/**
* @param array|NormalizedProduction[] $productions
* @param ParseSets $firstSets
* @param SymbolSet $emptySet
*
* @return SetsGenerator
*/
public function generateFirstSets(array $productions, ParseSets $firstSets, SymbolSet $emptySet)
{
//add epsilon to the first sets of the non terminals which generate epsilon
foreach ($productions as $production) {
$lhs = $production->getLeftHandSide();
if ($emptySet->contains($lhs)) {
$firstSets->addEpsilon($lhs);
}
}
//initialize the obvious first sets of productions which start with a terminal
foreach ($productions as $production) {
$symbol = $production->getFirstSymbol();
if ($symbol->getType() == Symbol::TYPE_TERMINAL) {
$firstSets->addTerminal($production->getLeftHandSide(), $symbol);
}
}
//initialize first sets for productions which contain several non-terminals
do {
$changes = false;
foreach ($productions as $production) {
$updateSet = new ArraySet();
$rhs = $production->getRightHandSide();
$this->firstSetCalculator->processSymbolList($updateSet, $rhs, $firstSets);
$nonTerminal = $production->getLeftHandSide();
$changes |= $firstSets->addAllTerminals($nonTerminal, $updateSet);
}
} while ($changes);
return $this;
}
/**
* @param NormalizedProduction $production
* @param SymbolSet $alreadyAdded
*
* @return bool
*/
private function indirectlyGeneratesEpsilon(NormalizedProduction $production, SymbolSet $alreadyAdded)
{
$rhsItems = $production->getRightHandSide();
/** @var Symbol $item */
$item = reset($rhsItems);
$answer = $alreadyAdded->contains($item);
for (next($rhsItems); $answer && !is_null(key($rhsItems)); next($rhsItems)) {
$item = current($rhsItems);
$answer = $answer && $alreadyAdded->contains($item);
}
$answer = is_null(key($rhsItems)) && $answer;
return $answer;
}
/**
* Calculates the first set of $list and adds it's elements to $set
*
* @param \Helstern\Nomsky\Grammar\Symbol\SymbolSet $newFirstSet
* @param array|NormalizedProduction[] $list
* @param \Helstern\Nomsky\GrammarAnalysis\ParseSets\ParseSets $firstSets
*
* @return bool if the epsilon symbol was added to $set
*/
public function processSymbolList(SymbolSet $newFirstSet, array $list, ParseSets $firstSets)
{
if (0 == count($list)) {
$newFirstSet->add(new EpsilonSymbol());
return true;
}
if (1 == count($list)) {
/** @var Symbol $symbol */
$symbol = $list[0];
return $this->processSymbol($newFirstSet, $symbol, $firstSets);
}
$symbolIsEpsilon = SymbolIsEpsilon::singletonInstance();
$symbolsIsNonTerminal = SymbolTypeEquals::newInstanceMatchingNonTerminals();
/** @var SymbolSet $lastSet */
$lastSet = null;
$epsilonCounter = new MatchCountingInterceptor($symbolIsEpsilon);
$acceptor = Inverter::newInstance($epsilonCounter);
//we assume there is no epsilon symbol in the $list list
$previousSymbol = reset($list);
//add the first non-terminal and continue past this one later on
if ($symbolsIsNonTerminal->matchSymbol($previousSymbol)) {
$lastSet = $firstSets->filterTerminalSet($previousSymbol, $acceptor);
$newFirstSet->addAll($lastSet);
} else {
$newFirstSet->add($previousSymbol);
}
//as long as the previous symbol was a non-terminal, process the next symbol
for (next($list); !is_null(key($list)) && $symbolsIsNonTerminal->matchSymbol($previousSymbol) && $epsilonCounter->getMatchCount() > 0; next($list)) {
$previousSymbol = current($list);
$epsilonCounter = new MatchCountingInterceptor($symbolIsEpsilon);
$acceptor = Inverter::newInstance($epsilonCounter);
if ($symbolsIsNonTerminal->matchSymbol($previousSymbol)) {
$lastSet = $firstSets->filterTerminalSet($previousSymbol, $acceptor);
$newFirstSet->addAll($lastSet);
} else {
$newFirstSet->add($previousSymbol);
}
}
if (is_null(key($list)) && $symbolsIsNonTerminal->matchSymbol($previousSymbol) && $epsilonCounter->getMatchCount() > 0) {
$newFirstSet->add(new EpsilonSymbol());
return true;
}
return false;
}
/**
* @param Symbol $nonTerminal
* @param Symbol $terminal
* @throws \Exception
* @return null
*/
private function assertKnownSymbols(Symbol $nonTerminal, Symbol $terminal)
{
$knownNonTerminal = $this->nonTerminals->contains($nonTerminal);
$knownTerminal = $this->terminals->contains($terminal);
if ($knownNonTerminal && $knownTerminal) {
return true;
}
if (!$knownNonTerminal && !$knownTerminal) {
$assertionMessage = 'Uknown symbols: non-terminal %s and terminal %s';
$assertionMessage = sprintf($assertionMessage, $nonTerminal->toString(), $terminal->toString());
} else {
if ($knownTerminal) {
$assertionMessage = 'Uknown non-terminal: %s';
$assertionMessage = sprintf($assertionMessage, $nonTerminal->toString());
} else {
$assertionMessage = 'Uknown terminal: %s';
$assertionMessage = sprintf($assertionMessage, $terminal->toString());
}
}
throw new \RuntimeException($assertionMessage);
}