taoqu/App/Api/Model/Dataoke.php

<?php


namespace Heanup\App\Api\Model;


use QL\QueryList;
use function GuzzleHttp\Psr7\parse_query;

class Dataoke
{


    public static function parse($content)
    {
        $ql = QueryList::html($content);

// 元数据采集规则
        $rules = [
            'link' => ["a", "href"]
        ];
// 切片选择器
        $rt = $ql->find("a")->attrs('href');
        $data = $rt->all();
        $data = array_unique($data);
        foreach ($data as $url) {
//            if (strstr($url, "tao.php") === false) {
//                continue;
//            }
            $query=parse_url($url);

            $t = parse_query($query['query']);
            if ($t['kw']) {
                $q=[];
                if ($t['cid']){
                    $q['cid']=$t['cid'];
                }
                if ($t['pid']){
                    $q['pid']=$t['pid'];
                }
                if ($t['bids']){
                    $q['bids']=$t['bids'];
                }

                //elseif ($t['cid']){
//                    $new_url="/kw/" . $t['kw'] . ".html?cid=".$t['cid'];
//                } else{
//                    $d=$t;
//                    unset($d['kw']);
//                }
                $new_url="/kw/" . $t['kw'] . ".html?".http_build_query($q);

                $content = str_replace($url, $new_url, $content);
            }else{
                if ($t['cid']) {
                    $content = str_replace($url, "/category/" . $t['cid'] . ".html", $content);
                }else{
                    if ($t['id']) {
                        $content = str_replace($url, "/product/" . $t['id'] . ".html", $content);
                    }
                }

            }

            ///index.php?r=index
//            if ($url=="/index.php?r=index"){
//                $content = str_replace($url, "/" , $content);
//            }

        }
        $content=str_replace('"/tao.php?r=p"', '"/pages/fengqiang.html"', $content);
        $content=str_replace('"/tao.php?r=nine"', '"/pages/nine.html"', $content);
        $content=str_replace('"/tao.php?r=ddq"', '"/pages/dongdong.html"', $content);
        $content=str_replace('"/tao.php?r=l"', '"/pages/haohuo.html"', $content);
        $content=str_replace('"/tao.php?r=index/index&u=1350274"', '"/"', $content);
        $content=str_replace('href="/tao.php?r=index/index"', 'href="/"', $content);
        $content=str_replace('https://cmsstatic.ffquan.cn//js/new_home.js', "/static/new_home.js", $content);
//        $content=str_replace('index.php', "tao.php", $content);
        return $content;
    }

}