/**
* @private
*/
function getSpamProbability($blogId, $topic, $text, $userName, $userEmail, $userUrl)
{
$tokenizer = new BayesianTokenizer();
$tokensTopic = $tokenizer->addContextMark($tokenizer->tokenize($topic), TOKEN_TOPIC_MARK);
$tokensText = $tokenizer->tokenize($text);
$tokensUserName = $tokenizer->addContextMark($tokenizer->tokenize($userName), TOKEN_USER_NAME_MARK);
$tokensUserEmail = $tokenizer->addContextMark($tokenizer->tokenize($userEmail), TOKEN_USER_EMAIL_MARK);
$tokensUserUrl = $tokenizer->addContextMark($tokenizer->tokenize($userUrl), TOKEN_USER_URL_MARK);
$tokens = array_merge($tokensTopic, $tokensText, $tokensUserName, $tokensUserEmail, $tokensUserUrl);
$significantTokens = BayesianFilter::_getMostSignificantTokens($blogId, $tokens);
return BayesianFilter::_getBayesProbability($significantTokens);
}
/**
* Processes the pipeline, using the request and blogInfo
* objects as given in the constructor.
*/
function process()
{
global $_pLogPipelineRegisteredFilters;
// check if the pipeline is enabled
$config =& Config::getConfig();
if ($config->getValue("security_pipeline_enabled") == false) {
// pipeline is disabled, so everything's fine
return new PipelineResult(true);
}
// boolean to indicate whether we should run the bayesian filter
// This is a hack. We don't want to run the bayesian filter on
// a message that will be rejected by another filter. The only way
// to know this is to run after all of the other filters have run.
//
// Ideally, once a message is rejected, we would want to let all
// of the filters know about it. This way they could do something
// interesting (i.e. train the message as spam, report the ip address
// and urls to dnsbl services)
// default it to false.
$runBayesianFilter = false;
// if enabled, then check all the filters
foreach ($_pLogPipelineRegisteredFilters as $filterClass) {
if (strcmp($filterClass, "BayesianFilter") == 0) {
$runBayesianFilter = true;
}
// create an instance of the filter
$pipelineRequest = new PipelineRequest($this->_httpRequest, $this->_blogInfo);
$filter = new $filterClass($pipelineRequest);
// and execute it...
$result = $filter->filter();
// if there was an error, we better say so now
// and quite, making sure that we're keeping the
// error code
if (!$result->isValid()) {
$this->_result = $result;
return $result;
}
}
if ($runBayesianFilter) {
// create an instance of the filter
$pipelineRequest = new PipelineRequest($this->_httpRequest, $this->_blogInfo);
$filter = new BayesianFilter($pipelineRequest);
// and execute it...
$result = $filter->filter();
}
$this->_result = $result;
return $result;
}
