Automation/app/Libraries/MyCrawler/Mangboard/YamoonCrawler.php
2024-09-18 13:51:42 +09:00

126 lines
7.6 KiB
PHP

<?php
namespace App\Libraries\MyCrawler\Mangboard;
use App\Entities\Mangboard\UserEntity;
use Symfony\Component\DomCrawler\Crawler;
class YamoonCrawler extends MangboardCrawler
{
public function __construct(string $host, string $board_name, UserEntity $user_entity)
{
parent::__construct($host, $board_name, $user_entity);
}
//작성내용
// <div class="panel panel-default">
// <div class="text-center panel-heading-local-title text-bold">요즘 화제라는 명품 목걸이</div>
// <div style="margin:5px 10px;">
// <span class="pull-left dropdown">CAT7478</span>
// <span class="pull-right">| 추천 (8) | 조회 (268)</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">2024-09-16 09:52:39</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_18297097311726447898684.webp'" class="fr-fic fr-dii" src="/newboard/yamoonfreeboard/uploads/humor/mceu_18297097311726447898684.webp" alt=""></p>
// <p><img title="" class="cloudzoom" data-cloudzoom="zoomImage:'/newboard/yamoonfreeboard/uploads/humor/mceu_41105156321726447902977.png'" class="fr-fic fr-dii" src="/newboard/yamoonfreeboard/uploads/humor/mceu_41105156321726447902977.png" alt=""></p>
// <p>&nbsp;</p>
// <p>&nbsp;</p>
// <p>전화기선 짤라서 목걸이 만들어도 위화감이 전혀 없을것같은</p>
// <p>&nbsp;</p>
// <p>디자인이군요</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>
protected function detail_page(int $cnt, array $listInfo): array
{
$response = $this->getMySocket()->getContent("/newboard/yamoonboard/" . $listInfo['detail_url']);
//작성시간
// $selector = $this->getSelector($response, getenv("yamoon.view.date.tag"));
// $listInfo['date'] = trim($selector->text());
// if ($this->getDebug()) {
// throw new \Exception(
// sprintf(
// "\n--------------%s Debug--------------\n%s\n%s\n---------------------------------------\n",
// __FUNCTION__,
// var_export($listInfo, true),
// $selector->html()
// )
// );
// }
//작성내용
$tag = getenv("yamoon.view.content.tag");
$selector = $this->getSelector($response, $tag);
$media_urls = $this->getUrlsByMediaType($selector, "img", "src");
$media_urls = $this->getUrlsByMediaType($selector, "video", "src", $media_urls);
if ($this->getDebug()) {
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);
}
return $listInfo;
}
//리스트 내용
// <td class="listvisited mobile-td subject-view">
// <a href="board-read.asp?fullboardname=yamoonfreeboard&mtablename=humor&num=89372&ref=85575&page=1" class="ya-tooltip mobile-bold mobile-height" title="<p><br><br><video autoplay=&quot;autoplay&quot; loop=&quot;loop&quot; muted=&quot;&quot; controls=&quot;controls&quot; width=&quot;560&quot;&quot; height=&quot; &quot;> <source src=&quot; https://files.bepick.net/bbs/2024/09/c2a20ab5771cbb934940551859fce1c8_769966583.mp4 &quot;> </video><br><br><br></p">
// 졸고 있는 여군</a>
// <i class="fa fa-commenting-o" aria-hidden="true"></i> <span class="color-red small">6</span>
// <span class="visible-xs visible-sm small"><i class="fa fa-user-o" aria-hidden="true"></i> yeeyuu | <i class="fa fa-thumbs-o-up" aria-hidden="true"></i> 6 | <i class="fa fa-eye" aria-hidden="true"></i> 369 | No 89372 | 2024-09-13</span>
// </td>
public function execute(int $max_limit): void
{
try {
if ($this->getDebug()) {
$url = getenv("yamoon.view.test.url.{$this->_board_name}");
log_message("notice", __FUNCTION__ . "=> DEBUG 게시물 {$url} 작업시작");
$this->detail_page(1, ['detail_url' => $url]);
log_message("notice", __FUNCTION__ . "=> DEBUG 게시물 {$url} 작업종료");
} else {
$listInfos = [];
$response = $this->getMySocket()->getContent(getenv("yamoon.list.url.{$this->_board_name}"));
$this->getSelector($response, getenv("yamoon.list.tag"))->each(
function (Crawler $node) use (&$listInfos): void {
$link_node = $node->filter(getenv("yamoon.list.item.link.tag"));
$detail_url = $link_node->attr("href");
$title = $link_node->text();
$info_node = $node->filter(getenv("yamoon.list.item.info.tag"));
$infos = explode("|", $info_node->text());
$listInfos[] = ['title' => $title, 'detail_url' => $detail_url, 'nickname' => trim($infos[0]), 'hit' => trim($infos[2]), 'date' => trim($infos[4])];
}
);
if (!count($listInfos)) {
throw new \Exception("Target URL이 없습니다.");
}
$this->main_process($max_limit, $listInfos);
}
log_message("notice", __FUNCTION__ . " 작업이 완료되었습니다.");
} catch (\Exception $e) {
log_message("warning", sprintf(
"\n---%s 오류---\n%s\n-----------------------------------------\n",
__FUNCTION__,
$e->getMessage()
));
}
}
}