' : '(?=[\r\n])';
751: $regex = '/' . $start . '([ \t]*)(' . $end . ')?/';
752: return preg_replace_callback(
753: $regex,
754: function ($matches) use ($indentLen, $indentChar, $attributes) {
755: $prefix = substr($matches[1], 0, $indentLen);
756: if (false !== strpos($prefix, $indentChar === " " ? "\t" : " ")) {
757: $this->emitError(new Error(
758: 'Invalid indentation - tabs and spaces cannot be mixed', $attributes
759: ));
760: } elseif (strlen($prefix) < $indentLen && !isset($matches[2])) {
761: $this->emitError(new Error(
762: 'Invalid body indentation level ' .
763: '(expecting an indentation level of at least ' . $indentLen . ')',
764: $attributes
765: ));
766: }
767: return substr($matches[0], strlen($prefix));
768: },
769: $string
770: );
771: }
772:
773: protected function parseDocString(
774: string $startToken, $contents, string $endToken,
775: array $attributes, array $endTokenAttributes, bool $parseUnicodeEscape
776: ) {
777: $kind = strpos($startToken, "'") === false
778: ? String_::KIND_HEREDOC : String_::KIND_NOWDOC;
779:
780: $regex = '/\A[bB]?<<<[ \t]*[\'"]?([a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*)[\'"]?(?:\r\n|\n|\r)\z/';
781: $result = preg_match($regex, $startToken, $matches);
782: assert($result === 1);
783: $label = $matches[1];
784:
785: $result = preg_match('/\A[ \t]*/', $endToken, $matches);
786: assert($result === 1);
787: $indentation = $matches[0];
788:
789: $attributes['kind'] = $kind;
790: $attributes['docLabel'] = $label;
791: $attributes['docIndentation'] = $indentation;
792:
793: $indentHasSpaces = false !== strpos($indentation, " ");
794: $indentHasTabs = false !== strpos($indentation, "\t");
795: if ($indentHasSpaces && $indentHasTabs) {
796: $this->emitError(new Error(
797: 'Invalid indentation - tabs and spaces cannot be mixed',
798: $endTokenAttributes
799: ));
800:
801: // Proceed processing as if this doc string is not indented
802: $indentation = '';
803: }
804:
805: $indentLen = \strlen($indentation);
806: $indentChar = $indentHasSpaces ? " " : "\t";
807:
808: if (\is_string($contents)) {
809: if ($contents === '') {
810: return new String_('', $attributes);
811: }
812:
813: $contents = $this->stripIndentation(
814: $contents, $indentLen, $indentChar, true, true, $attributes
815: );
816: $contents = preg_replace('~(\r\n|\n|\r)\z~', '', $contents);
817:
818: if ($kind === String_::KIND_HEREDOC) {
819: $contents = String_::parseEscapeSequences($contents, null, $parseUnicodeEscape);
820: }
821:
822: return new String_($contents, $attributes);
823: } else {
824: assert(count($contents) > 0);
825: if (!$contents[0] instanceof Node\Scalar\EncapsedStringPart) {
826: // If there is no leading encapsed string part, pretend there is an empty one
827: $this->stripIndentation(
828: '', $indentLen, $indentChar, true, false, $contents[0]->getAttributes()
829: );
830: }
831:
832: $newContents = [];
833: foreach ($contents as $i => $part) {
834: if ($part instanceof Node\Scalar\EncapsedStringPart) {
835: $isLast = $i === \count($contents) - 1;
836: $part->value = $this->stripIndentation(
837: $part->value, $indentLen, $indentChar,
838: $i === 0, $isLast, $part->getAttributes()
839: );
840: $part->value = String_::parseEscapeSequences($part->value, null, $parseUnicodeEscape);
841: if ($isLast) {
842: $part->value = preg_replace('~(\r\n|\n|\r)\z~', '', $part->value);
843: }
844: if ('' === $part->value) {
845: continue;
846: }
847: }
848: $newContents[] = $part;
849: }
850: return new Encapsed($newContents, $attributes);
851: }
852: }
853:
854: /**
855: * Create attributes for a zero-length common-capturing nop.
856: *
857: * @param Comment[] $comments
858: * @return array
859: */
860: protected function createCommentNopAttributes(array $comments) {
861: $comment = $comments[count($comments) - 1];
862: $commentEndLine = $comment->getEndLine();
863: $commentEndFilePos = $comment->getEndFilePos();
864: $commentEndTokenPos = $comment->getEndTokenPos();
865:
866: $attributes = ['comments' => $comments];
867: if (-1 !== $commentEndLine) {
868: $attributes['startLine'] = $commentEndLine;
869: $attributes['endLine'] = $commentEndLine;
870: }
871: if (-1 !== $commentEndFilePos) {
872: $attributes['startFilePos'] = $commentEndFilePos + 1;
873: $attributes['endFilePos'] = $commentEndFilePos;
874: }
875: if (-1 !== $commentEndTokenPos) {
876: $attributes['startTokenPos'] = $commentEndTokenPos + 1;
877: $attributes['endTokenPos'] = $commentEndTokenPos;
878: }
879: return $attributes;
880: }
881:
882: /** @param ElseIf_|Else_ $node */
883: protected function fixupAlternativeElse($node) {
884: // Make sure a trailing nop statement carrying comments is part of the node.
885: $numStmts = \count($node->stmts);
886: if ($numStmts !== 0 && $node->stmts[$numStmts - 1] instanceof Nop) {
887: $nopAttrs = $node->stmts[$numStmts - 1]->getAttributes();
888: if (isset($nopAttrs['endLine'])) {
889: $node->setAttribute('endLine', $nopAttrs['endLine']);
890: }
891: if (isset($nopAttrs['endFilePos'])) {
892: $node->setAttribute('endFilePos', $nopAttrs['endFilePos']);
893: }
894: if (isset($nopAttrs['endTokenPos'])) {
895: $node->setAttribute('endTokenPos', $nopAttrs['endTokenPos']);
896: }
897: }
898: }
899:
900: protected function checkClassModifier($a, $b, $modifierPos) {
901: try {
902: Class_::verifyClassModifier($a, $b);
903: } catch (Error $error) {
904: $error->setAttributes($this->getAttributesAt($modifierPos));
905: $this->emitError($error);
906: }
907: }
908:
909: protected function checkModifier($a, $b, $modifierPos) {
910: // Jumping through some hoops here because verifyModifier() is also used elsewhere
911: try {
912: Class_::verifyModifier($a, $b);
913: } catch (Error $error) {
914: $error->setAttributes($this->getAttributesAt($modifierPos));
915: $this->emitError($error);
916: }
917: }
918:
919: protected function checkParam(Param $node) {
920: if ($node->variadic && null !== $node->default) {
921: $this->emitError(new Error(
922: 'Variadic parameter cannot have a default value',
923: $node->default->getAttributes()
924: ));
925: }
926: }
927:
928: protected function checkTryCatch(TryCatch $node) {
929: if (empty($node->catches) && null === $node->finally) {
930: $this->emitError(new Error(
931: 'Cannot use try without catch or finally', $node->getAttributes()
932: ));
933: }
934: }
935:
936: protected function checkNamespace(Namespace_ $node) {
937: if (null !== $node->stmts) {
938: foreach ($node->stmts as $stmt) {
939: if ($stmt instanceof Namespace_) {
940: $this->emitError(new Error(
941: 'Namespace declarations cannot be nested', $stmt->getAttributes()
942: ));
943: }
944: }
945: }
946: }
947:
948: private function checkClassName($name, $namePos) {
949: if (null !== $name && $name->isSpecialClassName()) {
950: $this->emitError(new Error(
951: sprintf('Cannot use \'%s\' as class name as it is reserved', $name),
952: $this->getAttributesAt($namePos)
953: ));
954: }
955: }
956:
957: private function checkImplementedInterfaces(array $interfaces) {
958: foreach ($interfaces as $interface) {
959: if ($interface->isSpecialClassName()) {
960: $this->emitError(new Error(
961: sprintf('Cannot use \'%s\' as interface name as it is reserved', $interface),
962: $interface->getAttributes()
963: ));
964: }
965: }
966: }
967:
968: protected function checkClass(Class_ $node, $namePos) {
969: $this->checkClassName($node->name, $namePos);
970:
971: if ($node->extends && $node->extends->isSpecialClassName()) {
972: $this->emitError(new Error(
973: sprintf('Cannot use \'%s\' as class name as it is reserved', $node->extends),
974: $node->extends->getAttributes()
975: ));
976: }
977:
978: $this->checkImplementedInterfaces($node->implements);
979: }
980:
981: protected function checkInterface(Interface_ $node, $namePos) {
982: $this->checkClassName($node->name, $namePos);
983: $this->checkImplementedInterfaces($node->extends);
984: }
985:
986: protected function checkEnum(Enum_ $node, $namePos) {
987: $this->checkClassName($node->name, $namePos);
988: $this->checkImplementedInterfaces($node->implements);
989: }
990:
991: protected function checkClassMethod(ClassMethod $node, $modifierPos) {
992: if ($node->flags & Class_::MODIFIER_STATIC) {
993: switch ($node->name->toLowerString()) {
994: case '__construct':
995: $this->emitError(new Error(
996: sprintf('Constructor %s() cannot be static', $node->name),
997: $this->getAttributesAt($modifierPos)));
998: break;
999: case '__destruct':
1000: $this->emitError(new Error(
1001: sprintf('Destructor %s() cannot be static', $node->name),
1002: $this->getAttributesAt($modifierPos)));
1003: break;
1004: case '__clone':
1005: $this->emitError(new Error(
1006: sprintf('Clone method %s() cannot be static', $node->name),
1007: $this->getAttributesAt($modifierPos)));
1008: break;
1009: }
1010: }
1011:
1012: if ($node->flags & Class_::MODIFIER_READONLY) {
1013: $this->emitError(new Error(
1014: sprintf('Method %s() cannot be readonly', $node->name),
1015: $this->getAttributesAt($modifierPos)));
1016: }
1017: }
1018:
1019: protected function checkClassConst(ClassConst $node, $modifierPos) {
1020: if ($node->flags & Class_::MODIFIER_STATIC) {
1021: $this->emitError(new Error(
1022: "Cannot use 'static' as constant modifier",
1023: $this->getAttributesAt($modifierPos)));
1024: }
1025: if ($node->flags & Class_::MODIFIER_ABSTRACT) {
1026: $this->emitError(new Error(
1027: "Cannot use 'abstract' as constant modifier",
1028: $this->getAttributesAt($modifierPos)));
1029: }
1030: if ($node->flags & Class_::MODIFIER_READONLY) {
1031: $this->emitError(new Error(
1032: "Cannot use 'readonly' as constant modifier",
1033: $this->getAttributesAt($modifierPos)));
1034: }
1035: }
1036:
1037: protected function checkProperty(Property $node, $modifierPos) {
1038: if ($node->flags & Class_::MODIFIER_ABSTRACT) {
1039: $this->emitError(new Error('Properties cannot be declared abstract',
1040: $this->getAttributesAt($modifierPos)));
1041: }
1042:
1043: if ($node->flags & Class_::MODIFIER_FINAL) {
1044: $this->emitError(new Error('Properties cannot be declared final',
1045: $this->getAttributesAt($modifierPos)));
1046: }
1047: }
1048:
1049: protected function checkUseUse(UseUse $node, $namePos) {
1050: if ($node->alias && $node->alias->isSpecialClassName()) {
1051: $this->emitError(new Error(
1052: sprintf(
1053: 'Cannot use %s as %s because \'%2$s\' is a special class name',
1054: $node->name, $node->alias
1055: ),
1056: $this->getAttributesAt($namePos)
1057: ));
1058: }
1059: }
1060: }
1061: