php-parser/lib/PhpParser/ParserAbstract.php
Nikita Popov 6996db1e3a Build node attributes inside semantic action methods
Minor performance improvement for parsing, also allows to access
attributes with higher granulity in the parser, though this is not
currently done.

* #n can now be used to access the stack position of a token. $n
  is the same as $this->semStack[#n]. (Post-translate $n will
  actually be the stack position.)
* $attributeStack is now $this->startAttributeStack and
  $endAttributes is now $this->endAttributes.
* Attributes for a node are now computed inside the individual
  reduction methods, instead of being passed as a parameter.
  Accessible through the attributes() macro.
2015-04-26 11:55:59 +02:00

451 lines
18 KiB
PHP

<?php
namespace PhpParser;
/*
* This parser is based on a skeleton written by Moriyoshi Koizumi, which in
* turn is based on work by Masato Bito.
*/
abstract class ParserAbstract
{
const SYMBOL_NONE = -1;
/*
* The following members will be filled with generated parsing data:
*/
/** @var int Size of $tokenToSymbol map */
protected $tokenToSymbolMapSize;
/** @var int Size of $action table */
protected $actionTableSize;
/** @var int Size of $goto table */
protected $gotoTableSize;
/** @var int Symbol number signifying an invalid token */
protected $invalidSymbol;
/** @var int Symbol number of error recovery token */
protected $errorSymbol;
/** @var int Action number signifying default action */
protected $defaultAction;
/** @var int Rule number signifying that an unexpected token was encountered */
protected $unexpectedTokenRule;
protected $YY2TBLSTATE;
protected $YYNLSTATES;
/** @var array Map of lexer tokens to internal symbols */
protected $tokenToSymbol;
/** @var array Map of symbols to their names */
protected $symbolToName;
/** @var array Names of the production rules (only necessary for debugging) */
protected $productions;
/** @var array Map of states to a displacement into the $action table. The corresponding action for this
* state/symbol pair is $action[$actionBase[$state] + $symbol]. If $actionBase[$state] is 0, the
action is defaulted, i.e. $actionDefault[$state] should be used instead. */
protected $actionBase;
/** @var array Table of actions. Indexed according to $actionBase comment. */
protected $action;
/** @var array Table indexed analogously to $action. If $actionCheck[$actionBase[$state] + $symbol] != $symbol
* then the action is defaulted, i.e. $actionDefault[$state] should be used instead. */
protected $actionCheck;
/** @var array Map of states to their default action */
protected $actionDefault;
/** @var array Map of non-terminals to a displacement into the $goto table. The corresponding goto state for this
* non-terminal/state pair is $goto[$gotoBase[$nonTerminal] + $state] (unless defaulted) */
protected $gotoBase;
/** @var array Table of states to goto after reduction. Indexed according to $gotoBase comment. */
protected $goto;
/** @var array Table indexed analogously to $goto. If $gotoCheck[$gotoBase[$nonTerminal] + $state] != $nonTerminal
* then the goto state is defaulted, i.e. $gotoDefault[$nonTerminal] should be used. */
protected $gotoCheck;
/** @var array Map of non-terminals to the default state to goto after their reduction */
protected $gotoDefault;
/** @var array Map of rules to the non-terminal on their left-hand side, i.e. the non-terminal to use for
* determining the state to goto after reduction. */
protected $ruleToNonTerminal;
/** @var array Map of rules to the length of their right-hand side, which is the number of elements that have to
* be popped from the stack(s) on reduction. */
protected $ruleToLength;
/*
* The following members are part of the parser state:
*/
/** @var Lexer Lexer that is used when parsing */
protected $lexer;
/** @var mixed Temporary value containing the result of last semantic action (reduction) */
protected $semValue;
/** @var int Position in stacks (state stack, semantic value stack, attribute stack) */
protected $stackPos;
/** @var array Semantic value stack (contains values of tokens and semantic action results) */
protected $semStack;
/** @var array[] Start attribute stack */
protected $startAttributeStack;
/** @var array End attributes of last *shifted* token */
protected $endAttributes;
/** @var bool Whether to throw on first error */
protected $throwOnError;
/** @var Error[] Errors collected during last parse */
protected $errors;
/**
* Creates a parser instance.
*
* @param Lexer $lexer A lexer
* @param array $options Options array. The boolean 'throwOnError' option determined whether an exception should be
* thrown on first error, or if the parser should try to continue parsing the remaining code
* and build a partial AST.
*/
public function __construct(Lexer $lexer, array $options = array()) {
$this->lexer = $lexer;
$this->errors = array();
$this->throwOnError = isset($options['throwOnError']) ? $options['throwOnError'] : true;
}
/**
* Get array of errors that occurred during the last parse.
*
* This method may only return multiple errors if the throwOnError option is disabled.
*
* @return Error[]
*/
public function getErrors() {
return $this->errors;
}
/**
* Parses PHP code into a node tree.
*
* @param string $code The source code to parse
*
* @return Node[] Array of statements
*/
public function parse($code) {
$this->lexer->startLexing($code);
$this->errors = array();
// We start off with no lookahead-token
$symbol = self::SYMBOL_NONE;
// The attributes for a node are taken from the first and last token of the node.
// From the first token only the startAttributes are taken and from the last only
// the endAttributes. Both are merged using the array union operator (+).
$startAttributes = array('startLine' => 1);
$this->endAttributes = array();
// In order to figure out the attributes for the starting token, we have to keep
// them in a stack
$this->startAttributeStack = array($startAttributes);
// Start off in the initial state and keep a stack of previous states
$state = 0;
$stateStack = array($state);
// Semantic value stack (contains values of tokens and semantic action results)
$this->semStack = array();
// Current position in the stack(s)
$this->stackPos = 0;
$errorState = 0;
for (;;) {
//$this->traceNewState($state, $symbol);
if ($this->actionBase[$state] == 0) {
$rule = $this->actionDefault[$state];
} else {
if ($symbol === self::SYMBOL_NONE) {
// Fetch the next token id from the lexer and fetch additional info by-ref.
// The end attributes are fetched into a temporary variable and only set once the token is really
// shifted (not during read). Otherwise you would sometimes get off-by-one errors, when a rule is
// reduced after a token was read but not yet shifted.
$tokenId = $this->lexer->getNextToken($tokenValue, $startAttributes, $nextEndAttributes);
// map the lexer token id to the internally used symbols
$symbol = $tokenId >= 0 && $tokenId < $this->tokenToSymbolMapSize
? $this->tokenToSymbol[$tokenId]
: $this->invalidSymbol;
if ($symbol === $this->invalidSymbol) {
throw new \RangeException(sprintf(
'The lexer returned an invalid token (id=%d, value=%s)',
$tokenId, $tokenValue
));
}
$this->startAttributeStack[$this->stackPos] = $startAttributes;
//$this->traceRead($symbol);
}
$idx = $this->actionBase[$state] + $symbol;
if ((($idx >= 0 && $idx < $this->actionTableSize && $this->actionCheck[$idx] == $symbol)
|| ($state < $this->YY2TBLSTATE
&& ($idx = $this->actionBase[$state + $this->YYNLSTATES] + $symbol) >= 0
&& $idx < $this->actionTableSize && $this->actionCheck[$idx] == $symbol))
&& ($action = $this->action[$idx]) != $this->defaultAction) {
/*
* >= YYNLSTATES: shift and reduce
* > 0: shift
* = 0: accept
* < 0: reduce
* = -YYUNEXPECTED: error
*/
if ($action > 0) {
/* shift */
//$this->traceShift($symbol);
++$this->stackPos;
$stateStack[$this->stackPos] = $state = $action;
$this->semStack[$this->stackPos] = $tokenValue;
$this->startAttributeStack[$this->stackPos] = $startAttributes;
$this->endAttributes = $nextEndAttributes;
$symbol = self::SYMBOL_NONE;
if ($errorState) {
--$errorState;
}
if ($action < $this->YYNLSTATES) {
continue;
}
/* $yyn >= YYNLSTATES means shift-and-reduce */
$rule = $action - $this->YYNLSTATES;
} else {
$rule = -$action;
}
} else {
$rule = $this->actionDefault[$state];
}
}
for (;;) {
if ($rule === 0) {
/* accept */
//$this->traceAccept();
return $this->semValue;
} elseif ($rule !== $this->unexpectedTokenRule) {
/* reduce */
//$this->traceReduce($rule);
try {
$this->{'reduceRule' . $rule}();
} catch (Error $e) {
if (-1 === $e->getStartLine() && isset($startAttributes['startLine'])) {
$e->setStartLine($startAttributes['startLine']);
}
throw $e;
}
/* Goto - shift nonterminal */
$this->stackPos -= $this->ruleToLength[$rule];
$nonTerminal = $this->ruleToNonTerminal[$rule];
$idx = $this->gotoBase[$nonTerminal] + $stateStack[$this->stackPos];
if ($idx >= 0 && $idx < $this->gotoTableSize && $this->gotoCheck[$idx] == $nonTerminal) {
$state = $this->goto[$idx];
} else {
$state = $this->gotoDefault[$nonTerminal];
}
++$this->stackPos;
$stateStack[$this->stackPos] = $state;
$this->semStack[$this->stackPos] = $this->semValue;
$this->startAttributeStack[$this->stackPos] = $startAttributes;
} else {
/* error */
switch ($errorState) {
case 0:
$msg = $this->getErrorMessage($symbol, $state);
$error = new Error($msg, $startAttributes + $nextEndAttributes);
$this->errors[] = $error;
if ($this->throwOnError) {
throw $error;
}
// Break missing intentionally
case 1:
case 2:
$errorState = 3;
// Pop until error-expecting state uncovered
while (!(
(($idx = $this->actionBase[$state] + $this->errorSymbol) >= 0
&& $idx < $this->actionTableSize && $this->actionCheck[$idx] == $this->errorSymbol)
|| ($state < $this->YY2TBLSTATE
&& ($idx = $this->actionBase[$state + $this->YYNLSTATES] + $this->errorSymbol) >= 0
&& $idx < $this->actionTableSize && $this->actionCheck[$idx] == $this->errorSymbol)
) || ($action = $this->action[$idx]) == $this->defaultAction) { // Not totally sure about this
if ($this->stackPos <= 0) {
throw new Error('Could not recover from error');
}
$state = $stateStack[--$this->stackPos];
//$this->tracePop($state);
}
//$this->traceShift($this->errorSymbol);
$stateStack[++$this->stackPos] = $state = $action;
break;
case 3:
if ($symbol === 0) {
// Reached EOF
throw new Error('Could not recover from error');
}
//$this->traceDiscard($symbol);
$symbol = self::SYMBOL_NONE;
break 2;
}
}
if ($state < $this->YYNLSTATES) {
break;
}
/* >= YYNLSTATES means shift-and-reduce */
$rule = $state - $this->YYNLSTATES;
}
}
}
protected function getErrorMessage($symbol, $state) {
$expectedString = '';
if ($expected = $this->getExpectedTokens($state)) {
$expectedString = ', expecting ' . implode(' or ', $expected);
}
return 'Syntax error, unexpected ' . $this->symbolToName[$symbol] . $expectedString;
}
protected function getExpectedTokens($state) {
$expected = array();
$base = $this->actionBase[$state];
foreach ($this->symbolToName as $symbol => $name) {
$idx = $base + $symbol;
if ($idx >= 0 && $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol
|| $state < $this->YY2TBLSTATE
&& ($idx = $this->actionBase[$state + $this->YYNLSTATES] + $symbol) >= 0
&& $idx < $this->actionTableSize && $this->actionCheck[$idx] === $symbol
) {
if ($this->action[$idx] != $this->unexpectedTokenRule) {
if (count($expected) == 4) {
/* Too many expected tokens */
return array();
}
$expected[] = $name;
}
}
}
return $expected;
}
/*
* Tracing functions used for debugging the parser.
*/
protected function traceNewState($state, $symbol) {
echo '% State ' . $state
. ', Lookahead ' . ($symbol == self::SYMBOL_NONE ? '--none--' : $this->symbolToName[$symbol]) . "\n";
}
protected function traceRead($symbol) {
echo '% Reading ' . $this->symbolToName[$symbol] . "\n";
}
protected function traceShift($symbol) {
echo '% Shift ' . $this->symbolToName[$symbol] . "\n";
}
protected function traceAccept() {
echo "% Accepted.\n";
}
protected function traceReduce($n) {
echo '% Reduce by (' . $n . ') ' . $this->productions[$n] . "\n";
}
protected function tracePop($state) {
echo '% Recovering, uncovered state ' . $state . "\n";
}
protected function traceDiscard($symbol) {
echo '% Discard ' . $this->symbolToName[$symbol] . "\n";
}
/*
* Helper functions invoked by semantic actions
*/
/**
* Moves statements of semicolon-style namespaces into $ns->stmts and checks various error conditions.
*
* @param Node[] $stmts
* @return Node[]
*/
protected function handleNamespaces(array $stmts) {
$style = $this->getNamespacingStyle($stmts);
if (null === $style) {
// not namespaced, nothing to do
return $stmts;
} elseif ('brace' === $style) {
// For braced namespaces we only have to check that there are no invalid statements between the namespaces
$afterFirstNamespace = false;
foreach ($stmts as $stmt) {
if ($stmt instanceof Node\Stmt\Namespace_) {
$afterFirstNamespace = true;
} elseif (!$stmt instanceof Node\Stmt\HaltCompiler && $afterFirstNamespace) {
throw new Error('No code may exist outside of namespace {}', $stmt->getLine());
}
}
return $stmts;
} else {
// For semicolon namespaces we have to move the statements after a namespace declaration into ->stmts
$resultStmts = array();
$targetStmts =& $resultStmts;
foreach ($stmts as $stmt) {
if ($stmt instanceof Node\Stmt\Namespace_) {
$stmt->stmts = array();
$targetStmts =& $stmt->stmts;
$resultStmts[] = $stmt;
} elseif ($stmt instanceof Node\Stmt\HaltCompiler) {
// __halt_compiler() is not moved into the namespace
$resultStmts[] = $stmt;
} else {
$targetStmts[] = $stmt;
}
}
return $resultStmts;
}
}
private function getNamespacingStyle(array $stmts) {
$style = null;
$hasNotAllowedStmts = false;
foreach ($stmts as $stmt) {
if ($stmt instanceof Node\Stmt\Namespace_) {
$currentStyle = null === $stmt->stmts ? 'semicolon' : 'brace';
if (null === $style) {
$style = $currentStyle;
if ($hasNotAllowedStmts) {
throw new Error('Namespace declaration statement has to be the very first statement in the script', $stmt->getLine());
}
} elseif ($style !== $currentStyle) {
throw new Error('Cannot mix bracketed namespace declarations with unbracketed namespace declarations', $stmt->getLine());
}
} elseif (!$stmt instanceof Node\Stmt\Declare_ && !$stmt instanceof Node\Stmt\HaltCompiler) {
$hasNotAllowedStmts = true;
}
}
return $style;
}
}