public function scws($key)
{
if (function_exists('scws')) {
$so = scws_new();
$so->set_charset('utf-8');
$so->add_dict(ini_get('scws.default.fpath') . '/dict.utf8.xdb');
//自定义词库
//$so->add_dict(APP_PATH . '/library/dict/scws.txt', SCWS_XDICT_TXT);
$so->set_rule(ini_get('scws.default.fpath') . '/rules.utf8.ini');
$so->set_ignore(true);
$so->set_multi(true);
$so->set_duality(true);
} else {
require_once APP_PATH . '/library/scws/pscws4.class.php';
$so = new PSCWS4('utf-8');
$so->set_dict(APP_PATH . '/library/scws/etc/dict.utf8.xdb');
$so->set_rule(APP_PATH . '/library/scws/etc/rules.utf8.ini');
$so->set_multi(true);
$so->set_ignore(true);
$so->set_duality(true);
}
$keys = str_replace(array(" ", " ", "\t", "\n", "\r"), array("", "", "", "", ""), $key);
$so->send_text($keys);
$words_array = $so->get_result();
$words = '';
foreach ($words_array as $v) {
$words = $words . '|"' . $v['word'] . '"';
}
$so->close();
return $words = trim($words, '|');
}
/**
* [fenci 获取分词属性]
* @param [type] $text [description]
* @return [type] [description]
*/
public function fenci($text)
{
vendor('pscws4.class#pscws4');
$pscws = new PSCWS4();
$pscws->set_charset('utf-8');
$pscws->send_text($text);
$pscws->set_ignore(true);
$some = $pscws->get_result();
return $this->makeAttr($some);
}
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);
}
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;
}
/**
* 分词
* @param string $source_string 源字符串
* @return array 分好的单词数组
*/
private function wordSplit($source_string)
{
$words = array();
$pscws_path = APPPATH . 'third_party/pscws4/';
require_once $pscws_path . 'pscws4.class.php';
$dicts = array('UTF-8' => $pscws_path . 'dict.utf8.xdb');
$rules = array('UTF-8' => $pscws_path . 'etc/rules.utf8.ini');
$charset = strtoupper($this->config->item('charset'));
$pscws = new PSCWS4($charset);
$dict = $dicts[$charset];
$rule = $rules[$charset];
$pscws->set_dict($dict);
$pscws->set_rule($rule);
$pscws->set_ignore(true);
$pscws->send_text($source_string);
for ($some_words = $pscws->get_result(); $some_words !== false; $some_words = $pscws->get_result()) {
foreach ($some_words as $one_word) {
$words[] = $one_word['word'];
}
}
$pscws->close();
return $words;
}
public function get_tags($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;
}
/**
* 文本分词
* @param string $text 需要分词的文本
* @return array
*/
public static function segmentAll($text)
{
$list = array();
if (empty($text)) {
return $list;
}
//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/rules.utf8.ini');
scws_set_ignore($sh, true);
scws_send_text($sh, $text);
while ($words = scws_get_result($sh)) {
foreach ($words as $word) {
$list[] = $word['word'];
}
}
scws_close($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);
$pscws->send_text($text);
while ($words = $pscws->get_result()) {
foreach ($words as $word) {
$list[] = $word['word'];
}
}
$pscws->close();
}
return $list;
}
function pscws()
{
$pscws_path = APPPATH . 'third_party/pscws4/';
require_once $pscws_path . 'pscws4.class.php';
$pscws = new PSCWS4('utf8');
$pscws->set_dict($pscws_path . 'dict.utf8.xdb');
$pscws->set_rule($pscws_path . 'etc/rules.utf8.ini');
$pscws->set_ignore(true);
$text = '我是华东政法大学的学生';
echo $text;
$pscws->send_text($text);
$words = array();
while ($some = $pscws->get_result()) {
foreach ($some as $one) {
array_push($words, $one['word']);
}
}
var_dump($words);
$display_text = implode(' ', $words);
echo $display_text;
$pscws->close();
$this->load->sidebar_loaded = true;
}
function get_keywords3($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);
}
