public function testConvertPaperExample()
{
// Example 2. Let W = {abccde, cccad, bfegg, bfehi}.
$automaton = new slCountingSingleOccurenceAutomaton();
$automaton->learn(array('a', 'b', 'c', 'c', 'd', 'e'));
$automaton->learn(array('c', 'c', 'c', 'a', 'd'));
$automaton->learn(array('b', 'f', 'e', 'g', 'g'));
$automaton->learn(array('b', 'f', 'e', 'h', 'i'));
$converter = new slChareConverter();
$regexp = $converter->convertAutomaton($automaton);
$this->assertEquals(new slRegularExpressionSequence(new slRegularExpressionRepeatedAtLeastOnce(new slRegularExpressionChoice(new slRegularExpressionElement('a'), new slRegularExpressionElement('b'), new slRegularExpressionElement('c'))), new slRegularExpressionChoice(new slRegularExpressionElement('d'), new slRegularExpressionElement('f')), new slRegularExpressionOptional(new slRegularExpressionElement('e')), new slRegularExpressionRepeated(new slRegularExpressionElement('g')), new slRegularExpressionOptional(new slRegularExpressionElement('h')), new slRegularExpressionOptional(new slRegularExpressionElement('i'))), $regexp);
}
/**
* Build regular expression
*
* Builds the regilar expression from the provided automaton, where each
* node associates with one of the equivalency classes, and a topologically
* sorted list of these nodes.
*
* The provided automaton must be an instance of the
* slCountingSingleOccurenceAutomaton, so t at it provides information on
* how often the token occur in the learned strings.
*
* @param slCountingSingleOccurenceAutomaton $automaton
* @param array $classes
* @return slRegularExpression
*/
protected function buildRegularExpression(slCountingSingleOccurenceAutomaton $automaton, array $classes)
{
// xsd:all may only occur outmost and makes only sense for equivalence
// classes with more then one elment.
if (count($classes) === 1) {
$class = reset($classes);
$term = $eClasses = $this->equivalenceClasses[$class];
$count = $automaton->getOccurenceSum($eClasses);
$nodes = $automaton->getNodes();
$generalCount = $automaton->getGeneralOccurences($eClasses);
if (count($term) > 1 && $generalCount['max'] === 1 && $count['max'] > 1) {
// Inference all
$term = new slRegularExpressionAll(array_map(function ($term) use($nodes) {
return new slRegularExpressionElement($nodes[$term]);
}, $term));
$term->minOccurences = $generalCount['min'];
return $term;
}
}
return parent::buildRegularExpression($automaton, $classes);
}
/**
* Convert automaton to regular expression
*
* @param slAutomaton $automaton
* @return slRegularExpression
*/
protected function convertRegularExpression($automaton)
{
// Convert automatons
$converter = new slSoreConverter();
if (($expression = $converter->convertAutomaton($automaton)) !== false) {
return $expression;
}
$converter = new slChareConverter();
return $converter->convertAutomaton($automaton);
}
