Chris@13: <?php declare(strict_types=1);
Chris@0: 
Chris@0: namespace PhpParser;
Chris@0: 
Chris@0: /*
Chris@0:  * This parser is based on a skeleton written by Moriyoshi Koizumi, which in
Chris@0:  * turn is based on work by Masato Bito.
Chris@0:  */
Chris@13: use PhpParser\Node\Expr;
Chris@17: use PhpParser\Node\Expr\Cast\Double;
Chris@0: use PhpParser\Node\Name;
Chris@0: use PhpParser\Node\Param;
Chris@17: use PhpParser\Node\Scalar\Encapsed;
Chris@0: use PhpParser\Node\Scalar\LNumber;
Chris@0: use PhpParser\Node\Scalar\String_;
Chris@0: use PhpParser\Node\Stmt\Class_;
Chris@0: use PhpParser\Node\Stmt\ClassConst;
Chris@0: use PhpParser\Node\Stmt\ClassMethod;
Chris@0: use PhpParser\Node\Stmt\Interface_;
Chris@0: use PhpParser\Node\Stmt\Namespace_;
Chris@0: use PhpParser\Node\Stmt\Property;
Chris@0: use PhpParser\Node\Stmt\TryCatch;
Chris@0: use PhpParser\Node\Stmt\UseUse;
Chris@13: use PhpParser\Node\VarLikeIdentifier;
Chris@0: 
Chris@0: abstract class ParserAbstract implements Parser
Chris@0: {
Chris@0:     const SYMBOL_NONE = -1;
Chris@0: 
Chris@0:     /*
Chris@0:      * The following members will be filled with generated parsing data:
Chris@0:      */
Chris@0: 
Chris@0:     /** @var int Size of $tokenToSymbol map */
Chris@0:     protected $tokenToSymbolMapSize;
Chris@0:     /** @var int Size of $action table */
Chris@0:     protected $actionTableSize;
Chris@0:     /** @var int Size of $goto table */
Chris@0:     protected $gotoTableSize;
Chris@0: 
Chris@0:     /** @var int Symbol number signifying an invalid token */
Chris@0:     protected $invalidSymbol;
Chris@0:     /** @var int Symbol number of error recovery token */
Chris@0:     protected $errorSymbol;
Chris@0:     /** @var int Action number signifying default action */
Chris@0:     protected $defaultAction;
Chris@0:     /** @var int Rule number signifying that an unexpected token was encountered */
Chris@0:     protected $unexpectedTokenRule;
Chris@0: 
Chris@0:     protected $YY2TBLSTATE;
Chris@13:     /** @var int Number of non-leaf states */
Chris@13:     protected $numNonLeafStates;
Chris@0: 
Chris@13:     /** @var int[] Map of lexer tokens to internal symbols */
Chris@0:     protected $tokenToSymbol;
Chris@13:     /** @var string[] Map of symbols to their names */
Chris@0:     protected $symbolToName;
Chris@0:     /** @var array Names of the production rules (only necessary for debugging) */
Chris@0:     protected $productions;
Chris@0: 
Chris@13:     /** @var int[] Map of states to a displacement into the $action table. The corresponding action for this
Chris@0:      *             state/symbol pair is $action[$actionBase[$state] + $symbol]. If $actionBase[$state] is 0, the
Chris@0:                    action is defaulted, i.e. $actionDefault[$state] should be used instead. */
Chris@0:     protected $actionBase;
Chris@13:     /** @var int[] Table of actions. Indexed according to $actionBase comment. */
Chris@0:     protected $action;
Chris@13:     /** @var int[] Table indexed analogously to $action. If $actionCheck[$actionBase[$state] + $symbol] != $symbol
Chris@0:      *             then the action is defaulted, i.e. $actionDefault[$state] should be used instead. */
Chris@0:     protected $actionCheck;
Chris@13:     /** @var int[] Map of states to their default action */
Chris@0:     protected $actionDefault;
Chris@13:     /** @var callable[] Semantic action callbacks */
Chris@13:     protected $reduceCallbacks;
Chris@0: 
Chris@13:     /** @var int[] Map of non-terminals to a displacement into the $goto table. The corresponding goto state for this
Chris@0:      *             non-terminal/state pair is $goto[$gotoBase[$nonTerminal] + $state] (unless defaulted) */
Chris@0:     protected $gotoBase;
Chris@13:     /** @var int[] Table of states to goto after reduction. Indexed according to $gotoBase comment. */
Chris@0:     protected $goto;
Chris@13:     /** @var int[] Table indexed analogously to $goto. If $gotoCheck[$gotoBase[$nonTerminal] + $state] != $nonTerminal
Chris@0:      *             then the goto state is defaulted, i.e. $gotoDefault[$nonTerminal] should be used. */
Chris@0:     protected $gotoCheck;
Chris@13:     /** @var int[] Map of non-terminals to the default state to goto after their reduction */
Chris@0:     protected $gotoDefault;
Chris@0: 
Chris@13:     /** @var int[] Map of rules to the non-terminal on their left-hand side, i.e. the non-terminal to use for
Chris@0:      *             determining the state to goto after reduction. */
Chris@0:     protected $ruleToNonTerminal;
Chris@13:     /** @var int[] Map of rules to the length of their right-hand side, which is the number of elements that have to
Chris@0:      *             be popped from the stack(s) on reduction. */
Chris@0:     protected $ruleToLength;
Chris@0: 
Chris@0:     /*
Chris@0:      * The following members are part of the parser state:
Chris@0:      */
Chris@0: 
Chris@0:     /** @var Lexer Lexer that is used when parsing */
Chris@0:     protected $lexer;
Chris@0:     /** @var mixed Temporary value containing the result of last semantic action (reduction) */
Chris@0:     protected $semValue;
Chris@0:     /** @var array Semantic value stack (contains values of tokens and semantic action results) */
Chris@0:     protected $semStack;
Chris@0:     /** @var array[] Start attribute stack */
Chris@0:     protected $startAttributeStack;
Chris@0:     /** @var array[] End attribute stack */
Chris@0:     protected $endAttributeStack;
Chris@0:     /** @var array End attributes of last *shifted* token */
Chris@0:     protected $endAttributes;
Chris@0:     /** @var array Start attributes of last *read* token */
Chris@0:     protected $lookaheadStartAttributes;
Chris@0: 
Chris@0:     /** @var ErrorHandler Error handler */
Chris@0:     protected $errorHandler;
Chris@0:     /** @var int Error state, used to avoid error floods */
Chris@0:     protected $errorState;
Chris@0: 
Chris@0:     /**
Chris@13:      * Initialize $reduceCallbacks map.
Chris@13:      */
Chris@13:     abstract protected function initReduceCallbacks();
Chris@13: 
Chris@13:     /**
Chris@0:      * Creates a parser instance.
Chris@0:      *
Chris@13:      * Options: Currently none.
Chris@13:      *
Chris@0:      * @param Lexer $lexer A lexer
Chris@13:      * @param array $options Options array.
Chris@0:      */
Chris@13:     public function __construct(Lexer $lexer, array $options = []) {
Chris@0:         $this->lexer = $lexer;
Chris@0: 
Chris@0:         if (isset($options['throwOnError'])) {
Chris@0:             throw new \LogicException(
Chris@0:                 '"throwOnError" is no longer supported, use "errorHandler" instead');
Chris@0:         }
Chris@13: 
Chris@13:         $this->initReduceCallbacks();
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses PHP code into a node tree.
Chris@0:      *
Chris@0:      * If a non-throwing error handler is used, the parser will continue parsing after an error
Chris@0:      * occurred and attempt to build a partial AST.
Chris@0:      *
Chris@0:      * @param string $code The source code to parse
Chris@0:      * @param ErrorHandler|null $errorHandler Error handler to use for lexer/parser errors, defaults
Chris@0:      *                                        to ErrorHandler\Throwing.
Chris@0:      *
Chris@13:      * @return Node\Stmt[]|null Array of statements (or null non-throwing error handler is used and
Chris@13:      *                          the parser was unable to recover from an error).
Chris@0:      */
Chris@13:     public function parse(string $code, ErrorHandler $errorHandler = null) {
Chris@0:         $this->errorHandler = $errorHandler ?: new ErrorHandler\Throwing;
Chris@0: 
Chris@0:         $this->lexer->startLexing($code, $this->errorHandler);
Chris@13:         $result = $this->doParse();
Chris@0: 
Chris@13:         // Clear out some of the interior state, so we don't hold onto unnecessary
Chris@13:         // memory between uses of the parser
Chris@13:         $this->startAttributeStack = [];
Chris@13:         $this->endAttributeStack = [];
Chris@13:         $this->semStack = [];
Chris@13:         $this->semValue = null;
Chris@13: 
Chris@13:         return $result;
Chris@13:     }
Chris@13: 
Chris@13:     protected function doParse() {
Chris@0:         // We start off with no lookahead-token
Chris@0:         $symbol = self::SYMBOL_NONE;
Chris@0: 
Chris@0:         // The attributes for a node are taken from the first and last token of the node.
Chris@0:         // From the first token only the startAttributes are taken and from the last only
Chris@0:         // the endAttributes. Both are merged using the array union operator (+).
Chris@13:         $startAttributes = [];
Chris@13:         $endAttributes = [];
Chris@0:         $this->endAttributes = $endAttributes;
Chris@0: 
Chris@0:         // Keep stack of start and end attributes
Chris@13:         $this->startAttributeStack = [];
Chris@13:         $this->endAttributeStack = [$endAttributes];
Chris@0: 
Chris@0:         // Start off in the initial state and keep a stack of previous states
Chris@0:         $state = 0;
Chris@13:         $stateStack = [$state];
Chris@0: 
Chris@0:         // Semantic value stack (contains values of tokens and semantic action results)
Chris@13:         $this->semStack = [];
Chris@0: 
Chris@0:         // Current position in the stack(s)
Chris@13:         $stackPos = 0;
Chris@0: 
Chris@0:         $this->errorState = 0;
Chris@0: 
Chris@0:         for (;;) {
Chris@0:             //$this->traceNewState($state, $symbol);
Chris@0: 
Chris@13:             if ($this->actionBase[$state] === 0) {
Chris@0:                 $rule = $this->actionDefault[$state];
Chris@0:             } else {
Chris@0:                 if ($symbol === self::SYMBOL_NONE) {
Chris@0:                     // Fetch the next token id from the lexer and fetch additional info by-ref.
Chris@0:                     // The end attributes are fetched into a temporary variable and only set once the token is really
Chris@0:                     // shifted (not during read). Otherwise you would sometimes get off-by-one errors, when a rule is
Chris@0:                     // reduced after a token was read but not yet shifted.
Chris@0:                     $tokenId = $this->lexer->getNextToken($tokenValue, $startAttributes, $endAttributes);
Chris@0: 
Chris@0:                     // map the lexer token id to the internally used symbols
Chris@0:                     $symbol = $tokenId >= 0 && $tokenId < $this->tokenToSymbolMapSize
Chris@0:                         ? $this->tokenToSymbol[$tokenId]
Chris@0:                         : $this->invalidSymbol;
Chris@0: 
Chris@0:                     if ($symbol === $this->invalidSymbol) {
Chris@0:                         throw new \RangeException(sprintf(
Chris@0:                             'The lexer returned an invalid token (id=%d, value=%s)',
Chris@0:                             $tokenId, $tokenValue
Chris@0:                         ));
Chris@0:                     }
Chris@0: 
Chris@0:                     // This is necessary to assign some meaningful attributes to /* empty */ productions. They'll get
Chris@0:                     // the attributes of the next token, even though they don't contain it themselves.
Chris@13:                     $this->startAttributeStack[$stackPos+1] = $startAttributes;
Chris@13:                     $this->endAttributeStack[$stackPos+1] = $endAttributes;
Chris@0:                     $this->lookaheadStartAttributes = $startAttributes;
Chris@0: 
Chris@0:                     //$this->traceRead($symbol);
Chris@0:                 }
Chris@0: 
Chris@0:                 $idx = $this->actionBase[$state] + $symbol;
Chris@13:                 if ((($idx >= 0 && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol)
Chris@0:                      || ($state < $this->YY2TBLSTATE
Chris@13:                          && ($idx = $this->actionBase[$state + $this->numNonLeafStates] + $symbol) >= 0
Chris@13:                          && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol))
Chris@13:                     && ($action = $this->action[$idx]) !== $this->defaultAction) {
Chris@0:                     /*
Chris@13:                      * >= numNonLeafStates: shift and reduce
Chris@0:                      * > 0: shift
Chris@0:                      * = 0: accept
Chris@0:                      * < 0: reduce
Chris@0:                      * = -YYUNEXPECTED: error
Chris@0:                      */
Chris@0:                     if ($action > 0) {
Chris@0:                         /* shift */
Chris@0:                         //$this->traceShift($symbol);
Chris@0: 
Chris@13:                         ++$stackPos;
Chris@13:                         $stateStack[$stackPos] = $state = $action;
Chris@13:                         $this->semStack[$stackPos] = $tokenValue;
Chris@13:                         $this->startAttributeStack[$stackPos] = $startAttributes;
Chris@13:                         $this->endAttributeStack[$stackPos] = $endAttributes;
Chris@0:                         $this->endAttributes = $endAttributes;
Chris@0:                         $symbol = self::SYMBOL_NONE;
Chris@0: 
Chris@0:                         if ($this->errorState) {
Chris@0:                             --$this->errorState;
Chris@0:                         }
Chris@0: 
Chris@13:                         if ($action < $this->numNonLeafStates) {
Chris@0:                             continue;
Chris@0:                         }
Chris@0: 
Chris@13:                         /* $yyn >= numNonLeafStates means shift-and-reduce */
Chris@13:                         $rule = $action - $this->numNonLeafStates;
Chris@0:                     } else {
Chris@0:                         $rule = -$action;
Chris@0:                     }
Chris@0:                 } else {
Chris@0:                     $rule = $this->actionDefault[$state];
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:             for (;;) {
Chris@0:                 if ($rule === 0) {
Chris@0:                     /* accept */
Chris@0:                     //$this->traceAccept();
Chris@0:                     return $this->semValue;
Chris@0:                 } elseif ($rule !== $this->unexpectedTokenRule) {
Chris@0:                     /* reduce */
Chris@0:                     //$this->traceReduce($rule);
Chris@0: 
Chris@0:                     try {
Chris@13:                         $this->reduceCallbacks[$rule]($stackPos);
Chris@0:                     } catch (Error $e) {
Chris@0:                         if (-1 === $e->getStartLine() && isset($startAttributes['startLine'])) {
Chris@0:                             $e->setStartLine($startAttributes['startLine']);
Chris@0:                         }
Chris@0: 
Chris@0:                         $this->emitError($e);
Chris@0:                         // Can't recover from this type of error
Chris@0:                         return null;
Chris@0:                     }
Chris@0: 
Chris@0:                     /* Goto - shift nonterminal */
Chris@13:                     $lastEndAttributes = $this->endAttributeStack[$stackPos];
Chris@13:                     $stackPos -= $this->ruleToLength[$rule];
Chris@0:                     $nonTerminal = $this->ruleToNonTerminal[$rule];
Chris@13:                     $idx = $this->gotoBase[$nonTerminal] + $stateStack[$stackPos];
Chris@13:                     if ($idx >= 0 && $idx < $this->gotoTableSize && $this->gotoCheck[$idx] === $nonTerminal) {
Chris@0:                         $state = $this->goto[$idx];
Chris@0:                     } else {
Chris@0:                         $state = $this->gotoDefault[$nonTerminal];
Chris@0:                     }
Chris@0: 
Chris@13:                     ++$stackPos;
Chris@13:                     $stateStack[$stackPos]     = $state;
Chris@13:                     $this->semStack[$stackPos] = $this->semValue;
Chris@13:                     $this->endAttributeStack[$stackPos] = $lastEndAttributes;
Chris@0:                 } else {
Chris@0:                     /* error */
Chris@0:                     switch ($this->errorState) {
Chris@0:                         case 0:
Chris@0:                             $msg = $this->getErrorMessage($symbol, $state);
Chris@0:                             $this->emitError(new Error($msg, $startAttributes + $endAttributes));
Chris@0:                             // Break missing intentionally
Chris@0:                         case 1:
Chris@0:                         case 2:
Chris@0:                             $this->errorState = 3;
Chris@0: 
Chris@0:                             // Pop until error-expecting state uncovered
Chris@0:                             while (!(
Chris@0:                                 (($idx = $this->actionBase[$state] + $this->errorSymbol) >= 0
Chris@13:                                     && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $this->errorSymbol)
Chris@0:                                 || ($state < $this->YY2TBLSTATE
Chris@13:                                     && ($idx = $this->actionBase[$state + $this->numNonLeafStates] + $this->errorSymbol) >= 0
Chris@13:                                     && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $this->errorSymbol)
Chris@13:                             ) || ($action = $this->action[$idx]) === $this->defaultAction) { // Not totally sure about this
Chris@13:                                 if ($stackPos <= 0) {
Chris@0:                                     // Could not recover from error
Chris@0:                                     return null;
Chris@0:                                 }
Chris@13:                                 $state = $stateStack[--$stackPos];
Chris@0:                                 //$this->tracePop($state);
Chris@0:                             }
Chris@0: 
Chris@0:                             //$this->traceShift($this->errorSymbol);
Chris@13:                             ++$stackPos;
Chris@13:                             $stateStack[$stackPos] = $state = $action;
Chris@0: 
Chris@0:                             // We treat the error symbol as being empty, so we reset the end attributes
Chris@0:                             // to the end attributes of the last non-error symbol
Chris@13:                             $this->endAttributeStack[$stackPos] = $this->endAttributeStack[$stackPos - 1];
Chris@13:                             $this->endAttributes = $this->endAttributeStack[$stackPos - 1];
Chris@0:                             break;
Chris@0: 
Chris@0:                         case 3:
Chris@0:                             if ($symbol === 0) {
Chris@0:                                 // Reached EOF without recovering from error
Chris@0:                                 return null;
Chris@0:                             }
Chris@0: 
Chris@0:                             //$this->traceDiscard($symbol);
Chris@0:                             $symbol = self::SYMBOL_NONE;
Chris@0:                             break 2;
Chris@0:                     }
Chris@0:                 }
Chris@0: 
Chris@13:                 if ($state < $this->numNonLeafStates) {
Chris@0:                     break;
Chris@0:                 }
Chris@0: 
Chris@13:                 /* >= numNonLeafStates means shift-and-reduce */
Chris@13:                 $rule = $state - $this->numNonLeafStates;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         throw new \RuntimeException('Reached end of parser loop');
Chris@0:     }
Chris@0: 
Chris@0:     protected function emitError(Error $error) {
Chris@0:         $this->errorHandler->handleError($error);
Chris@0:     }
Chris@0: 
Chris@13:     /**
Chris@13:      * Format error message including expected tokens.
Chris@13:      *
Chris@13:      * @param int $symbol Unexpected symbol
Chris@13:      * @param int $state  State at time of error
Chris@13:      *
Chris@13:      * @return string Formatted error message
Chris@13:      */
Chris@13:     protected function getErrorMessage(int $symbol, int $state) : string {
Chris@0:         $expectedString = '';
Chris@0:         if ($expected = $this->getExpectedTokens($state)) {
Chris@0:             $expectedString = ', expecting ' . implode(' or ', $expected);
Chris@0:         }
Chris@0: 
Chris@0:         return 'Syntax error, unexpected ' . $this->symbolToName[$symbol] . $expectedString;
Chris@0:     }
Chris@0: 
Chris@13:     /**
Chris@13:      * Get limited number of expected tokens in given state.
Chris@13:      *
Chris@13:      * @param int $state State
Chris@13:      *
Chris@13:      * @return string[] Expected tokens. If too many, an empty array is returned.
Chris@13:      */
Chris@13:     protected function getExpectedTokens(int $state) : array {
Chris@13:         $expected = [];
Chris@0: 
Chris@0:         $base = $this->actionBase[$state];
Chris@0:         foreach ($this->symbolToName as $symbol => $name) {
Chris@0:             $idx = $base + $symbol;
Chris@0:             if ($idx >= 0 && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol
Chris@0:                 || $state < $this->YY2TBLSTATE
Chris@13:                 && ($idx = $this->actionBase[$state + $this->numNonLeafStates] + $symbol) >= 0
Chris@0:                 && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol
Chris@0:             ) {
Chris@13:                 if ($this->action[$idx] !== $this->unexpectedTokenRule
Chris@13:                     && $this->action[$idx] !== $this->defaultAction
Chris@13:                     && $symbol !== $this->errorSymbol
Chris@0:                 ) {
Chris@13:                     if (count($expected) === 4) {
Chris@0:                         /* Too many expected tokens */
Chris@13:                         return [];
Chris@0:                     }
Chris@0: 
Chris@0:                     $expected[] = $name;
Chris@0:                 }
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return $expected;
Chris@0:     }
Chris@0: 
Chris@0:     /*
Chris@0:      * Tracing functions used for debugging the parser.
Chris@0:      */
Chris@0: 
Chris@0:     /*
Chris@0:     protected function traceNewState($state, $symbol) {
Chris@0:         echo '% State ' . $state
Chris@0:             . ', Lookahead ' . ($symbol == self::SYMBOL_NONE ? '--none--' : $this->symbolToName[$symbol]) . "\n";
Chris@0:     }
Chris@0: 
Chris@0:     protected function traceRead($symbol) {
Chris@0:         echo '% Reading ' . $this->symbolToName[$symbol] . "\n";
Chris@0:     }
Chris@0: 
Chris@0:     protected function traceShift($symbol) {
Chris@0:         echo '% Shift ' . $this->symbolToName[$symbol] . "\n";
Chris@0:     }
Chris@0: 
Chris@0:     protected function traceAccept() {
Chris@0:         echo "% Accepted.\n";
Chris@0:     }
Chris@0: 
Chris@0:     protected function traceReduce($n) {
Chris@0:         echo '% Reduce by (' . $n . ') ' . $this->productions[$n] . "\n";
Chris@0:     }
Chris@0: 
Chris@0:     protected function tracePop($state) {
Chris@0:         echo '% Recovering, uncovered state ' . $state . "\n";
Chris@0:     }
Chris@0: 
Chris@0:     protected function traceDiscard($symbol) {
Chris@0:         echo '% Discard ' . $this->symbolToName[$symbol] . "\n";
Chris@0:     }
Chris@0:     */
Chris@0: 
Chris@0:     /*
Chris@0:      * Helper functions invoked by semantic actions
Chris@0:      */
Chris@0: 
Chris@0:     /**
Chris@0:      * Moves statements of semicolon-style namespaces into $ns->stmts and checks various error conditions.
Chris@0:      *
Chris@13:      * @param Node\Stmt[] $stmts
Chris@13:      * @return Node\Stmt[]
Chris@0:      */
Chris@13:     protected function handleNamespaces(array $stmts) : array {
Chris@0:         $hasErrored = false;
Chris@0:         $style = $this->getNamespacingStyle($stmts);
Chris@0:         if (null === $style) {
Chris@0:             // not namespaced, nothing to do
Chris@0:             return $stmts;
Chris@0:         } elseif ('brace' === $style) {
Chris@0:             // For braced namespaces we only have to check that there are no invalid statements between the namespaces
Chris@0:             $afterFirstNamespace = false;
Chris@0:             foreach ($stmts as $stmt) {
Chris@0:                 if ($stmt instanceof Node\Stmt\Namespace_) {
Chris@0:                     $afterFirstNamespace = true;
Chris@0:                 } elseif (!$stmt instanceof Node\Stmt\HaltCompiler
Chris@0:                         && !$stmt instanceof Node\Stmt\Nop
Chris@0:                         && $afterFirstNamespace && !$hasErrored) {
Chris@0:                     $this->emitError(new Error(
Chris@0:                         'No code may exist outside of namespace {}', $stmt->getAttributes()));
Chris@0:                     $hasErrored = true; // Avoid one error for every statement
Chris@0:                 }
Chris@0:             }
Chris@0:             return $stmts;
Chris@0:         } else {
Chris@0:             // For semicolon namespaces we have to move the statements after a namespace declaration into ->stmts
Chris@13:             $resultStmts = [];
Chris@0:             $targetStmts =& $resultStmts;
Chris@13:             $lastNs = null;
Chris@0:             foreach ($stmts as $stmt) {
Chris@0:                 if ($stmt instanceof Node\Stmt\Namespace_) {
Chris@13:                     if ($lastNs !== null) {
Chris@13:                         $this->fixupNamespaceAttributes($lastNs);
Chris@13:                     }
Chris@0:                     if ($stmt->stmts === null) {
Chris@13:                         $stmt->stmts = [];
Chris@0:                         $targetStmts =& $stmt->stmts;
Chris@0:                         $resultStmts[] = $stmt;
Chris@0:                     } else {
Chris@0:                         // This handles the invalid case of mixed style namespaces
Chris@0:                         $resultStmts[] = $stmt;
Chris@0:                         $targetStmts =& $resultStmts;
Chris@0:                     }
Chris@13:                     $lastNs = $stmt;
Chris@0:                 } elseif ($stmt instanceof Node\Stmt\HaltCompiler) {
Chris@0:                     // __halt_compiler() is not moved into the namespace
Chris@0:                     $resultStmts[] = $stmt;
Chris@0:                 } else {
Chris@0:                     $targetStmts[] = $stmt;
Chris@0:                 }
Chris@0:             }
Chris@13:             if ($lastNs !== null) {
Chris@13:                 $this->fixupNamespaceAttributes($lastNs);
Chris@13:             }
Chris@0:             return $resultStmts;
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@13:     private function fixupNamespaceAttributes(Node\Stmt\Namespace_ $stmt) {
Chris@13:         // We moved the statements into the namespace node, as such the end of the namespace node
Chris@13:         // needs to be extended to the end of the statements.
Chris@13:         if (empty($stmt->stmts)) {
Chris@13:             return;
Chris@13:         }
Chris@13: 
Chris@13:         // We only move the builtin end attributes here. This is the best we can do with the
Chris@13:         // knowledge we have.
Chris@13:         $endAttributes = ['endLine', 'endFilePos', 'endTokenPos'];
Chris@13:         $lastStmt = $stmt->stmts[count($stmt->stmts) - 1];
Chris@13:         foreach ($endAttributes as $endAttribute) {
Chris@13:             if ($lastStmt->hasAttribute($endAttribute)) {
Chris@13:                 $stmt->setAttribute($endAttribute, $lastStmt->getAttribute($endAttribute));
Chris@13:             }
Chris@13:         }
Chris@13:     }
Chris@13: 
Chris@13:     /**
Chris@13:      * Determine namespacing style (semicolon or brace)
Chris@13:      *
Chris@13:      * @param Node[] $stmts Top-level statements.
Chris@13:      *
Chris@13:      * @return null|string One of "semicolon", "brace" or null (no namespaces)
Chris@13:      */
Chris@0:     private function getNamespacingStyle(array $stmts) {
Chris@0:         $style = null;
Chris@0:         $hasNotAllowedStmts = false;
Chris@0:         foreach ($stmts as $i => $stmt) {
Chris@0:             if ($stmt instanceof Node\Stmt\Namespace_) {
Chris@0:                 $currentStyle = null === $stmt->stmts ? 'semicolon' : 'brace';
Chris@0:                 if (null === $style) {
Chris@0:                     $style = $currentStyle;
Chris@0:                     if ($hasNotAllowedStmts) {
Chris@0:                         $this->emitError(new Error(
Chris@0:                             'Namespace declaration statement has to be the very first statement in the script',
Chris@0:                             $stmt->getLine() // Avoid marking the entire namespace as an error
Chris@0:                         ));
Chris@0:                     }
Chris@0:                 } elseif ($style !== $currentStyle) {
Chris@0:                     $this->emitError(new Error(
Chris@0:                         'Cannot mix bracketed namespace declarations with unbracketed namespace declarations',
Chris@0:                         $stmt->getLine() // Avoid marking the entire namespace as an error
Chris@0:                     ));
Chris@0:                     // Treat like semicolon style for namespace normalization
Chris@0:                     return 'semicolon';
Chris@0:                 }
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             /* declare(), __halt_compiler() and nops can be used before a namespace declaration */
Chris@0:             if ($stmt instanceof Node\Stmt\Declare_
Chris@0:                 || $stmt instanceof Node\Stmt\HaltCompiler
Chris@0:                 || $stmt instanceof Node\Stmt\Nop) {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             /* There may be a hashbang line at the very start of the file */
Chris@13:             if ($i === 0 && $stmt instanceof Node\Stmt\InlineHTML && preg_match('/\A#!.*\r?\n\z/', $stmt->value)) {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             /* Everything else if forbidden before namespace declarations */
Chris@0:             $hasNotAllowedStmts = true;
Chris@0:         }
Chris@0:         return $style;
Chris@0:     }
Chris@0: 
Chris@13:     /**
Chris@13:      * Fix up parsing of static property calls in PHP 5.
Chris@13:      *
Chris@13:      * In PHP 5 A::$b[c][d] and A::$b[c][d]() have very different interpretation. The former is
Chris@13:      * interpreted as (A::$b)[c][d], while the latter is the same as A::{$b[c][d]}(). We parse the
Chris@13:      * latter as the former initially and this method fixes the AST into the correct form when we
Chris@13:      * encounter the "()".
Chris@13:      *
Chris@13:      * @param  Node\Expr\StaticPropertyFetch|Node\Expr\ArrayDimFetch $prop
Chris@13:      * @param  Node\Arg[] $args
Chris@13:      * @param  array      $attributes
Chris@13:      *
Chris@13:      * @return Expr\StaticCall
Chris@13:      */
Chris@13:     protected function fixupPhp5StaticPropCall($prop, array $args, array $attributes) : Expr\StaticCall {
Chris@13:         if ($prop instanceof Node\Expr\StaticPropertyFetch) {
Chris@13:             $name = $prop->name instanceof VarLikeIdentifier
Chris@13:                 ? $prop->name->toString() : $prop->name;
Chris@13:             $var = new Expr\Variable($name, $prop->name->getAttributes());
Chris@13:             return new Expr\StaticCall($prop->class, $var, $args, $attributes);
Chris@13:         } elseif ($prop instanceof Node\Expr\ArrayDimFetch) {
Chris@13:             $tmp = $prop;
Chris@13:             while ($tmp->var instanceof Node\Expr\ArrayDimFetch) {
Chris@13:                 $tmp = $tmp->var;
Chris@13:             }
Chris@13: 
Chris@13:             /** @var Expr\StaticPropertyFetch $staticProp */
Chris@13:             $staticProp = $tmp->var;
Chris@13: 
Chris@13:             // Set start attributes to attributes of innermost node
Chris@13:             $tmp = $prop;
Chris@13:             $this->fixupStartAttributes($tmp, $staticProp->name);
Chris@13:             while ($tmp->var instanceof Node\Expr\ArrayDimFetch) {
Chris@13:                 $tmp = $tmp->var;
Chris@13:                 $this->fixupStartAttributes($tmp, $staticProp->name);
Chris@13:             }
Chris@13: 
Chris@13:             $name = $staticProp->name instanceof VarLikeIdentifier
Chris@13:                 ? $staticProp->name->toString() : $staticProp->name;
Chris@13:             $tmp->var = new Expr\Variable($name, $staticProp->name->getAttributes());
Chris@13:             return new Expr\StaticCall($staticProp->class, $prop, $args, $attributes);
Chris@13:         } else {
Chris@13:             throw new \Exception;
Chris@13:         }
Chris@13:     }
Chris@13: 
Chris@13:     protected function fixupStartAttributes(Node $to, Node $from) {
Chris@13:         $startAttributes = ['startLine', 'startFilePos', 'startTokenPos'];
Chris@13:         foreach ($startAttributes as $startAttribute) {
Chris@13:             if ($from->hasAttribute($startAttribute)) {
Chris@13:                 $to->setAttribute($startAttribute, $from->getAttribute($startAttribute));
Chris@13:             }
Chris@13:         }
Chris@13:     }
Chris@13: 
Chris@0:     protected function handleBuiltinTypes(Name $name) {
Chris@0:         $scalarTypes = [
Chris@0:             'bool'     => true,
Chris@0:             'int'      => true,
Chris@0:             'float'    => true,
Chris@0:             'string'   => true,
Chris@0:             'iterable' => true,
Chris@0:             'void'     => true,
Chris@0:             'object'   => true,
Chris@0:         ];
Chris@0: 
Chris@0:         if (!$name->isUnqualified()) {
Chris@0:             return $name;
Chris@0:         }
Chris@0: 
Chris@13:         $lowerName = $name->toLowerString();
Chris@13:         if (!isset($scalarTypes[$lowerName])) {
Chris@13:             return $name;
Chris@13:         }
Chris@13: 
Chris@13:         return new Node\Identifier($lowerName, $name->getAttributes());
Chris@0:     }
Chris@0: 
Chris@13:     /**
Chris@13:      * Get combined start and end attributes at a stack location
Chris@13:      *
Chris@13:      * @param int $pos Stack location
Chris@13:      *
Chris@13:      * @return array Combined start and end attributes
Chris@13:      */
Chris@13:     protected function getAttributesAt(int $pos) : array {
Chris@0:         return $this->startAttributeStack[$pos] + $this->endAttributeStack[$pos];
Chris@0:     }
Chris@0: 
Chris@17:     protected function getFloatCastKind(string $cast): int
Chris@17:     {
Chris@17:         $cast = strtolower($cast);
Chris@17:         if (strpos($cast, 'float') !== false) {
Chris@17:             return Double::KIND_FLOAT;
Chris@17:         }
Chris@17: 
Chris@17:         if (strpos($cast, 'real') !== false) {
Chris@17:             return Double::KIND_REAL;
Chris@17:         }
Chris@17: 
Chris@17:         return Double::KIND_DOUBLE;
Chris@17:     }
Chris@17: 
Chris@0:     protected function parseLNumber($str, $attributes, $allowInvalidOctal = false) {
Chris@0:         try {
Chris@0:             return LNumber::fromString($str, $attributes, $allowInvalidOctal);
Chris@0:         } catch (Error $error) {
Chris@0:             $this->emitError($error);
Chris@0:             // Use dummy value
Chris@0:             return new LNumber(0, $attributes);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@13:     /**
Chris@13:      * Parse a T_NUM_STRING token into either an integer or string node.
Chris@13:      *
Chris@13:      * @param string $str        Number string
Chris@13:      * @param array  $attributes Attributes
Chris@13:      *
Chris@13:      * @return LNumber|String_ Integer or string node.
Chris@13:      */
Chris@13:     protected function parseNumString(string $str, array $attributes) {
Chris@0:         if (!preg_match('/^(?:0|-?[1-9][0-9]*)$/', $str)) {
Chris@0:             return new String_($str, $attributes);
Chris@0:         }
Chris@0: 
Chris@0:         $num = +$str;
Chris@0:         if (!is_int($num)) {
Chris@0:             return new String_($str, $attributes);
Chris@0:         }
Chris@0: 
Chris@0:         return new LNumber($num, $attributes);
Chris@0:     }
Chris@0: 
Chris@17:     protected function stripIndentation(
Chris@17:         string $string, int $indentLen, string $indentChar,
Chris@17:         bool $newlineAtStart, bool $newlineAtEnd, array $attributes
Chris@17:     ) {
Chris@17:         if ($indentLen === 0) {
Chris@17:             return $string;
Chris@17:         }
Chris@17: 
Chris@17:         $start = $newlineAtStart ? '(?:(?<=\n)|\A)' : '(?<=\n)';
Chris@17:         $end = $newlineAtEnd ? '(?:(?=[\r\n])|\z)' : '(?=[\r\n])';
Chris@17:         $regex = '/' . $start . '([ \t]*)(' . $end . ')?/';
Chris@17:         return preg_replace_callback(
Chris@17:             $regex,
Chris@17:             function ($matches) use ($indentLen, $indentChar, $attributes) {
Chris@17:                 $prefix = substr($matches[1], 0, $indentLen);
Chris@17:                 if (false !== strpos($prefix, $indentChar === " " ? "\t" : " ")) {
Chris@17:                     $this->emitError(new Error(
Chris@17:                         'Invalid indentation - tabs and spaces cannot be mixed', $attributes
Chris@17:                     ));
Chris@17:                 } elseif (strlen($prefix) < $indentLen && !isset($matches[2])) {
Chris@17:                     $this->emitError(new Error(
Chris@17:                         'Invalid body indentation level ' .
Chris@17:                         '(expecting an indentation level of at least ' . $indentLen . ')',
Chris@17:                         $attributes
Chris@17:                     ));
Chris@17:                 }
Chris@17:                 return substr($matches[0], strlen($prefix));
Chris@17:             },
Chris@17:             $string
Chris@17:         );
Chris@17:     }
Chris@17: 
Chris@17:     protected function parseDocString(
Chris@17:         string $startToken, $contents, string $endToken,
Chris@17:         array $attributes, array $endTokenAttributes, bool $parseUnicodeEscape
Chris@17:     ) {
Chris@17:         $kind = strpos($startToken, "'") === false
Chris@17:             ? String_::KIND_HEREDOC : String_::KIND_NOWDOC;
Chris@17: 
Chris@17:         $regex = '/\A[bB]?<<<[ \t]*[\'"]?([a-zA-Z_\x7f-\xff][a-zA-Z0-9_\x7f-\xff]*)[\'"]?(?:\r\n|\n|\r)\z/';
Chris@17:         $result = preg_match($regex, $startToken, $matches);
Chris@17:         assert($result === 1);
Chris@17:         $label = $matches[1];
Chris@17: 
Chris@17:         $result = preg_match('/\A[ \t]*/', $endToken, $matches);
Chris@17:         assert($result === 1);
Chris@17:         $indentation = $matches[0];
Chris@17: 
Chris@17:         $attributes['kind'] = $kind;
Chris@17:         $attributes['docLabel'] = $label;
Chris@17:         $attributes['docIndentation'] = $indentation;
Chris@17: 
Chris@17:         $indentHasSpaces = false !== strpos($indentation, " ");
Chris@17:         $indentHasTabs = false !== strpos($indentation, "\t");
Chris@17:         if ($indentHasSpaces && $indentHasTabs) {
Chris@17:             $this->emitError(new Error(
Chris@17:                 'Invalid indentation - tabs and spaces cannot be mixed',
Chris@17:                 $endTokenAttributes
Chris@17:             ));
Chris@17: 
Chris@17:             // Proceed processing as if this doc string is not indented
Chris@17:             $indentation = '';
Chris@17:         }
Chris@17: 
Chris@17:         $indentLen = \strlen($indentation);
Chris@17:         $indentChar = $indentHasSpaces ? " " : "\t";
Chris@17: 
Chris@17:         if (\is_string($contents)) {
Chris@17:             if ($contents === '') {
Chris@17:                 return new String_('', $attributes);
Chris@17:             }
Chris@17: 
Chris@17:             $contents = $this->stripIndentation(
Chris@17:                 $contents, $indentLen, $indentChar, true, true, $attributes
Chris@17:             );
Chris@17:             $contents = preg_replace('~(\r\n|\n|\r)\z~', '', $contents);
Chris@17: 
Chris@17:             if ($kind === String_::KIND_HEREDOC) {
Chris@17:                 $contents = String_::parseEscapeSequences($contents, null, $parseUnicodeEscape);
Chris@17:             }
Chris@17: 
Chris@17:             return new String_($contents, $attributes);
Chris@17:         } else {
Chris@17:             assert(count($contents) > 0);
Chris@17:             if (!$contents[0] instanceof Node\Scalar\EncapsedStringPart) {
Chris@17:                 // If there is no leading encapsed string part, pretend there is an empty one
Chris@17:                 $this->stripIndentation(
Chris@17:                     '', $indentLen, $indentChar, true, false, $contents[0]->getAttributes()
Chris@17:                 );
Chris@17:             }
Chris@17: 
Chris@17:             $newContents = [];
Chris@17:             foreach ($contents as $i => $part) {
Chris@17:                 if ($part instanceof Node\Scalar\EncapsedStringPart) {
Chris@17:                     $isLast = $i === \count($contents) - 1;
Chris@17:                     $part->value = $this->stripIndentation(
Chris@17:                         $part->value, $indentLen, $indentChar,
Chris@17:                         $i === 0, $isLast, $part->getAttributes()
Chris@17:                     );
Chris@17:                     $part->value = String_::parseEscapeSequences($part->value, null, $parseUnicodeEscape);
Chris@17:                     if ($isLast) {
Chris@17:                         $part->value = preg_replace('~(\r\n|\n|\r)\z~', '', $part->value);
Chris@17:                     }
Chris@17:                     if ('' === $part->value) {
Chris@17:                         continue;
Chris@17:                     }
Chris@17:                 }
Chris@17:                 $newContents[] = $part;
Chris@17:             }
Chris@17:             return new Encapsed($newContents, $attributes);
Chris@17:         }
Chris@17:     }
Chris@17: 
Chris@0:     protected function checkModifier($a, $b, $modifierPos) {
Chris@0:         // Jumping through some hoops here because verifyModifier() is also used elsewhere
Chris@0:         try {
Chris@0:             Class_::verifyModifier($a, $b);
Chris@0:         } catch (Error $error) {
Chris@0:             $error->setAttributes($this->getAttributesAt($modifierPos));
Chris@0:             $this->emitError($error);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkParam(Param $node) {
Chris@0:         if ($node->variadic && null !== $node->default) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 'Variadic parameter cannot have a default value',
Chris@0:                 $node->default->getAttributes()
Chris@0:             ));
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkTryCatch(TryCatch $node) {
Chris@0:         if (empty($node->catches) && null === $node->finally) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 'Cannot use try without catch or finally', $node->getAttributes()
Chris@0:             ));
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkNamespace(Namespace_ $node) {
Chris@13:         if ($node->name && $node->name->isSpecialClassName()) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 sprintf('Cannot use \'%s\' as namespace name', $node->name),
Chris@0:                 $node->name->getAttributes()
Chris@0:             ));
Chris@0:         }
Chris@0: 
Chris@0:         if (null !== $node->stmts) {
Chris@0:             foreach ($node->stmts as $stmt) {
Chris@0:                 if ($stmt instanceof Namespace_) {
Chris@0:                     $this->emitError(new Error(
Chris@0:                         'Namespace declarations cannot be nested', $stmt->getAttributes()
Chris@0:                     ));
Chris@0:                 }
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkClass(Class_ $node, $namePos) {
Chris@13:         if (null !== $node->name && $node->name->isSpecialClassName()) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 sprintf('Cannot use \'%s\' as class name as it is reserved', $node->name),
Chris@0:                 $this->getAttributesAt($namePos)
Chris@0:             ));
Chris@0:         }
Chris@0: 
Chris@13:         if ($node->extends && $node->extends->isSpecialClassName()) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 sprintf('Cannot use \'%s\' as class name as it is reserved', $node->extends),
Chris@0:                 $node->extends->getAttributes()
Chris@0:             ));
Chris@0:         }
Chris@0: 
Chris@0:         foreach ($node->implements as $interface) {
Chris@13:             if ($interface->isSpecialClassName()) {
Chris@0:                 $this->emitError(new Error(
Chris@0:                     sprintf('Cannot use \'%s\' as interface name as it is reserved', $interface),
Chris@0:                     $interface->getAttributes()
Chris@0:                 ));
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkInterface(Interface_ $node, $namePos) {
Chris@13:         if (null !== $node->name && $node->name->isSpecialClassName()) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 sprintf('Cannot use \'%s\' as class name as it is reserved', $node->name),
Chris@0:                 $this->getAttributesAt($namePos)
Chris@0:             ));
Chris@0:         }
Chris@0: 
Chris@0:         foreach ($node->extends as $interface) {
Chris@13:             if ($interface->isSpecialClassName()) {
Chris@0:                 $this->emitError(new Error(
Chris@0:                     sprintf('Cannot use \'%s\' as interface name as it is reserved', $interface),
Chris@0:                     $interface->getAttributes()
Chris@0:                 ));
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkClassMethod(ClassMethod $node, $modifierPos) {
Chris@0:         if ($node->flags & Class_::MODIFIER_STATIC) {
Chris@13:             switch ($node->name->toLowerString()) {
Chris@0:                 case '__construct':
Chris@0:                     $this->emitError(new Error(
Chris@0:                         sprintf('Constructor %s() cannot be static', $node->name),
Chris@0:                         $this->getAttributesAt($modifierPos)));
Chris@0:                     break;
Chris@0:                 case '__destruct':
Chris@0:                     $this->emitError(new Error(
Chris@0:                         sprintf('Destructor %s() cannot be static', $node->name),
Chris@0:                         $this->getAttributesAt($modifierPos)));
Chris@0:                     break;
Chris@0:                 case '__clone':
Chris@0:                     $this->emitError(new Error(
Chris@0:                         sprintf('Clone method %s() cannot be static', $node->name),
Chris@0:                         $this->getAttributesAt($modifierPos)));
Chris@0:                     break;
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkClassConst(ClassConst $node, $modifierPos) {
Chris@0:         if ($node->flags & Class_::MODIFIER_STATIC) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 "Cannot use 'static' as constant modifier",
Chris@0:                 $this->getAttributesAt($modifierPos)));
Chris@0:         }
Chris@0:         if ($node->flags & Class_::MODIFIER_ABSTRACT) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 "Cannot use 'abstract' as constant modifier",
Chris@0:                 $this->getAttributesAt($modifierPos)));
Chris@0:         }
Chris@0:         if ($node->flags & Class_::MODIFIER_FINAL) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 "Cannot use 'final' as constant modifier",
Chris@0:                 $this->getAttributesAt($modifierPos)));
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkProperty(Property $node, $modifierPos) {
Chris@0:         if ($node->flags & Class_::MODIFIER_ABSTRACT) {
Chris@0:             $this->emitError(new Error('Properties cannot be declared abstract',
Chris@0:                 $this->getAttributesAt($modifierPos)));
Chris@0:         }
Chris@0: 
Chris@0:         if ($node->flags & Class_::MODIFIER_FINAL) {
Chris@0:             $this->emitError(new Error('Properties cannot be declared final',
Chris@0:                 $this->getAttributesAt($modifierPos)));
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function checkUseUse(UseUse $node, $namePos) {
Chris@13:         if ($node->alias && $node->alias->isSpecialClassName()) {
Chris@0:             $this->emitError(new Error(
Chris@0:                 sprintf(
Chris@0:                     'Cannot use %s as %s because \'%2$s\' is a special class name',
Chris@0:                     $node->name, $node->alias
Chris@0:                 ),
Chris@0:                 $this->getAttributesAt($namePos)
Chris@0:             ));
Chris@0:         }
Chris@0:     }
Chris@0: }