289 lines
13 KiB
PHP
289 lines
13 KiB
PHP
<?php
|
|
|
|
namespace App\Libraries\MyCrawler;
|
|
|
|
|
|
use App\Libraries\MyCrawlerLibrary;
|
|
use App\Libraries\MySocket\WebSocket;
|
|
use App\Libraries\MyStorage\MangboardStorage;
|
|
use App\Libraries\Mangboard\BoardsLibrary;
|
|
use App\Libraries\Mangboard\BoardLibrary;
|
|
use App\Libraries\Mangboard\FileLibrary;
|
|
use App\Libraries\Mangboard\ImageLibrary;
|
|
use App\Entities\Mangboard\UserEntity;
|
|
use Symfony\Component\DomCrawler\Crawler;
|
|
use App\Traits\FileTrait;
|
|
|
|
class YamapCrawler extends MyCrawlerLibrary
|
|
{
|
|
use FileTrait;
|
|
private $_mySocket = null;
|
|
private $_myStorage = null;
|
|
private $_storages = [];
|
|
private $_category = "";
|
|
private $_user_entity = null;
|
|
private $_boards_library = null;
|
|
private $_board_library = null;
|
|
private $_file_library = null;
|
|
private $_image_library = null;
|
|
public function __construct(string $category)
|
|
{
|
|
parent::__construct();
|
|
$this->_category = $category;
|
|
}
|
|
public function getMySocket()
|
|
{
|
|
if ($this->_mySocket === null) {
|
|
$this->_mySocket = new WebSocket(getenv('yamap.host.url'));
|
|
}
|
|
return $this->_mySocket;
|
|
}
|
|
public function getMyStorage()
|
|
{
|
|
if ($this->_myStorage === null) {
|
|
$this->_myStorage = new MangboardStorage($this->getCategory());
|
|
}
|
|
return $this->_myStorage;
|
|
}
|
|
public function getBoardsLibrary(): BoardsLibrary
|
|
{
|
|
// $test = $this->getBoard();
|
|
// echo "TEST:{$test}\n";
|
|
// $temp = getenv("mangboard.storage.{$this->getBoard()}.name");
|
|
// echo "Temp:{$temp}\n";
|
|
// exit;
|
|
if ($this->_boards_library === null) {
|
|
$this->_boards_library = new BoardsLibrary(
|
|
$this->getCategory(),
|
|
$this->getUserEntity()
|
|
);
|
|
}
|
|
return $this->_boards_library;
|
|
}
|
|
public function getBoardLibrary(): BoardLibrary
|
|
{
|
|
if ($this->_board_library === null) {
|
|
$this->_board_library = new BoardLibrary(
|
|
$this->getBoardsLibrary()->getEntity(),
|
|
$this->getUserEntity()
|
|
);
|
|
}
|
|
return $this->_board_library;
|
|
}
|
|
public function getFileLibrary(): FileLibrary
|
|
{
|
|
if ($this->_file_library === null) {
|
|
$this->_file_library = new FileLibrary(
|
|
$this->getBoardsLibrary()->getEntity(),
|
|
$this->getUserEntity()
|
|
);
|
|
}
|
|
return $this->_file_library;
|
|
}
|
|
public function getImageLibrary(): ImageLibrary
|
|
{
|
|
if ($this->_image_library === null) {
|
|
$this->_image_library = new ImageLibrary();
|
|
}
|
|
return $this->_image_library;
|
|
}
|
|
public function getUserEntity(): UserEntity
|
|
{
|
|
if ($this->_user_entity === null) {
|
|
throw new \Exception("사용자정보가 없습니다.");
|
|
}
|
|
return $this->_user_entity;
|
|
}
|
|
public function setUserEntity(UserEntity $user_entity): void
|
|
{
|
|
$this->_user_entity = $user_entity;
|
|
}
|
|
public function getCategory(): string
|
|
{
|
|
if ($this->_category == "") {
|
|
throw new \Exception("저장할 Category가 정의되지 않았습니다.");
|
|
}
|
|
return $this->_category;
|
|
}
|
|
private function save(int $file_sequence, string $mediaType, string $file_name, string $content): void
|
|
{
|
|
log_message("debug", __FUNCTION__ . " 원본파일 {$file_name} 작업 시작");
|
|
$this->getMyStorage()->setOriginName($file_name);
|
|
$this->getMyStorage()->setOriginContent($content);
|
|
$this->getMyStorage()->setOriginType($mediaType);
|
|
$this->getMyStorage()->setOriginSequence($file_sequence);
|
|
$this->_storages[] = $this->getMyStorage()->save();
|
|
}
|
|
//Yamap ViewPage의 이미지나영상데이터가 있으면 Dodownload 한다.
|
|
private function download(string $mediaType, string $url): array
|
|
{
|
|
$file_names = explode('/', $url);
|
|
if (!is_array($file_names) || !count($file_names)) {
|
|
throw new \Exception("URL이 파일명 형식이 아닙니다 : " . $this->getMySocket()->getHost() . $url);
|
|
}
|
|
$file_name = array_pop($file_names);
|
|
$temps = explode(".", $file_name);
|
|
$file_ext = array_pop($temps);
|
|
if (!$this->isFileType_FileTrait($file_ext, $mediaType)) {
|
|
throw new \Exception("파일명 형식이 {$mediaType}가 아닙니다");
|
|
}
|
|
$content = $this->getMySocket()->getContent($url);
|
|
log_message("notice", "{$file_name} 파일이 다운로드되었습니다!");
|
|
return array($file_name, $content);
|
|
}
|
|
private function mediaContent(array $urls): void
|
|
{
|
|
$file_sequence = 1;
|
|
$this->_storages = []; //CreateBoard에서 사용을 위해 DetailPage마다 초기화
|
|
// log_message("debug", var_export($urls, true));
|
|
foreach ($urls as $mediaType => $media_urls) {
|
|
foreach ($media_urls as $url) {
|
|
try {
|
|
if ($url === null) {
|
|
continue;
|
|
}
|
|
list($file_name, $content) = $this->download($mediaType, $url);
|
|
$this->save($file_sequence, $mediaType, $file_name, $content);
|
|
$file_sequence++;
|
|
log_message("notice", __FUNCTION__ . " OriginType->{$mediaType} 작업 완료");
|
|
} catch (\Exception $e) {
|
|
log_message("warning", sprintf(
|
|
"\n---%s,OriginType->%s 오류---\n%s\n-----------------------------------------\n",
|
|
__FUNCTION__,
|
|
$mediaType,
|
|
$e->getMessage()
|
|
));
|
|
}
|
|
}
|
|
}
|
|
if (!count($this->_storages)) {
|
|
throw new \Exception("Download된 Content가 없습니다.");
|
|
}
|
|
}
|
|
//Yamap ViewPage의 이미지나영상데이터가 있으면 URL과MediaType을 가져온다
|
|
private function getUrlsByDetailPageMediaType(string $mediaType, Crawler $selector, array $options, array $urls = []): array
|
|
{
|
|
$urls[$mediaType] = [];
|
|
$selector->filter($options["tag"])->each(
|
|
function (Crawler $node) use (&$mediaType, &$options, &$urls): void {
|
|
$url = $node->attr($options["attr"]);
|
|
log_message("debug", "getUrlsByDetailPageMediaType-> {$mediaType}[{$options["attr"]}]:{$url}");
|
|
if (!is_null($url)) {
|
|
$urls[$mediaType][] = $url;
|
|
}
|
|
}
|
|
);
|
|
return $urls;
|
|
}
|
|
private function detailPage(array $listInfo): array
|
|
{
|
|
// <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> </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>
|
|
$response = $this->getMySocket()->getContent($listInfo['detail_url']);
|
|
$tag = getenv("yamap.view.content.tag");
|
|
$selector = $this->getSelector($response, $tag);
|
|
log_message("debug", "\n-----------detailPage Tag: {$tag}---------------\n{$selector->html()}\n---------------------------\n");
|
|
$urls = $this->getUrlsByDetailPageMediaType("image", $selector, ["tag" => "img", "attr" => "src"]);
|
|
$urls = $this->getUrlsByDetailPageMediaType("video", $selector, ["tag" => "video", "attr" => "src"], $urls);
|
|
log_message("debug", "\n-------------------------\n" . var_export($urls, true) . "\n-----------------------\n");
|
|
log_message("notice", "-----------" . __FUNCTION__ . " 작업완료--------");
|
|
return array($listInfo, $urls);
|
|
}
|
|
private function mainPage(string $url): array
|
|
{
|
|
$listInfos = [];
|
|
$response = $this->getMySocket()->getContent($url);
|
|
$selector = $this->getSelector($response, getenv("yamap.list.tag"));
|
|
//div.bbs_item를 가진 객체를 찾아서 같은 형식의 객체(sibling)를 배열로 넘김
|
|
// log_message("debug", sprintf("\n-------------MainPage------------\n%s\n--------------------------\n", $selector->html()));
|
|
$selector->filter(getenv("yamap.list.item.tag"))->each(
|
|
function (Crawler $node) use (&$listInfos): void {
|
|
//bbs_item에서 span.g_nickname 객체를 찾아서 작성자가 "관리자" 아닌지 확인 후 Return Bool
|
|
$nickname = $node->filter(getenv("yamap.list.item.nickname.tag"))->text();
|
|
$hit = $node->filter(getenv("yamap.list.item.hit.tag"))->text();
|
|
$date = $node->filter(getenv("yamap.list.item.date.tag"))->text();
|
|
if ($nickname != getenv("yamap.list.item.nickname.except")) {
|
|
//작성자가 "관리자"가 아니 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->children()->last()->text();
|
|
$listInfos[] = ['title' => $title, 'nickname' => $nickname, 'detail_url' => $detail_url, 'date' => $date, 'hit' => $hit];
|
|
}
|
|
}
|
|
);
|
|
if (!count($listInfos)) {
|
|
throw new \Exception("Target URL이 없습니다.");
|
|
}
|
|
log_message("notice", __FUNCTION__ . " 작업 완료");
|
|
return $listInfos;
|
|
}
|
|
public function execute(): void
|
|
{
|
|
if ($this->getDebug()) {
|
|
$listInfos = [
|
|
'title' => getenv("yamap.view.test.title"),
|
|
'nickname' => getenv("yamap.view.test.nickname"),
|
|
'detail_url' => getenv("yamap.view.test.url"),
|
|
'time' => date("Y-m-d H:i:s"),
|
|
'hit' => 1,
|
|
];
|
|
} else {
|
|
$listInfos = $this->mainPage(getenv("yamap.list.url." . $this->getCategory()));
|
|
}
|
|
//Limit가 0이면 $listInfos 갯수만큼 다하고, LIMIT 갯수 혹은 item의 갯수중 작은수만큼 한다.
|
|
$max_limit = intval(getenv("yamap.list.max_limit"));
|
|
if ($max_limit) {
|
|
$max_limit = count($listInfos) <= $max_limit ? count($listInfos) : $max_limit;
|
|
} else {
|
|
$max_limit = count($listInfos);
|
|
}
|
|
$total = count($listInfos);
|
|
$i = 1;
|
|
foreach ($listInfos as $listInfo) {
|
|
if ($i <= $max_limit) {
|
|
try {
|
|
log_message("notice", "게시물 {$i}번째/{$total}개중 {$listInfo["nickname"]} 작업시작");
|
|
list($listInfo, $urls) = $this->detailPage($listInfo);
|
|
$this->mediaContent($urls);
|
|
//File DB 및 Board DB 등록작업
|
|
$board_entity = $this->getBoardLibrary()->createByCrawler($i, $listInfo, $this->_storages);
|
|
$this->getFileLibrary()->createByCrawler($board_entity, $this->_storages);
|
|
$this->getImageLibrary()->createByCrawler($board_entity, $this->_storages);
|
|
log_message("notice", "게시물 {$i}번째/{$total}개중 {$listInfo["nickname"]} 작업완료.");
|
|
$i++;
|
|
} catch (\Exception $e) {
|
|
log_message("debug", $e->getMessage());
|
|
}
|
|
}
|
|
}
|
|
log_message("notice", "Crawler->" . __FUNCTION__ . " 작업이 완료되었습니다.");
|
|
}
|
|
}
|