public function init()
{
scws_set_charset($this->tool, "utf8");
scws_set_dict($this->tool, self::ROOT . "etc/dict.utf8.xdb");
scws_set_rule($this->tool, self::ROOT . "etc/rules.utf8.ini");
scws_set_ignore($this->tool, true);
//清楚标点
scws_set_multi($this->tool, self::SCWS_MULTI_DUALITY);
#scws_set_multi($this->tool,self::SCWS_MULTI_SHORT);
scws_set_duality($this->tool, false);
}
/**
* get tops content of the input, get number of word from return
* @param string $content input content
* @param int $tops number of words
*/
public static function keywords($content, $type = '', $tops = 10)
{
$obj = scws_open();
//load dic path and rules
$dicpath = empty($type) ? C('dictionay.defaultdic') : C('dictionay.' . $type);
$rules = C('dictionay.defaultrule');
if (empty($dicpath) || empty($rules)) {
throw_exception("Load dictionary node: dic file Error");
}
//var_dump($type,$dicpath);
scws_set_charset($obj, 'utf8');
scws_set_dict($obj, $dicpath, SCWS_XDICT_TXT);
scws_set_duality($obj, true);
scws_set_rule($obj, $rules);
scws_send_text($obj, $content);
return scws_get_tops($obj, $tops);
}
/**
* 分词处理，取出词频最高的词组，并可以指定词性进行查找
*
* @param string $str
* @param int $limit
* 可选参数，返回的词的最大数量，缺省是 10
* @param string $attr
* 可选参数，是一系列词性组成的字符串，各词性之间以半角的逗号隔开， 这表示返回的词性必须在列表中，如果以~开头，则表示取反，词性必须不在列表中，缺省为NULL，返回全部词性，不过滤。
* @return multitype:
*/
function scwsTop($str, $limit = 10, $attr = null)
{
if (!function_exists('scws_open')) {
return false;
}
$rst = array();
$str = preg_replace("/[\\s\t\r\n]+/", '', $str);
if (!empty($str)) {
$sh = scws_open();
scws_set_charset($sh, 'utf8');
scws_set_ignore($sh, true);
scws_set_multi($sh, SCWS_MULTI_SHORT | SCWS_MULTI_DUALITY);
scws_set_duality($sh, true);
scws_send_text($sh, $str);
$rst = scws_get_tops($sh, $limit, $attr);
scws_close($sh);
}
return $rst;
}
