Automation/app/Libraries/MyCrawler/YamapLibrary.php
2024-09-07 15:13:03 +09:00

126 lines
4.9 KiB
PHP

<?php
namespace App\Libraries\MyCrawler;
use Symfony\Component\DomCrawler\Crawler;
use App\Libraries\MySocket\WebLibrary;
use App\Libraries\MyStorage\FileLibrary;
class YamapLibrary extends MyCrawlerLibrary
{
private $_mySocket = null;
private $_myStorage = null;
public function __construct()
{
parent::__construct();
}
public function getMySocket()
{
if ($this->_mySocket === null) {
$this->_mySocket = new WebLibrary(getenv('yamap.host.url'));
}
return $this->_mySocket;
}
public function getMyStorage()
{
if ($this->_myStorage === null) {
$this->_myStorage = new FileLibrary(getenv('yamap.storage.upload.path'));
}
return $this->_myStorage;
}
private function mainPage(string $url): array
{
$crawler = $this->getContent($url, getenv("yamap.list.tag"));
$item_tag = getenv("yamap.list.item.tag");
$item_link_tag = getenv("yamap.list.item.link.tag");
$item_nickname_tag = getenv("yamap.list.item.nickname.tag");
$item_nickname_except = getenv("yamap.list.item.nickname.except");
$lists = [];
//div.bbs_item를 가진 객체를 찾아서 같은 형식의 객체(sibling)를 배열로 넘김
$crawler->filter($item_tag)->each(
function (Crawler $node) use (
&$item_link_tag,
&$item_nickname_tag,
&$item_nickname_except,
&$lists
): void {
//bbs_item에서 span.g_nickname 객체를 찾아서 작성자가 "관리자" 아닌지 확인 후 Return Bool
$nickname = $node->filter($item_nickname_tag)->text();
log_message("debug", $item_nickname_tag . ":" . $nickname);
if ($nickname != $item_nickname_except) {
//작성자가 "관리자"가 아니 게시물이면 해당 bbs_item에서 a.list_subject 객체를 찾아서
$link_node = $node->filter($item_link_tag);
$url = $link_node->attr("href");
$title = $link_node->children()->last()->text();
$lists[] = ['title' => $title, 'nickname' => $nickname, 'url' => $url];
}
}
);
if (!count($lists)) {
throw new \Exception("Target URL이 없습니다.");
}
return array($lists[0]["title"], $lists[0]["nickname"], $lists[0]["url"]);
}
private function detailPage(string $url): array
{
$crawler = $this->getContent($url, getenv("yamap.view.content.tag"));
$mediaInfos = [];
$mediaTags = [];
//3. Image 처리
$downloadInfos = $this->download($crawler, ["tag" => "img", "attr" => "src"]);
foreach ($downloadInfos as $downloadInfo) {
if ($this->getMySocket()->isContainsHttpOrHttps($downloadInfo['orignal'])) {
$mediaTags[] = $downloadInfos['orignal'];
} else {
$mediaTags[] = sprintf(
"<img src=\"/%s/%s/%s\" alt=\"%s\">",
$this->getMyStorage()->getUploadPath(),
$downloadInfo["path"],
$downloadInfo["fileName"],
$downloadInfo["fileName"]
);
};
$mediaInfos[] = $downloadInfo;
}
//4. Video(mp4) 처리
$downloadInfos = $this->download($crawler, ["tag" => "video", "attr" => "src"]);
foreach ($downloadInfos as $downloadInfo) {
if ($this->getMySocket()->isContainsHttpOrHttps($downloadInfo['orignal'])) {
$mediaTags[] = $downloadInfos['orignal'];
} else {
$mediaTags[] = sprintf(
"<video src=\"/%s/%s/%s\" alt=\"%s\" controls autoplay>",
$this->getMyStorage()->getUploadPath(),
$downloadInfo["path"],
$downloadInfo["fileName"],
$downloadInfo["fileName"]
);
};
$mediaInfos[] = $downloadInfo;
}
log_message("debug", "-----mediaTags-----");
log_message("debug", var_export($mediaTags, true));
return array($mediaInfos, $mediaTags);
}
public function execute(): array
{
//1. 해당사이트 MainPage 처리
if ($this->getDebug()) {
$title = getenv("yamap.view.test.title");
$nickname = getenv("yamap.view.test.nickname");
$detail_url = getenv("yamap.view.test.url");
} else {
list($title, $nickname, $detail_url) = $this->mainPage(getenv("yamap.list.url"));
}
//2. DetailPage 처리 : bbs_view > div.contents 가진 객체를 찾아서 처리
list($mediaInfos, $mediaTags) = $this->detailPage($detail_url);
return array($title, $nickname, $mediaInfos, $mediaTags);
}
}