/**
* @param string $rule
* @return string
* @throws Exceptions\ParserException
*/
public function parse($rule)
{
$this->output = '';
$this->operator = \null;
$this->values = \null;
$this->operatorRequired = \false;
foreach (new AST($this->tokenizer->tokenize($rule), $this->variables) as $token) {
switch ($token->getGroup()) {
case Constants::GROUP_VALUE:
$this->assignVariableValueFromToken($token);
break;
case Constants::GROUP_LOGICAL:
$this->assignLogicalToken($token);
continue 2;
case Constants::GROUP_PARENTHESES:
$this->assignParentheses($token);
continue 2;
case Constants::GROUP_OPERATOR:
$this->assignOperator($token);
continue 2;
case Constants::GROUP_COMMENT:
case Constants::GROUP_SPACE:
continue 2;
default:
throw new Exceptions\ParserException(sprintf('Unknown token "%s" at position %d on line %d', $token->getValue(), $token->getPosition(), $token->getLine()));
}
$this->parseExpression();
}
$this->assertSyntaxSeemsOkay();
return $this->output;
}
/**
* Classifies a text and returns the probability (score) per label
*
* @param string $text
* @return array
*/
public function classify($text)
{
$totalDocCount = array_sum($this->docs);
$tokens = $this->tokenizer->tokenize($text);
$scores = array();
foreach ($this->labels as $label => $labelCount) {
$logSum = 0;
$docCount = $this->docs[$label];
$inversedDocCount = $totalDocCount - $docCount;
if (0 === $inversedDocCount) {
continue;
}
foreach ($tokens as $token) {
$totalTokenCount = isset($this->tokens[$token]) ? $this->tokens[$token] : 0;
if (0 === $totalTokenCount) {
continue;
}
$tokenCount = isset($this->data[$label][$token]) ? $this->data[$label][$token] : 0;
$inversedTokenCount = $this->inversedTokenCount($token, $label);
$tokenProbabilityPositive = $tokenCount / $docCount;
$tokenProbabilityNegative = $inversedTokenCount / $inversedDocCount;
$probability = $tokenProbabilityPositive / ($tokenProbabilityPositive + $tokenProbabilityNegative);
$probability = (1 * 0.5 + $totalTokenCount * $probability) / (1 + $totalTokenCount);
if (0 === $probability) {
$probability = 0.01;
} elseif (1 === $probability) {
$probability = 0.99;
}
$logSum += log(1 - $probability) - log($probability);
}
$scores[$label] = 1 / (1 + exp($logSum));
}
arsort($scores, SORT_NUMERIC);
return $scores;
}
/**
* @param string $string
* @return string
* @throws Exceptions\HighlighterException
*/
public function highlightString($string)
{
try {
$tokens = $this->tokenizer->tokenize($string);
} catch (Exceptions\TokenizerException $e) {
throw new Exceptions\HighlighterException('Unable to highlight string', 0, $e);
}
return $this->highlightTokens($tokens);
}
/**
* @param string $class
* @param string $regex
* @param int $priority
*/
public function registerToken($class, $regex, $priority = null)
{
$this->tokenizer->registerToken($class, $regex, $priority);
}