Automation/app/Controllers/Mangboard/Crawler/YamapCrawler.php

<?php

namespace App\Controllers\Mangboard\Crawler;

use App\Controllers\Mangboard\CrawlerController;
use CodeIgniter\HTTP\RequestInterface;
use CodeIgniter\HTTP\ResponseInterface;
use Psr\Log\LoggerInterface;
use Symfony\Component\DomCrawler\Crawler;

class YamapCrawler extends CrawlerController
{
    public function initController(RequestInterface $request, ResponseInterface $response, LoggerInterface $logger)
    {
        parent::initController($request, $response, $logger);
    }
    final protected function getHost(): string
    {
        return getenv("yamap.host.url");
    }
    protected function detail_content_process(int $cnt, array $listInfo): array
    {
        $response   = $this->getMySocket()->getContent($listInfo['detail_url']);
        $selector   = $this->getSelector($response, getenv("yamap.view.content.tag"));
        $formDatas = [];
        $formDatas['image_path'] = "";
        $formDatas['content'] = $selector->html();
        //File DB 및 Board DB 등록작업등
        $this->getBoardModel()->createByCrawler(
            $this->getBoardsEntity(),
            $this->getUserEntity(),
            $cnt,
            $listInfo,
            [],
            $formDatas
        );
        log_message("notice", __FUNCTION__ . " 작업이 완료되었습니다.");
        return $listInfo;
    }
    protected function detail_download_process(int $cnt, array $listInfo): array
    {
        $response   = $this->getMySocket()->getContent($listInfo['detail_url']);
        $selector   = $this->getSelector($response, getenv("yamap.view.content.tag"));
        $media_urls = $this->getUrlsByMediaType($selector, "img", "src");
        $media_urls = $this->getUrlsByMediaType($selector, "video", "src", $media_urls);
        if ($this->isDebug) {
            throw new \Exception(sprintf(
                "\n--------------%s Debug--------------\n%s%s\n---------------------------------------\n",
                __FUNCTION__,
                var_export($listInfo, true),
                var_export($media_urls, true)
            ));
        } else {
            // Image 나 Video 소스들의 url을 가져와서 실제 다운받는 처리
            $storages = $this->media_process($media_urls);
            if (!count($storages)) {
                throw new \Exception("등록할 자료가 없습니다.");
            }
            $this->backend_process($cnt, $listInfo, $storages);
        }
        log_message("notice", __FUNCTION__ . " 작업이 완료되었습니다.");
        return $listInfo;
    }
    //리스트내용
    // 				<div class="panel panel-default">
    // 					<div class="text-center panel-heading-local-title text-bold">요즘 패션</div>
    // 					<div style="margin:5px 10px;">
    // 						<span class="pull-left dropdown">
    // 						괴강고귀
    // 						</span>
    // 						<span class="pull-right">
    // 							 | 추천 (14) | 조회 (432)
    // 						</span>
    // 						<div class="clearfix"></div>
    // 						<hr class="hr-xs-xs">
    // 						<span>
    // 							<a href="javascript:void(0);" id="incfont"><i class="fa fa-plus fa-fw" aria-hidden="true"></i></a><a href="javascript:void(0);" id="decfont"><i class="fa fa-minus fa-fw margin-left-5" aria-hidden="true"></i></a>
    // 						</span>
    // 						<span class="pull-right">2024-09-14 01:53:45
    // 						</span>
    // 						<div class="clearfix"></div>
    // 						<hr class="margin-top-5 margin-bottom-20">
    // 						<div class="fr-view margin-bottom-30" id="read-content" style="word-break:break-all;">
    // 							<p><img title="" class="cloudzoom" data-cloudzoom="zoomImage:'/newboard/yamoonfreeboard/uploads/humor/mceu_86177012011726246415487.jpg'" class="fr-fic fr-dii" src="/newboard/yamoonfreeboard/uploads/humor/mceu_86177012011726246415487.jpg" alt=""></p>
    // <p>&nbsp;</p>
    // 						</div>
    // 					</div>
    // 					<div class="margin-10">
    // 							<a href="javascript:void(0)" onclick="javascript:window.open('https://twitter.com/intent/tweet?text='+encodeURIComponent(document.title)+'%20-%20'+encodeURIComponent(document.URL), 'twittersharedialog', 'menubar=no,toolbar=no,resizable=yes,scrollbars=yes,height=300,width=600');return false;" target="_blank"> <i class="fa fa-twitter-square fa-lg ya-tooltip" title="트위터 공유하기"></i></a>
    // 							<a href="javascript:void(0)" onclick="javascript:window.open('https://www.facebook.com/sharer/sharer.php?u='+encodeURIComponent(document.URL)+'&t='+encodeURIComponent(document.title), 'facebooksharedialog', 'menubar=no,toolbar=no,resizable=yes,scrollbars=yes,height=300,width=600');return false;" target="_blank"> <i class="fa fa-facebook-square fa-lg ya-tooltip" title="페이스북 공유하기"></i></a>
    // 					</div>
    // 					<div id="freesubframe"></div>
    // 				</div>  
    public function execute(string $board_name, string $user_id = null, ...$params): void
    {
        try {
            //추가옵션
            $this->isDebug = in_array('debug', $params);
            $this->isCopy = in_array('copy', $params);
            $this->setBoardName($board_name);
            $this->login_process($user_id);
            //실행
            $listInfos   = [];
            if ($this->isDebug) {
                $listInfo = [];
                $listInfo['title'] = 'test_title';
                $listInfo['nickname'] = 'test_name';
                $listInfo['hit'] = 1;
                $listInfo['date']   = date("Y-m-d H:i:s");
                $listInfo['detail_url'] = getenv("yamap.view.test.url.{$this->getBoardName()}");
                $listInfos[] = $listInfo;
            } else {
                $response = $this->getMySocket()->getContent(getenv("yamap.list.url.{$this->getBoardName()}"));
                $selector = $this->getSelector($response, getenv("inven.list.tag.{$this->getBoardName()}"));
                $selector->filter(getenv("yamap.list.item.tag"))->each(
                    function (Crawler $node) use (&$listInfos): void {
                        $hit      = $node->filter(getenv("yamap.list.item.hit.tag"))->text();
                        $date     = date("Y") . "-" . $node->filter(getenv("yamap.list.item.date.tag"))->text();
                        $nickname = $node->filter(getenv("yamap.list.item.nickname.tag"))->text();
                        //작성자가 "관리자"가 아닌 게시물이면 해당 bbs_item에서 a.list_subject 객체를 찾아서
                        $link_node   = $node->filter(getenv("yamap.list.item.link.tag"));
                        $detail_url  = $link_node->attr("href");
                        $title       = $link_node->text();
                        $listInfos[] = ['title' => $title, 'nickname' => $nickname, 'detail_url' => $detail_url, 'date' => $date, 'hit' => $hit];
                    }
                );
            }
            if (!count($listInfos)) {
                throw new \Exception("Target URL이 없습니다.");
            }
            $this->list_process(intval(getenv("yamap.list.max_limit.{$this->getBoardName()}")), $listInfos);
            log_message("notice",   __FUNCTION__ . " 작업이 완료되었습니다.");
        } catch (\Exception $e) {
            log_message("warning", sprintf(
                "\n---%s 오류---\n%s\n-----------------------------------------\n",
                __FUNCTION__,
                $e->getMessage()
            ));
        }
    }
}