/**
* Optimize regular expression
*
* Tries to optimize the given regular expression. Returns true if the AST
* has been modified, and false otherwise.
*
* @param slRegularExpression $regularExpression
* @return bool
*/
public function optimize(slRegularExpression &$regularExpression)
{
if ($regularExpression instanceof slRegularExpressionMultiple && count($children = $regularExpression->getChildren()) === 1) {
$regularExpression = reset($children);
return true;
}
return $this->recurse($regularExpression);
}
/**
* Optimize regular expression
*
* Tries to optimize the given regular expression. Returns true if the AST
* has been modified, and false otherwise.
*
* @param slRegularExpression $regularExpression
* @return bool
*/
public function optimize(slRegularExpression &$regularExpression)
{
if ((($op1 = $regularExpression instanceof slRegularExpressionOptional) || $regularExpression instanceof slRegularExpressionRepeated) && (($op2 = $regularExpression->getChild() instanceof slRegularExpressionOptional) || $regularExpression->getChild() instanceof slRegularExpressionRepeated)) {
$class = $op1 && $op2 ? 'slRegularExpressionOptional' : 'slRegularExpressionRepeated';
$regularExpression = new $class($regularExpression->getChild()->getChild());
return true;
}
return $this->recurse($regularExpression);
}
/**
* Recurse through the container stack
*
* Recurse through the container stack and call the optimize method for
* each child.
*
* Returns true, if something has been modified, and false otherwise.
*
* @param slRegularExpression $regularExpression
* @return bool
*/
protected function recurse(slRegularExpression $regularExpression)
{
if (!$regularExpression instanceof slRegularExpressionContainer) {
return false;
}
$modified = false;
$children = $regularExpression->getChildren();
foreach ($children as &$child) {
$modified |= $this->optimize($child);
}
if ($modified) {
$regularExpression->setChildren($children);
}
return (bool) $modified;
}
/**
* Recursively filters out start and end markers
*
* Recursively filter out start and end markers from the regular expression
* structure, since they do not have any real meaning, but were required to
* create correct automatons.
*
* @param slRegularExpression $regularExpression
* @return slRegularExpression
*/
protected function filterStartEndMarkers(slRegularExpression $regularExpression)
{
if ($regularExpression instanceof slRegularExpressionMultiple) {
foreach ($regularExpression->getChildren() as $nr => $child) {
if ($child instanceof slRegularExpressionElement && ($child->getContent() === 0 || $child->getContent() === 1)) {
$regularExpression[$nr] = new slRegularExpressionEmpty();
} else {
$this->filterStartEndMarkers($child);
}
}
}
if ($regularExpression instanceof slRegularExpressionSingular) {
$child = $regularExpression->getChild();
if ($child instanceof slRegularExpressionElement && ($child->getContent() === 0 || $child->getContent() === 1)) {
$regularExpression->setChild(new slRegularExpressionEmpty());
} else {
$this->filterStartEndMarkers($child);
}
}
return $regularExpression;
}
/**
* Get all elements occureing in a regular expression
*
* @param slRegularExpression $regExp
* @return array
*/
protected function getChildElements(slRegularExpression $regExp)
{
if ($regExp instanceof slRegularExpressionElement) {
return array($regExp->getContent());
}
if (!$regExp instanceof slRegularExpressionContainer) {
return array();
}
$children = array();
foreach ($regExp->getChildren() as $child) {
$children = array_merge($children, $this->getChildElements($child));
}
return $children;
}
/**
* Extracts all types mentioned in a regular exression
*
* Returns an unique array with all types mentioned in the regular
* expression, to create proper mixed types in the DTD schema output.
*
* @param slRegularExpression $regularExpression
* @return array
*/
protected function extractTypes(slRegularExpression $regularExpression)
{
if ($regularExpression instanceof slRegularExpressionElement) {
return array($regularExpression->getContent());
}
if (!$regularExpression instanceof slRegularExpressionContainer) {
return array();
}
$types = array();
foreach ($regularExpression->getChildren() as $child) {
$types = array_merge($types, $this->extractTypes($child));
}
return array_unique($types);
}
