首页 > 代码库 > ThinkPHP3.2.3扩展之自动分词获取关键字

ThinkPHP3.2.3扩展之自动分词获取关键字

ThinkPHP自动获取关键词调用在线discuz词库

先按照下图路径放好插件

技术分享

        /**
     * 自动获取关键词(调用第三方插件)
     * @return [type] [description]
     * www.shouce.ren
     */
    public function keyword()
    {
        Vendor(‘autokeyword.AutoKeyword‘);
        $keyword = new \AutoKeyword();
        $str=‘自动获取关键词并发大数据我们大家好吃饭啦调用第三方插件‘;
        $title = trim($str);
        $keys=$keyword::discuz($title);
        var_dump($keys);
    }

插件源码:

AutoKeyword.php

<?php
 
/**
 * 分词
 * @author zhao jinhan <326196998@qq.com>
 * 
 */
class AutoKeyword
{
 
    /**
     * 使用discuz词库
     * @param unknown_type $title
     * @param unknown_type $content
     */
    public static function discuz ($title = ‘‘, $content = ‘‘)
    {
        $subjectenc = rawurlencode(strip_tags($title));
        $messageenc = rawurlencode(strip_tags(preg_replace("/\[.+?\]/U", ‘‘, $content)));
        $data = @implode(‘‘, file("http://keyword.discuz.com/related_kw.html?title=$subjectenc&content=$messageenc&ics=utf-8&ocs=utf-8"));
 
        if ($data) {
            $parser = xml_parser_create();
            xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
            xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
            xml_parse_into_struct($parser, $data, $values, $index);
            xml_parser_free($parser);
            $kws = array ();
            foreach ($values as $valuearray) {
                if ($valuearray[‘tag‘] == ‘kw‘ || $valuearray[‘tag‘] == ‘ekw‘)
                    $kws[] = trim($valuearray[‘value‘]);
            }
//             $return = ‘‘;
//             $dot=‘‘;
//             if ($kws) {
//                 foreach ($kws as $kw) {
//                     $kw = CHtml::encode(strip_tags($kw));
//                     $return .= $dot.$kw ;
//                     $dot = ‘,‘;
//                 }
//                 $return = trim($return);
//             }
 
            return $kws;
        }
 
    }

 

/**
     * 简易自定义获取关键词
     * @param string $title
     * @param string $content
     * @return JSON
     */
    public static function simple($title = ‘‘, $content=‘‘){
        $words = include_once dirname(__FILE__).DIRECTORY_SEPARATOR.‘words.php‘;
        $data = array();
        $max = 10; //最多匹配10个
        if($words){
            foreach((array)$words as $word){
                if(count($data) > $max){
                    break;
                }else{
                    if($title && preg_match(‘/‘.$word.‘/is‘, $title)){
                        array_push($data, $word);
                    }
                    if($content && preg_match(‘/‘.$word.‘/is‘, $content)){
                        if(!in_array($word, $data)){
                            array_push($data, $word);
                        }
                    }
                }
            }
        }
        return array_unique($data);
    }
 
 
}

word.php

<?php
/**
 * 词库(权重越高越靠前)
 */
return array(
    ‘php‘,
    ‘mysql‘,
    ‘web‘,
    ‘html‘,
    ‘js‘,
    ‘jquery‘,
    ‘sql‘,
    ‘myisam‘,
    ‘innodb‘,
    ‘apache‘,
    ‘nginx‘,
    ‘yii‘,
    ‘linux‘,
    ‘cms‘,
    ‘yiifcms‘,
    ‘redis‘,
    ‘memcache‘,
    ‘explain‘,
    ‘ueditor‘,
    ‘kindeditor‘,
    ‘api‘,
    ‘w3c‘,
    ‘高性能‘,
    ‘大数据‘,
    ‘存储过程‘,
    ‘事物‘,
    ‘触发器‘,
    ‘索引‘,
    ‘并发‘,
    ‘编程‘,
    ‘算法‘,
    ‘排序‘,
    ‘安装包‘,
    ‘下载‘,
    ‘服务器‘,
    ‘手册‘,
    ‘指南‘,
    ‘文章‘,
    ‘图集‘,
);

 

ThinkPHP3.2.3扩展之自动分词获取关键字