/**
* [keyword 获取关键词]
* @param [type] $text [文本]
* @param string $filter [过滤条件]
* @return [type] [description]
*/
public function keyword($text, $filter = '')
{
$filter = $this->makeFilter($filter);
vendor('pscws4.class#pscws4');
$pscws = new PSCWS4();
$pscws->set_charset('utf-8');
$pscws->send_text($text);
$some = $pscws->get_tops(10, $filter);
$_result = $this->makeAttr($some);
$filter_reg = explode(',', str_replace('~', '', $filter));
if (!empty($filter)) {
if ($filter != "~") {
foreach ($_result as $k => $v) {
if (in_array($v['attr'], $filter_reg)) {
$_res[] = $v;
}
}
} else {
$_res = $_result;
}
} else {
$_res = $_result;
}
return $_res;
}
function get_tags_arr($title)
{
$pscws = new PSCWS4();
$pscws->set_dict(APP_ROOT . '/scws/dict.utf8.xdb');
$pscws->set_rule(APP_ROOT . '/scws/rules.utf8.ini');
$pscws->set_ignore(true);
$pscws->send_text($title);
$words = $pscws->get_tops(10);
$tags = array();
foreach ($words as $val) {
$tags[] = $val['word'];
}
$pscws->close();
return $tags;
}
/**
* @param $str 要解析的字符串
* @param $num 获取的词的数量
* @func 实现分词功能
*/
public function run($str, $num)
{
$pscws = new \PSCWS4();
$pscws->set_dict(APP_ROOT . '/dict.utf8.xdb');
$pscws->set_rule(APP_ROOT . '/rules.utf8.ini');
$pscws->set_ignore(true);
$pscws->send_text($str);
$words = $pscws->get_tops($num);
$tags = array();
foreach ($words as $val) {
$tags[] = $val['word'];
}
$pscws->close();
return $tags;
}
public function get_tags_by_title($title, $num = 10)
{
vendor('pscws4.pscws4', '', '.class.php');
$pscws = new PSCWS4();
$pscws->set_dict(PIN_DATA_PATH . 'scws/dict.utf8.xdb');
$pscws->set_rule(PIN_DATA_PATH . 'scws/rules.utf8.ini');
$pscws->set_ignore(true);
$pscws->send_text($title);
$words = $pscws->get_tops($num);
$pscws->close();
$tags = array();
foreach ($words as $val) {
$tags[] = $val['word'];
}
return $tags;
}
public function ajax_getKeywords($title, $num = 5)
{
import('ORG.Pscws.Pscws4');
$pscws = new PSCWS4();
$pscws->set_dict(LIBRARY_PATH . 'ORG/Pscws/dict.utf8.xdb');
$pscws->set_rule(LIBRARY_PATH . 'ORG/Pscws/rules.utf8.ini');
$pscws->set_ignore(true);
$pscws->send_text($title);
$words = $pscws->get_tops($num);
$pscws->close();
$list = array();
foreach ($words as $value) {
$list[] = $value['word'];
}
echo json_encode(array('num' => count($list), 'list' => $list));
}
function getKeyword($title)
{
require './pscws.class.php';
$pscws = new PSCWS4();
$pscws->set_dict('./scws/dict.utf8.xdb');
$pscws->set_rule('./scws/rules.utf8.ini');
$pscws->set_ignore(true);
$pscws->send_text($title);
$words = $pscws->get_tops();
$res = '';
foreach ($words as $val) {
$res .= '|' . $val['word'];
}
$pscws->close();
return substr($res, 1);
}
public static function segments($arr, $num = 10)
{
$list = array();
if (empty($text)) {
return $list;
}
$words = array();
//检测是否已安装php_scws扩展
if (function_exists("scws_open")) {
$sh = scws_open();
scws_set_charset($sh, 'utf8');
scws_set_dict($sh, APP_ROOT_PATH . 'system/scws/dict.utf8.xdb');
scws_set_rule($sh, APP_ROOT_PATH . 'system/scws/rules.utf8.ini');
scws_set_ignore($sh, true);
foreach ($arr as $key => $text) {
scws_send_text($sh, $text);
$words[] = scws_get_tops($sh, $num);
}
scws_close($sh);
} else {
require_once APP_ROOT_PATH . 'system/scws/pscws4.class.php';
$pscws = new PSCWS4();
$pscws->set_dict(APP_ROOT_PATH . 'system/scws/dict.utf8.xdb');
$pscws->set_rule(APP_ROOT_PATH . 'system/scws/rules.utf8.ini');
$pscws->set_ignore(true);
foreach ($arr as $key => $text) {
$pscws->send_text($text);
$words[] = $pscws->get_tops($num);
}
$pscws->close();
}
for ($i = 0; $i < $num; $i++) {
foreach ($words as $item) {
if (isset($item[$i])) {
$word = $item[$i]['word'];
if (isset($list[$word])) {
$list[$word]++;
} else {
$list[$word] = 1;
}
}
}
}
$list = array_slice($list, 0, $num);
return array_keys($list);
}
function getkeyword($title, $contents = '')
{
// 加入头文件
require_once 'pscws4.class.php';
// 建立分词类对像, 参数为字符集, 默认为 gbk, 可在后面调用 set_charset 改变
$pscws = new PSCWS4('utf8');
$pscws->set_dict('../function/etc/dict.xdb');
$pscws->set_rule('../function/etc/rules.ini');
$pscws->set_duality(true);
$text = $title;
$pscws->send_text($text);
$tops = $pscws->get_tops(10, '');
foreach ($tops as $k) {
$keywords = $keywords . $k[word] . ' ';
}
return ' ' . trim($keywords);
}
public function get_tags_by_title($title, $num = 10)
{
//vendor('pscws4.pscws4', '', '.class.php');
import("Extend.pscws4.pscws4", APP_PATH . 'Lib/');
///分组后
/// 分组前 import("@.Extend.pscws4.pscws4");
// dump(APP_PATH . 'Extend/pscws4/scws/dict.utf8.xdb');
$pscws = new \PSCWS4();
$pscws->set_dict(APP_PATH . 'Lib/' . 'Extend/pscws4/scws/dict.utf8.xdb');
$pscws->set_rule(APP_PATH . 'Lib/' . 'Extend/pscws4/scws/rules.utf8.ini');
$pscws->set_ignore(true);
$pscws->send_text($title);
$words = $pscws->get_tops($num);
$pscws->close();
$tags = array();
foreach ($words as $val) {
$tags[] = $val['word'];
}
return $tags;
}
public function get_tags($title, $num = 10)
{
vendor('Pscws.Pscws4', '', '.class.php');
$pscws = new PSCWS4();
$pscws->set_dict(CONF_PATH . 'etc/dict.utf8.xdb');
$pscws->set_rule(CONF_PATH . 'etc/rules.utf8.ini');
$pscws->set_ignore(true);
$pscws->send_text($title);
$words = $pscws->get_tops($num);
$pscws->close();
$tags = array();
foreach ($words as $val) {
$tags[] = $val['word'];
}
return implode(',', $tags);
}
function get_keywords($title)
{
require './Common/scws/pscws4.class.php';
$pscws = new PSCWS4();
$pscws->set_dict('./Common/scws/scws/dict.utf8.xdb');
$pscws->set_rule('./Common/scws/scws/rules.utf8.ini');
$pscws->set_ignore(true);
$pscws->send_text($title);
$words = $pscws->get_tops(5);
$tags = array();
foreach ($words as $val) {
$tags[] = $val['word'];
}
$pscws->close();
return $tags;
}
//$cws->set_duality(true);
$cws->send_text($text);
if (php_sapi_name() != 'cli') {
header('Content-Type: text/plain');
}
echo "pscws version: ", $cws->version(), "\n";
echo "Segment result:\n\n";
while ($tmp = $cws->get_result()) {
$line = '';
foreach ($tmp as $w) {
if ($w['word'] == "\r") {
continue;
}
if ($w['word'] == "\n") {
$line = rtrim($line, ' ') . "\n";
} else {
$line .= $w['word'] . " ";
}
}
echo $line;
}
// top:
echo "Top words stats:\n\n";
$ret = array();
$ret = $cws->get_tops(10, 'r,v,p');
echo "No.\tWord\t\t\tAttr\tTimes\tRank\n------------------------------------------------------\n";
$i = 1;
foreach ($ret as $tmp) {
printf("%02d.\t%-16s\t%s\t%d\t%.2f\n", $i++, $tmp['word'], $tmp['attr'], $tmp['times'], $tmp['weight']);
}
$cws->close();
function get_keywords3($title, $num = 10)
{
$title = urldecode_utf8($title);
$title = trim_html($title, 1);
if (strlen($title) > 2400) {
$title = cutstr($title, 800, '');
}
include_once ROOT_PATH . 'web/lib/pscws4/pscws4.class.php';
$cws = new PSCWS4('utf8');
$cws->set_dict(ROOT_PATH . 'web/lib/pscws4/dict.utf8.xdb');
//$cws->set_dict(ROOT_PATH.'web/lib/pscws4/a.xdb');
$cws->set_rule(ROOT_PATH . 'web/lib/pscws4/rules.ini');
//$cws->set_multi(3);
$cws->set_ignore(true);
//$cws->set_debug(true);
//$cws->set_duality(true);
$cws->send_text($title);
$words = $cws->get_tops(10, 'r,v,p');
$cws->close();
$tags = array();
foreach ($words as $val) {
$tags[] = $val['word'];
}
$tags = implode(',', $tags);
return $tags;
}
private function get_tags($title, $num = 10)
{
vendor("Pscws.Pscws4", "", ".class.php");
$pscws = new PSCWS4();
$pscws->set_dict(CONF_PATH . "etc/dict.utf8.xdb");
$pscws->set_rule(CONF_PATH . "etc/rules.utf8.ini");
$pscws->set_ignore(true);
$pscws->send_text($title);
$words = $pscws->get_tops($num);
$pscws->close();
$tags = array();
foreach ($words as $val) {
$tags[] = $val["word"];
}
return implode(",", $tags);
}
