hebei-chengben-car/modules/admin/controllers/KaoyanController.php

<?php
namespace app\modules\admin\controllers;
use app\common\controllers\BController;
use Yii;
class KaoyanController extends BController
{

    public $layout = 'main';
    //数据采集
    public function actionData()
    {
        $do = Yii::$app->request->get('do');

        //采集学校数据
        if($do=='school')
        {
            $sql = "select * from {{%collect_content}} where status=1 order by id asc limit 1";
            $schoolResult = Yii::$app->db->createCommand($sql)->queryOne();
            if(empty($schoolResult)){exit("导入院校完成");}
            $schoolInfo = string2array($schoolResult['data']);
            $sql = "insert into {{%ky_school}} set region_name='".$schoolInfo['region']."',title='".$schoolInfo['title']."',special='".$schoolInfo['special']."',logo='".addslashes($schoolInfo['logo'])."',owner='".$schoolInfo['owner']."'";
            Yii::$app->db->createCommand($sql)->execute();
            $sql = "update {{%collect_content}} set status=2 where id = ".$schoolResult['id'];
            Yii::$app->db->createCommand($sql)->execute();
            echo "<script>";
            echo "window.location.reload();";
            echo "</script>";
            exit;

        }

        //采集专业数据
        if($do=='major')
        {
            $sql = "select * from {{%ky_major}} where status=0 and parent_id=0 order by id asc limit 1";
            $parentMajor = Yii::$app->db->createCommand($sql)->queryOne();
            if(empty($parentMajor)){exit("采集专业完成");}
            $parentCode = $parentMajor['type'].$parentMajor['code'];
            $data = ['method'=>'subCategoryMl','key'=>$parentCode];
            $result = https_request('https://yz.chsi.com.cn/zyk/specialityCategory.do',[],$data);
            preg_match_all("/<li[^>]*id='(.*?)'[^>]*>(.*)<\/li>/i",$result,$matches);
            foreach($matches[1] as $k=>$v)
            {
                $code = substr($v,2);
                $title = str_replace('&#xe6a2;','',strip_tags($matches[2][$k]));
                $sql = "insert into {{%ky_major}} set status=0,parent_id=".$parentMajor['id'].",code='".$code."',type='".$parentMajor['type']."',title='".$title."'";
                Yii::$app->db->createCommand($sql)->execute();
            }
            $sql = "update {{%ky_major}} set status=1 where id = ".$parentMajor['id'];
            Yii::$app->db->createCommand($sql)->execute();
            echo "<script>";
            echo "window.location.reload();";
            echo "</script>";
            exit;

        }

        //采集专业数据
        if($do=='childmajor')
        {
            $sql = "select * from {{%ky_major}} where status=0  order by id asc limit 1";
            $parentMajor = Yii::$app->db->createCommand($sql)->queryOne();
            if(empty($parentMajor)){exit("采集专业完成");}
            $parentCode = $parentMajor['type'].$parentMajor['code'];
            $data = ['method'=>'subCategoryXk','key'=>$parentCode];
            $result = https_request('https://yz.chsi.com.cn/zyk/specialityCategory.do',[],$data);
            $result = $this->compressHtml($result);
            preg_match_all('/<tr.*?>(.*?)<\/tr>/im',$result,$matches);
            foreach($matches[1] as $k=>$v)
            {
                if($k==0)continue;
                preg_match_all('/<td.*?>(.*?)<\/td>/im',$v,$matchesChild);
                $code = $matchesChild[1][1];
                $title = str_replace('&#xe6a2;','',strip_tags($matchesChild[1][0]));
                $sql = "insert into {{%ky_major}} set status=1,parent_id=".$parentMajor['id'].",code='".$code."',type='".$parentMajor['type']."',title='".$title."'";
                Yii::$app->db->createCommand($sql)->execute();
            }
            $sql = "update {{%ky_major}} set status=1 where id = ".$parentMajor['id'];
            Yii::$app->db->createCommand($sql)->execute();
            echo "<script>";
            echo "window.location.reload();";
            echo "</script>";
            exit;
        }

        //采集院校专业数据
        if($do=='schoolmajor')
        {
            $sql = "select * from {{%ky_major}} where status=0  order by id asc limit 1";
            $major = Yii::$app->db->createCommand($sql)->queryOne();
            if(empty($major)){exit("采集完成");}
            $url =  "https://yz.chsi.com.cn/zyk/specialityDetail.do?zymc=".urlencode($major['title'])."&zydm=".$major['code']."&cckey=".$major['type']."&ssdm=&method=distribution";

            $result = $this->compressHtml(https_request($url));
            if(strpos($result,'很抱歉')===false)
            {
                $result = $this->cut_html($result,'<div class="tab-container zyk-zyfb-tab">','<div class="zyk-right">');
                preg_match_all("/<li[^>]*>(.*?)<\/li>/i",$result,$matches);
                if(!empty($matches[1]))foreach($matches[1] as $schoolTit)
                {
                    $schoolName = strip_tags($schoolTit);
                    $sql = "insert into {{%ky_school_major}} set school_title='".$schoolName."',major_title='".$major['title']."',major_code='".$major['code']."'";
                    Yii::$app->db->createCommand($sql)->execute();
                }
            }


            $sql = "update {{%ky_major}} set status=1 where id = ".$major['id'];
            Yii::$app->db->createCommand($sql)->execute();
            echo "<script>";
            echo "window.location.reload();";
            echo "</script>";
            exit;
        }
        //采集专业目录
        if($do=='majordir')
        {
            $sql = "select * from {{%ky_school_major}} where status=0  order by id asc limit 1";
            $unit = Yii::$app->db->createCommand($sql)->queryOne();
            if(empty($unit)){exit("采集完成");}
            $url = "https://yz.chsi.com.cn/zsml/querySchAction.do?dwmc=".urlencode($unit['school_title'])."&mlmc=&yjxkdm=".$unit['major_code'];
            $result = $this->compressHtml(https_request($url));
            $result = $this->cut_html($result,'<table class="ch-table more-content">','<div class="zsml-page-box">');
            if(strpos($result,'很抱歉')===false)
            {
                preg_match_all('/<a(.*?)href=[\'"](.*?)[\'"](.*?)>(.*?)<\/a>/i',$result,$matches);
                if(!empty($matches[4]))foreach($matches[4] as $k=>$linkTit)
                {
                    if($linkTit=='查看')
                    {
                        $url = 'https://yz.chsi.com.cn'.$matches[2][$k];
                        $sql = "insert into {{%ky_major_dir}} set school_title='".$unit['school_title']."',major_code='".$unit['major_code']."',url='".$url."'";
                        Yii::$app->db->createCommand($sql)->execute();
                    }


                }

            }

            $sql = "update {{%ky_school_major}} set status=1 where id = ".$unit['id'];
            Yii::$app->db->createCommand($sql)->execute();
            echo "<script>";
            echo "window.location.reload();";
            echo "</script>";
            exit;

        }

        //采集专业目录详情
        if($do=='dirdetail')
        {
            $sql = "select * from {{%ky_major_dir}} where status=0  order by id asc limit 1";
            $unit = Yii::$app->db->createCommand($sql)->queryOne();
            if(empty($unit)){exit("采集完成");}
            $url = $unit['url'];
            $result = $this->compressHtml(https_request($url));
            //$result = $this->cut_html($result,'<table class="ch-table more-content">','<div class="zsml-page-box">');
            $exam_way = $this->cut_html($result,'<td class="zsml-title">考试方式：</td><td class="zsml-summary">','</td>');
            $direction = $this->cut_html($result,'<td class="zsml-title">研究方向：</td><td class="zsml-summary">','</td>');
            $learn_way = $this->cut_html($result,'<td class="zsml-title">学习方式：</td><td class="zsml-summary">','</td>');
            $student_num = $this->cut_html($result,'<td class="zsml-title">拟招人数：</td><td class="zsml-summary">','</td>');
            $exam_content = $this->cut_html($result,'<div class="zsml-result">','</div><!-- 考试范围 end -->');
            $exam_content = str_replace('<span class="sub-msg">见招生简章</span>','',$exam_content);
            $sql = "update {{%ky_major_dir}} set status=1,exam_way='".$exam_way."',direction='".addslashes($direction)."',learn_way='".$learn_way."',student_num='".$student_num."',exam_content='".addslashes($exam_content)."' where id = ".$unit['id'];
            Yii::$app->db->createCommand($sql)->execute();

            echo "<script>";
            echo "window.location.reload();";
            echo "</script>";
            exit;

        }

        //采集科目
        if($do=='course')
        {
            $sql = "select * from {{%ky_major_dir}} where status=0 order by id asc limit 1";
            $unit = Yii::$app->db->createCommand($sql)->queryOne();
            if(empty($unit)){exit("采集完成");}
            $result = $unit['exam_content'];
            $result = str_replace('<td colspan="4"><div><span>或</span><hr class="zsml-line"></div></td>','',$result);
            $result = preg_replace('/<span[^>]*class="sub-msg"[^>]*>(.*?)<\/span>/i','',$result);
            preg_match_all('/<td.*?>(.*?)<\/td>/im',$result,$matches);
            if($matches[1])foreach($matches[1] as $course)
            {
                $short_name = str_replace(['(',')'],'',$course);
                $short_name = explode("（",$short_name);
                $short_name = $short_name[0];
                $sql = "select * from {{%ky_course}} where short_name='".$short_name."'";
                $row =  Yii::$app->db->createCommand($sql)->queryOne();
                if(empty($row))
                {
                    $sql = "insert into {{%ky_course}} set course='".$course."',short_name='".$short_name."'";
                    Yii::$app->db->createCommand($sql)->execute();
                }
            }

            $sql = "update {{%ky_major_dir}} set status=1 where id = ".$unit['id'];
            Yii::$app->db->createCommand($sql)->execute();
            echo "<script>";
            echo "window.location.reload();";
            echo "</script>";
            exit;

        }

    }
    //压缩html
    public  function compressHtml($str){
        $str = trim($str);
        $str = str_replace("\t","",$str);
        $str = str_replace("\r\n","",$str);
        $str = str_replace("\r","",$str);
        $str = str_replace("\n","",$str);
        $str=preg_replace("/>[ ]+/si",">",$str); //过滤>(">"号后面带空格)
        $str=preg_replace("/[ ]+</si","<",$str); //过滤<("<"号前面带空格)
        return trim($str);
    }
    public  function cut_html($html, $start, $end) {
        //if(strpos($html,$start)===false) return false;
        if (empty($html)||empty($start)||empty($end)) return false;
        $html = str_replace(array("\r", "\n"), "", $html);
        $start = str_replace(array("\r", "\n"), "", $start);
        $end = str_replace(array("\r", "\n"), "", $end);
        $html = explode(trim($start), $html);
        if(is_array($html)&&isset($html[1])) $html = explode(trim($end), $html[1]);
        return $html[0];
    }

}