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

<?php

namespace App\Controllers\Mangboard\Crawler;

use Symfony\Component\DomCrawler\Crawler;
use Psr\Log\LoggerInterface;
use CodeIgniter\HTTP\ResponseInterface;
use CodeIgniter\HTTP\RequestInterface;

class InvenCrawler extends MyCrawler
{
    public function initController(RequestInterface $request, ResponseInterface $response, LoggerInterface $logger)
    {
        parent::initController($request, $response, $logger);
    }
    final protected function getHost(): string
    {
        return getenv("inven.host.url");
    }
    protected function getUrlByMediaType(Crawler $node, string $media_type, string $attr): null|string
    {
        switch ($media_type) {
            case 'video':
                $url = parent::getUrlByMediaType($node, $media_type, $attr);
                //그래도 null이면 data-src로 추출해본다.
                if ($url === null) {
                    $attributes = $node->extract(['data-src']);
                    if (count($attributes)) {
                        $url = $attributes[0];
                    }
                }
                break;
            case 'img':
            default:
                $url = parent::getUrlByMediaType($node, $media_type,  $attr);
                break;
        }
        return $url;
    }
    //작성내용
    // <div class="articleContent">
    // <div id="imageCollectDiv" class="contentBody">
    //     <!-- ============== CONTENT ============== -->
    //     <div id="powerbbsContent">
    //         <div id="BBSImageHolderTop" style="text-align:center;">
    //              <img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1620925350.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 1080 / 1350;" loading="lazy" />
    //              <br><br><img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1587803007.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 1080 / 1350;" loading="lazy" />
    //              <br><br><img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1134295360.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 1080 / 1350;" loading="lazy" />
    //              <br><br><img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1481352611.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 1080 / 1350;" loading="lazy" />
    //              <br><br><img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1878651605.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 850 / 1063;" loading="lazy" />
    //              <br><br>
    //         </div>
    //         <div>^^</div>
    //     </div>
    //     <!-- ============== End CONTENT ============== -->
    // </div>    
    protected function getDetailSelector(array $listInfo): array
    {
        $response = $this->getMySocket()->getContent($listInfo['detail_url']);
        return array($this->getSelector($response, getenv("inven.view.content.tag")), $listInfo);
    }
    //리스트내용
    // <div class="board-list">
    //     <table>
    //         <tr class="lgtm">
    //             <td class="num"><span>1589</span></td>
    //             <td class="tit">
    //                 <div class="text-wrap">
    //                     <div>
    //                         <span class="user-icon">
    //                             <img src="https://upload3.inven.co.kr/upload/2024/06/12/icon/i1237935053.jpg" alt="유저 아이콘" loading="lazy">
    //                         </span>
    //                         <a class="subject-link" href="https://www.inven.co.kr/board/party/5951/1589">
    //                             <span class="board_name">[사진＆움짤]</span>스테이씨 윤
    //                         </a>
    //                     </div>
    //                     <span data-opinion-bbs-comeidx="5951" data-opinion-bbs-uid="1589" data-opinion-bbs-opi="1" class="con-comment">[1]</span>                                                        
    //                     <span class="con-icon board-img photo">사진</span>
    //                 </div>
    //             </td>
    //             <td class="user">
    //                 <img src="https://static.inven.co.kr/image_2011/member/level/1202/lv32.gif" alt="레벨 아이콘">
    //                 <span class="layerNickName" onclick="layerNickName('배수민', 'pbNickNameHandler'); ">배수민</span>                                            
    //             </td>
    //             <td class="date">09-15</td>
    //             <td class="view">1,502</td>
    //             <td class="reco">1</td>
    //         </tr>
    //     </table>
    // </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");
                $listInfo['detail_url'] = getenv("inven.view.test.url.{$this->getBoardName()}");
                $listInfos[] = $listInfo;
            } else {
                $response = $this->getMySocket()->getContent(getenv("inven.list.url.{$this->getBoardName()}"));
                $this->getSelector($response, getenv("inven.list.tag.{$this->getBoardName()}"))->each(
                    function (Crawler $node) use (&$listInfos): void {
                        $hit      = $node->filter(getenv("inven.list.item.hit.tag"))->text();
                        $date     = date("Y") . "-" . $node->filter(getenv("inven.list.item.date.tag"))->text();
                        $nickname = $node->filter(getenv("inven.list.item.nickname.tag"))->text();
                        $link_node   = $node->filter(getenv("inven.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.");
            }
            $this->list_process(intval(getenv("inven.list.max_limit.{$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()
            ));
        }
    }
}