/**
* -- Add function info here --
*/
function getBayesianTokensFromArray($blogId, $tokens, $degenerate = true)
{
$bayesianTokens = array();
foreach ($tokens as $token) {
$bayesianToken = $this->getBayesianTokenFromToken($blogId, $token);
if (!$bayesianToken) {
if ($degenerate) {
$bayesianToken = $this->getFarthestToken($blogId, BayesianToken::degenerate($token));
} else {
$bayesianToken = new BayesianToken($blogId, $token, 0, 0, TOKEN_DEFAULT_PROBABILITY);
}
}
array_push($bayesianTokens, $bayesianToken);
}
return $bayesianTokens;
}
/**
* -- Add function info here --
*/
function degenerate($token = false)
{
if ($token === false) {
$token = $this->getToken();
}
$degenerations = array();
$curToken = $token;
$prefix = true;
while ($prefix) {
if (ereg("^(" . TOKEN_TOPIC_MARK . "|" . TOKEN_URL_MARK . "|" . TOKEN_USER_NAME_MARK . "|" . TOKEN_USER_EMAIL_MARK . "|" . TOKEN_USER_URL_MARK . ")(.*)\$", $curToken, $regs)) {
$curToken = $regs[2];
$prefix = $regs[1];
} else {
$prefix = "";
}
$degenerations = array_merge($degenerations, BayesianToken::getBasicsDegeneration($curToken, $prefix));
if (ereg("([^!]+!)!+\$", $curToken, $regs)) {
$degenerations = array_merge($degenerations, BayesianToken::getBasicsDegeneration($regs[1], $prefix));
}
if (ereg("([^!]+)!+\$", $curToken, $regs)) {
$degenerations = array_merge($degenerations, BayesianToken::getBasicsDegeneration($regs[1], $prefix));
}
if ($prefix == "" && $curToken == $token) {
foreach ($degenerations as $degeneration) {
array_push($degenerations, TOKEN_TOPIC_MARK . $degeneration);
array_push($degenerations, TOKEN_URL_MARK . $degeneration);
array_push($degenerations, TOKEN_USER_NAME_MARK . $degeneration);
array_push($degenerations, TOKEN_USER_EMAIL_MARK . $degeneration);
array_push($degenerations, TOKEN_USER_URL_MARK . $degeneration);
}
}
}
return array_unique($degenerations);
}
/**
* @private
*/
function _tokenizeHtmlTag($tag)
{
$tokens = array();
preg_match_all("/([^=]+)=\\s*([^\\s>]+)/", $tag, $regs);
$count = count($regs[1]);
//foreach ($regs[2] as $value)
for ($i = 0; $i < $count; $i++) {
$value = $regs[2][$i];
$prefix = "";
if (eregi("(href|src)", $regs[1][$i])) {
$prefix = TOKEN_URL_MARK;
}
$token = $this->_unquoteToken($value);
$tokensTemp = $this->_tokenize($token);
foreach ($tokensTemp as $tokenTemp) {
if (BayesianToken::isValid($tokenTemp)) {
array_push($tokens, $prefix . $tokenTemp);
}
}
}
return $tokens;
}