collector_producer/app/JsonRpc/CollectorService.php

<?php
namespace App\JsonRpc;

use App\Model\ArticleData;
use App\Model\OldModel\Article as OldArticle;
use App\Model\OldModel\ArticleData as OldArticleData;
use App\Model\OldModel\Category;
use App\Model\Article;
use App\Model\Rule;
use App\Model\Web;
use Hyperf\DbConnection\Db;
use Hyperf\Di\Annotation\Inject;
use Hyperf\RpcServer\Annotation\RpcService;
use App\Tools\Result;
use QL\QueryList;
use Swoole\Coroutine;
use App\Service\GatherQueueService;


use function Hyperf\Support\retry;

#[RpcService(name: "CollectorService", protocol: "jsonrpc-http", server: "jsonrpc-http")]
class CollectorService implements CollectorServiceInterface
{
    #[Inject]
    protected GatherQueueService $Gservice;
    /**
     * 添加网站
     * @param array $data
     * @return array|mixed
     */
    public function addWeb(array $data): array
    {
        $where = [
            'name' => $data['name']
        ];
        $isweb = Web::where($where)->first();
        if(empty($isweb)){
            date_default_timezone_set('Asia/Shanghai');
            $web = Web::insert($data);
            
        }else{
            return Result::error('此网站已存在，不可重复添加！');
        }
        if(empty($web)){
            return Result::error('添加失败');
        }
        return Result::success('添加成功');
    }
    /**
     * 获取并搜索网站
     * @param array $data
     * @return array|mixed
     */
    public function getWeb(array $data): array
    {
        
        if(isset($data['keyWord'])){
            $where = [
                ['name','like','%'.$data['keyWord'].'%']
            ];
            $rep = Web::where($where)->limit($data['pageSize'])->orderBy("created_at","desc")->offset(($data['page']-1)*$data['pageSize'])->get();
            $count =  Web::where($where)->count();
            if($count==0){
                return Result::error('未查找到相关网站！');
            }
        }else{
            $rep = Web::limit($data['pageSize'])->orderBy("created_at","desc")->offset(($data['page']-1)*$data['pageSize'])->get();
            $count =  Web::count();
            
        }
        $data = [
            'rep' => $rep->toArray(),
            'count' => $count
        ];
        if(empty($rep)){
            return Result::error('您还未添加网站，请先去添加！');
            
        }
        
        return Result::success($data);
    }
    /**
     * 修改网站
     * @param array $data
     * @return array|mixed
     */
    public function upWeb(array $data): array
    {
        
        $web = Web::where('id',$data['id'])->first();
        if(empty($web)){
            return Result::error('请输入正确的网站id！');
            
        }else{
            date_default_timezone_set('Asia/Shanghai');
            $id = Web::where('id',$data['id'])->update($data);
            if(empty($id)){
                return Result::error('无法修改！');
            }
        }
        return Result::success($id);
    }
    /**
     * 删除网站
     * @param array $data
     * @return array|mixed
     */
    public function delWeb(array $data): array
    {
        $web = Web::where('id',$data['id'])->first();
        if(empty($web)){
            return Result::error('请输入正确的网站id！');
            
        }else{
            $where = [
                ['web_id','=',$data['id']]
            ];
            //判断此网站下是否规则任务
            $rule = Rule::where($where)->get();
            if(empty($rule)){
                //若没有直接删除网站
                $result['web'] = Web::where('id',$data['id'])->delete();
            }else{
                //若有，判断规则任务是否有已执行的
                $rule = Rule::where($where)->where('status',2)->get();
                // return Result::success($rule);
                if(!empty($rule->toArray())){
                    //若有已执行的任务规则，不可删除网站
                    return Result::error('该网站已有成功执行的任务规则，不可删除！');
                }else{
                    try {
                        Db::beginTransaction();
                        //若无已执行的任务规则，删除网站及相应的未执行的规则任务
                        $result['web'] = Web::where('id',$data['id'])->delete();
                        $result['rule'] = Rule::where($where)->delete();
                        Db::commit();
                    } catch(\Throwable $ex){
                        Db::rollBack();
                        var_dump($ex->getMessage());
                        return Result::error("删除失败",0);
                    }
                }        
            }
        }
        return Result::success($result);
    }

    /**
<<<<<<< HEAD
     * 添加任务规则
     * @param array $data
     * @return array|mixed
     */
    public function addRule(array $data): array
    {
        $web = Web::where('id',$data['web_id'])->get();
        if(empty($web->toArray())){
            return Result::error('请输入正确的网站id！');
            
        }else{
            $rulename = Rule::where('name',$data['name'])->get();
            //查找是否存在规则名称重复的
            if(empty($rulename->toArray())){
                //（若是多类型参数一起传过来则根据类型，只获取对应类型需要的参数）
                switch($data['type']){
                    case 1:
                        $rule = [
                            'name' => $data['name'],
                            'web_id' => $data['web_id'],
                            'first_url' => $data['first_url'],
                            'second_start' => $data['second_start'],
                            'second_num' => $data['second_num'],
                            'second_end' => $data['second_end'],
                            'end_pagenum' => $data['end_pagenum'],
                            'start' => $data['start'],
                            'title' => $data['title'],
                            'content' => $data['content']
                        ];
                        // var_dump("============1============");       
                        break;         
                    case 2:
                        $rule = [
                            'name' => $data['name'],
                            'web_id' => $data['web_id'],
                            'first_url' => $data['first_url'],
                            'parameter' => $data['parameter'],
                            'start' => $data['start'],
                            'title' => $data['title'],
                            'content' => $data['content']
                        ];  
                        // var_dump("============2============");       
                        break;        
                    default:
                        $rule = [
                            'name' => $data['name'],
                            'web_id' => $data['web_id'],
                            'diy_rule' => $data['diy_rule']
                        ];        
                        // var_dump("============3============");       
                        break;              
                }
                if(!empty($data['source']) && $data['type'] != 3){

                    $rule ['source'] = $data['source'];
                }
                if(isset($data['writer_class']) && $data['type'] != 3){
                    $rule ['writer_class'] = $data['writer_class'];
                }
                if(isset($data['writer']) && $data['type'] != 3){
                    $rule ['writer'] = $data['writer'];
                }
                date_default_timezone_set('Asia/Shanghai');
                //若不存在，根据网站类型添加到不行类型的规则表中
                $result = Rule::insertGetId($rule);
                
            }else{
                return Result::error('此任务已存在！');
            } 
             
        }
        return Result::success($result);
    }
    /**
     * 获取并搜索规则任务
     * @param array $data
     * @return array|mixed
     */

    public function getRule(array $data): array
    {

        $where = [];
        
        if(isset($data['web_id'])){
            $web = Web::where('id',$data['web_id'])->get();
            if(empty($web->toArray())){
                return Result::error('请输入正确的网站id!');
                
            }else{
                //若是根据网站跳转到的规则任务则存到$where数组中
                $where = [
                    ['web_id','=', $data['web_id']]
                ];
            }
        } 

        if(isset($data['keyWord'])){
            //若存在搜索词,则存到条件数组$where中
            $where = [
                ['name','like','%'.$data['keyWord'].'%']
            ];
        }
        if(empty($where)){
            $rep = Rule::withCount(relations:'arts')->limit($data['pageSize'])->orderBy("created_at","desc")->offset(($data['page']-1)*$data['pageSize'])->get();
        }else{
            $rep = Rule::withCount(relations:'arts')->where($where)->limit($data['pageSize'])->orderBy("created_at","desc")->offset(($data['page']-1)*$data['pageSize'])->get();
        }
        
        $count = Rule::where($where)->count();
        if($count==0){
            return Result::error('暂无相关规则任务！');
        }       
        
        $data = [
            'rep' => $rep->toArray(),
            'count' => $count
        ];

        return Result::success($data);

    }
    

     /**
     * 获取某个任务规则
     * @param array $data
     * @return array|mixed
     */
    public function getOneRule(array $data): array
    {
        $result = Rule::where('id',$data['id'])->first();
        if(empty($result)){
            return Result::error('请输入正确的规则任务id！');
            
        }else{
            return Result::success($result);
        } 

    }
    /**
     * 修改规则任务
     * @param array $data
     * @return array|mixed
     */
    public function upRule(array $data): array
    {
        $rule = Rule::where('id',$data['id'])->select('id')->first();
        unset($data['type']);
        if(empty($rule)){
            return Result::error('请输入正确的规则任务id！');
            
        }else{
            $rulename = Rule::where('id','!=',$rule['id'])->where('name',$data['name'])->select('name')->first();
            if(empty($rulename)){
                $result = Rule::where('id',$data['id'])->update($data);
            }else{
                return Result::error('已存在此任务规则名称！');
            }
        }
       
        return Result::success($result);
    }
    /**
     * 删除规则任务
     * @param array $data
     * @return array
     */
    public function delRule(array $data): array
    {
        $where = ['id' => $data['rule_id']];
        $rule = Rule::where($where)->first();
        if(empty($rule)){
            return Result::error('请输入正确的规则任务id！');
            
        }else{
            //查找是否存在已导入的文章
            $art_num = Article::where('rule_id',$data['rule_id'])->where('state',1)->count();
            if($art_num==0){
                //查找是否存在已采集但是未导入的文章
                $colart_num = Article::where('rule_id',$data['rule_id'])->where('state',0)->count();
                if($colart_num==0){
                    $result['rule'] = Rule::where($where)->delete();
                }else{
                    try {
                        Db::beginTransaction();
                        //若有已采集但未导入的文章，删除规则任务及相应的未导入的文章
                        $result['rule'] = Rule::where($where)->delete();
                        $result['art'] = Article::where('rule_id',$data['rule_id'])->delete();
                        Db::commit();
                    } catch(\Throwable $ex){
                        Db::rollBack();
                        var_dump($ex->getMessage());
                        return Result::error("删除失败",0);
                    }
                }
            }else{
                return Result::error('此规则任务下的文章已导入，不可删除！');
            }
        }
        return Result::success($result);
    }

    /**
     * 开始采集
     * @param array $data
     * @return array
     */

    public function sendCrawler(array $data): array
    {
       $result =  $this->Gservice->push($data,rand(5,20));
       return  Result::success([$result]);
    }

    /**
     * @param array $data
     * @return array
     */
    public function goCrawler(array $data): array
    {
        //通过规则id 查询规则类型
        $where = [
            'rule.id'=>$data['id']
        ];
        $info = Rule::where($where)->leftJoin('web','rule.web_id','web.id')
            ->select("rule.*","web.name as web_name","web.url as web_url","web.type as web_type")
            ->first();
        $info = $info->toArray();

        switch ($info['web_type']){
            case 1:
                var_dump("===========规则采集======",$info);
                Rule::where(['id'=>$data['id']])->update(['status'=>1]);
                $data['copyfrom'] = $info['web_name'];
                $data['author'] = $info['writer'];;
                $data['first_url'] = $info['first_url'];
                $data['second_start'] = $info['second_start'];
                $data['second_num'] = $info['second_num'];
                $data['second_end'] = $info['second_end'];
                $data['end_pagenum']= $info['end_pagenum'];
                $data['rule_id']= $data['id'];
                $data['admin_user_id']= $data['admin_user_id'];
//                $data['newUrlStr'] =
                $urlList = $this->addUrlArr($data);
                if($urlList){
                    foreach ($urlList as $val){
//                        var_dump("单列表地址：",$val);
                        $this->ruleCollection($val,$data);
                    }
                }
                Rule::where(['id'=>$data['id']])->update(['status'=>2]);
                break;
            case 2:
                Rule::where(['id'=>$data['id']])->update(['status'=>1]);
                $wecUrl = $info['first_url'];//'https://www.ndcpa.gov.cn/queryList';
                $parames = json_decode($info['parameter'],true);

//                var_dump($parames);die;

                $parames['webSiteCode'] = [trim($parames['webSiteCode'], "[]")]; //['jbkzzx'];//
                $parames['channelCode'] =  [trim($parames['channelCode'], "[]")]; // ['c100008'];//
                $other = [
                    'web_url'=>$info['web_url'],
                    'copyfrom'=>$info['web_name'],
                    'admin_user_id'=>$data['admin_user_id'],
                    'rule_id'=>$data['id'],
                    'writer'=>$info['writer'],
                ];
                var_dump("=======开始接口采集====",$parames);
//                die;
                $this->foreachCurl($wecUrl,$parames,$other);

                Rule::where(['id'=>$data['id']])->update(['status'=>2]);



        }
        return  Result::success([]);
    }

    /**
     * 把可采集的列表页连接 打包成一个大数组
     * @return void
     */
    public function addUrlArr($data)
    {
        $arrList = [];
        array_push($arrList,$data['first_url']);

            $exit = false;
            $i = 0;
            while(!$exit){
                $i++;
                $url = $data['second_start'].$i.$data['second_end'];
                $respon1 = Result::pageExists($url);
//                var_dump("采集地址：",$respon1,$url);
//                Coroutine::sleep(2);
                if ($i==intval($data['end_pagenum'])-1) {
                    $exit = true;
//                    Coroutine::exit(); // 退出循环
                }else{
                    array_push($arrList,$url);
                }
            }


        return $arrList;
    }

    
    /**
     * 按照规则采集数据
     * @return void
     */
    public function ruleCollection($url,$data)
    {
//        var_dump("采集参数：",$data);
        $list = QueryList::get($url);
        $dataList = $list->rules([
            'title' => ['a', 'text'],
            'link'  => ['a', 'href'],
        ])->range('.list1 li')->query()->getData();
//        var_dump("采集的内容：",$dataList);
//        var_dump("====",$dataList);die;
        $firstUrlArr =  explode("/", $url);
        array_pop($firstUrlArr);
        $firstUrlArr = implode('/',$firstUrlArr);

        $dataList = $dataList->toArray();
        if($dataList){
            foreach ($dataList as $tiem){
                $newUrl =  substr($tiem['link'], 1);
                $newUrlStr = $firstUrlArr.$newUrl;
                $detailContent = QueryList::get($newUrlStr);
                $detailData = $detailContent->rules([
                    'title'=>['h1','text'],
                    'content'=>['.TRS_UEDITOR','html'],
                ])->range(".news-details")->query()->getData();

                $detailData = $detailData->toArray();
//                var_dump("内容详情：",$detailData,$newUrlStr);
                if($detailData){
                    foreach ($detailData as $val){
//                        var_dump("进没进foreach:",$newUrlStr,$val);
                        $data['fromurl'] = $newUrlStr;
                        $data['title'] = $val['title'];
                        $data['content'] = $val['content'];
                        $data['newUrlStr'] = $newUrlStr;
                        $data['source'] = '';
                        $data['introduce'] = $val['title']??'';
                        $data['keyword'] = $val['title']??'';
                        $data['copyfrom'] = $data['copyfrom'];
                        $data['source'] = $data['source']??$data['copyfrom'];
                        $data['admin_user_id'] = $data['admin_user_id']??'';
                        $data['rule_id'] = $data['rule_id']??'';
//                        $data['copyfrom'] = $data['copyfrom'];
//                        var_dump("要插入的数据：",$data);
                        $this->insertArticleData($data);
                    }
                }

            }
//
        }
    }

    /**
     * 插入数据
     * @param $data
     * @return void
     */
    public function insertArticleData($data=[])
    {
        if($data){
            Db::beginTransaction();
            try{
                $articleInfo =  Article::where(['title'=>$data['title']])->first();
//                var_dump("获取详情：",$articleInfo,$data);
                if(empty($articleInfo)){
                    $insertData = [];
                    $insertData['fromurl'] =$data['newUrlStr'];
                    $insertData['oldtitle'] =$data['title'];
                    $insertData['title'] = $data['title'];
                    $insertData['copyfrom'] =  $data['copyfrom'];
                    $insertData['author'] = $data['author'];
                    $insertData['introduce'] = $data['title'];
                    $insertData['keyword'] = $data['title'];
                    $insertData['source'] = isset($data['source']) && $data['source']!=''? $data['source']:$data['copyfrom'];
                    $insertData['admin_user_id'] = $data['admin_user_id'];
                    $insertData['rule_id'] = $data['rule_id'];
//                    var_dump("插入Article：",$insertData);
                    $article_id = Article::insertGetId($insertData);
                    $insertDataDetail = [];
                    $insertDataDetail['article_id'] = $article_id;
                    $insertDataDetail['content'] = $data['content'];
//                    var_dump("插入ArticleData：",$insertDataDetail);
                    ArticleData::insertGetId($insertDataDetail);
//                        Coroutine::sleep(2);
//                    var_dump("插入成功一次：",$article_id,$insertDataDetail);
                }
                Db::commit();
            }catch (\Exception $e){
                Db::rollBack();
                var_dump("插入失败：",$e->getMessage());
            }

        }else{
            var_dump("没有数据可以插入：");
        }

    }

    /**
     * 分页采集
     * @return void
     */
    public function foreachCurl($wecUrl,$parames,$other,&$page=1)
    {
        $options = [
            CURLOPT_HEADER => true, // 如果想包含头部信息在响应中，可以设置为true
            CURLOPT_TIMEOUT => 30 // 设置请求超时时间为30秒
        ];
        $result = Result::http_post($wecUrl,$parames,$options);
        $result = json_decode($result['response'],true);
//        var_dump("获取数据：",$result);
        if($result['data'] && $result['data']['results']){
            $dataList  = $result['data']['results'];
//            var_dump("取数据结构体：",$dataList);
            foreach ($dataList as $val){
//                var_dump("进入循环插入：",$val);
                $newUrlStr = json_decode($val['source']['urls'],true);
                $newUrlStr = $other['web_url'].$newUrlStr['common'];
//                var_dump("来源地址：",$newUrlStr);
                $insertData = [
                    'newUrlStr'=>$newUrlStr,
                    'title'=>$val['source']['title']??'',
                    'source'=>$val['source']['contentSource']??'',
                    'copyfrom'=>$other['copyfrom']??'',
                    'content'=>$val['source']['content']['content']??'',
                    'admin_user_id'=>$other['admin_user_id']??'',
                    'rule_id'=>$other['rule_id']??'',
                    'author'=>$other['writer']??''
                ];
//                var_dump("调用插入数据方法,组装数据：",$insertData);
                $this->insertArticleData($insertData);
            }
        }
        $pages = intval($parames['current']);
        $pages =  $pages+1;
        $parames['current'] = $pages;
        $twoResult = Result::http_post($wecUrl,$parames,$options);
        if($result['data'] && $result['data']['results'] && count($result['data']['results'])>0){
//            var_dump("分页测试：",$parames,$parames['current']);
            $this->foreachCurl($wecUrl,$parames,$other,$pages);
        }
//        var_dump("正确的数据：",$result);
    }
    /**
     * 获取并搜索资讯
     * @param array $data
     * @return array
     */
    public function getInfo(array $data): array
    {
        $where = [
            ['rule_id','=',$data['rule_id']]
        ];
        //若存在条件参数都存到where数组
        if(isset($data['title'])){
            $where[] = ['title','like','%'.$data['title'].'%'];
        }
        if(isset($data['source'])){
            $art_source = Article::where($where)->get();
            if(!empty($art_source->toArray())){
                $where[] = ['source','=',$data['source']];
            } 
        }
        if(isset($data['state'])){
            $where[] = ['state','=',$data['state']];
        }
        //跨库查询栏目导航及采集的新闻
        $info = Article::query()
        ->where($where)
        ->with('category')
        ->orderBy("article.id","desc")
        ->limit($data['pageSize'])
        ->offset(($data['page']-1)*$data['pageSize'])->get();
        $count = Article::where($where)->count();
        if($count == 0){
            return Result::error('暂无资讯');
        }
        $data = [
            'rep' => $info->toArray(),
            'count' => $count
        ];
        return  Result::success($data);
    }
    /**
     * 获取某个资讯
     * @param array $data
     * @return array
     */
    public function getOneInfo(array $data): array
    {
        $where = ['id' => $data['art_id']];
        $inf = Article::where($where)->first();
        if($inf==null){
            return Result::error('请输入正确的资讯id！');
        }
        $info =  Article::where($where)
        ->leftJoin('article_data','article_id','id')
        ->select('article.*','article_data.content')
        ->first();
        if($inf['catid']!=null){
            $category = Category::where(['id'=>$info['catid']])->select('name')->first();
            $info['category'] = $category['name'];
        }
        
        return  Result::success($info);
    }
    /**
     * 修改资讯
     * @param array $data
     * @return array
     */
    public function upInfo(array $data): array
    {
        $id = $data['art_id'];
        $content = $data['content'];
        unset($data['art_id']);
        //去掉此元素
        unset($data['content']);
        //去掉此元素
        $info =  Article::where('id',$id)->first();  
        if($info==null){
            return Result::error('请输入正确的文章id！');
        }
        if($info['state']==1){
            return Result::error('此文章已导入 ，不可编辑！');
        }else{
            Db::beginTransaction();
            try{
                $info =  Article::where('id',$id)->update($data);  
                $art_data = ArticleData::where('article_id',$id)->update(['content'=>$content]);

                Db::commit();
            } catch(\Throwable $ex){
                Db::rollBack();
                var_dump($ex->getMessage());
                return Result::error("修改失败",0);
            }
            $data = [
                'info' => $info,
                'art_data' => $art_data
            ];
            return  Result::success($data);
        }
        
    }
    /**
    * 删除资讯
    * @param array $data
    * @return array
    */
    public function delInfo(array $data): array
    {
        $id = $data['art_id'];
        $info = Article::where('id',$id)->first();
        if($info==null){
            return Result::error('请输入正确的文章id！');
        }
        if($info['state']==1){
            return Result::error('此文章已导入，不可删除！'); 
        }else{
            Db::beginTransaction();
            try{
                $delinfo = Article::where('id',$id)->delete();
                $deldata = ArticleData::where('article_id',$id)->delete();
                Db::commit();
            } catch(\Throwable $ex){
                Db::rollBack();
                var_dump($ex->getMessage());
                return Result::error("删除失败",0);
            }
        }
        $data = [
            'delinfo' => $delinfo,
            'deldata' => $deldata
        ];
        return  Result::success($data);
    }
    /**
    * 关联导航池
    * @param array $data
    * @return array
    */
    public function addCatid(array $data): array
    {
        $id = $data['rule_id'];
        $art = Article::where('rule_id',$id)->select('id')->count();
        if($art==0){
            return Result::error('还未采集，请采集');
        }else{
            $info = Article::where('rule_id',$id)->where('state',0)->select('id')->get();
        if(empty($info->toArray())){
            return Result::error('所有文章都已导入，不可修改关联的导航池！'); 
        }else{
            //查找此规则任务下的文章是否已经有导入的文章
            $article = Article::where('rule_id',$id)->where('state',1)->select('id')->get();
            if(!empty($article->toArray())){
                //若有已导入的文章则直接复制之前已导入的导航池
                $catid = Article::whereIn('id',$article)->select('catid')->first();
                //若未导入的文章已经复制之前的导航，则无需修改
                $art_catid = Article::whereIn('id',$info)->whereNull('catid')->count();
                if($art_catid>0){
                    $result = Article::whereIn('id',$info)->update(['catid'=>$catid['catid']]);
                }else{
                    $result = ['已全部关联导航，无需再次关联！'];
                }
            }else{
                //若不存在已导入的文章则判断是否存在导航id
                if(isset($data['catid'])){
                    //若存在直接使用此导航id
                    $result = Article::whereIn('id',$info)->update(['catid'=>$data['catid']]);
                    
                }else{
                    //若不存在则返回所有导航栏目
                    $result = Category::select('id','name')->get();
                }
            }
              
        }
        }
        //查找此规则任务下的文章是否都已经导入
        
        return  Result::success($result);
    }
    /**
    * 导入文章
    * @param array $data
    * @return array
    */
    public function addArt(array $data): array
    {
        // var_dump("======@@@====");
        $where = [
            'rule_id' => $data['rule_id'],
            'state' => 0
        ];
        //获取某个规则任务下的已采集未导入的文章及文章详情
        $arts_id = Article::where($where)->wherenotNull('catid')->orderBy('id')->select('id')->get();
        $arts = Article::where($where)->wherenotNull('catid')->select('title','catid','level','introduce','keyword','author','copyfrom','fromurl','hits','islink','imgurl','admin_user_id','is_original')->orderBy('id')->get()->toArray();       
        $arts_data = ArticleData::whereIn('article_id',$arts_id)->select('content')->orderBy('article_id')->get()->toArray();
        // var_dump($article_data);
        $data = [
            'articles' => $arts,
            'art_content' => $arts_data
        ];
        Db::beginTransaction();
        try{
           
            $oldart = OldArticle::insert($arts);
            $oldart_data = OldArticleData::insert($arts_data);
            $upstate_art = Article::where($where)->wherenotNull('catid')->update(['state' => 1]);
            Db::commit();
            
        } catch(\Throwable $ex){
            Db::rollBack();
            var_dump($ex->getMessage());
            return Result::error($ex->getMessage(),0);
        }
        return Result::success($data);
    }
}