Automation/app/Libraries/MyCrawler/MyCrawler.php

<?php

namespace App\Libraries\MyCrawler;

use App\Libraries\CommonLibrary;
use Symfony\Component\DomCrawler\Crawler;
use App\Traits\FileTrait;

abstract class MyCrawler extends CommonLibrary
{
    use FileTrait;
    protected function __construct()
    {
        parent::__construct();
    }
    abstract protected function getMySocket();
    abstract protected function createMyStorage();
    final protected function getSelector(string $content, string $tag): Crawler
    {
        $crawler = new Crawler($content);
        if ($this->isDebug) {
            log_message("debug", __FUNCTION__ . "=> " . $tag);
        }
        $crawler->filter($tag);
        if ($this->isDebug) {
            log_message("debug", sprintf(
                "\n------------%s HTML-------------\n%s\n-----------------------------------------------------\n",
                __FUNCTION__,
                $crawler->filter($tag)->html()
            ));
        }
        return $crawler->filter($tag);
    }

    protected function changeURLByCrawler(string $url): string
    {
        return preg_match('/^[^?]+/', $url, $matches) ? $matches[0] : null;
    }
    protected function getUrlByMediaType(Crawler $node, string $media_type,  string $attr): null|string
    {
        switch ($media_type) {
            case 'video':
                try {
                    $url = $node->attr($attr); //<video src="test.mp4"></video> 또는 <video data-src="test.mp4"></video>
                } catch (\Exception) {
                    $url = $node->children()->attr("src"); //<video><source src="test.mp4"></source</video>
                }
                break;
            case 'img':
            default:
                $url = $node->attr($attr);
                break;
        }
        return $url;
    }
    protected function getUrlsByMediaType(Crawler $selector, string $media_type, string $attr, array $urls = []): array
    {
        log_message("notice", "-----------" . __FUNCTION__ . "=> {$media_type} 작업시작--------");
        $urls[$media_type] = [];
        $selector->filter($media_type)->each(
            function (Crawler $node) use (&$media_type, &$attr, &$urls): void {
                $url = $this->getUrlByMediaType($node, $media_type, $attr);
                if ($url !== null && preg_match('/^[^?]+/', $url, $matches)) {
                    $urls[$media_type][] = $this->changeURLByCrawler($matches[0]);
                } else {
                    log_message("debug", __FUNCTION__ . "-> {$media_type}:{$attr}\n");
                    //Node 모든 속성은 DOMElement 변환 후 반환가능
                    $domNode = $node->getNode(0);
                    if ($domNode->hasAttributes()) {
                        foreach ($domNode->attributes as $attr) {
                            log_message("debug", "{$attr->nodeName} = {$attr->nodeValue}");
                        }
                    }
                }
            }
        );
        log_message("notice", "-----------" . __FUNCTION__ . "=> {$media_type} 작업완료--------");
        return $urls;
    }
    private function media_save(int $file_sequence, string $media_type, string $file_name, string $content): mixed
    {
        log_message("debug", __FUNCTION__ . " 원본파일 {$file_name} 작업 시작");
        $storage = $this->createMyStorage();
        $storage->setOriginName($file_name);
        $storage->setOriginContent($content);
        $storage->setOriginType($media_type);
        $storage->setOriginSequence($file_sequence);
        return $storage->save();
    }
    //Yamap ViewPage의 이미지나영상데이터가 있으면 Dodownload 한다.
    private function media_download(string $media_type, 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, $media_type)) {
            throw new \Exception("파일명 형식이 {$media_type}가 아닙니다");
        }
        $content = $this->getMySocket()->getContent($url);
        log_message("notice", "{$file_name} 파일이 다운로드되었습니다!");
        return array($file_name, $content);
    }
    final protected function media_process(array $media_urls): array
    {
        $file_sequence = 1;
        $storages = []; //CreateBoard에서 사용을 위해 DetailPage마다 초기화
        foreach ($media_urls as $media_type => $urls) {
            $total = count($urls);
            foreach ($urls as $url) {
                log_message("notice", __FUNCTION__ . " {$file_sequence}번째/총:{$total} MediaType->{$media_type} 작업 시작");
                try {
                    list($file_name, $content) = $this->media_download($media_type, $url);
                    $storage = $this->media_save($file_sequence, $media_type, $file_name, $content);
                    log_message("debug", __FUNCTION__ . " {$file_sequence}번째/총:{$total} 결과=>" . $storage->getOriginName());
                    $storages[] = $storage;
                } catch (\Exception $e) {
                    log_message("warning", sprintf(
                        "\n---%s MediaType->%s {$file_sequence}번째/총:{$total} 오류---\n%s\n-----------------------------------------\n",
                        __FUNCTION__,
                        $media_type,
                        $e->getMessage()
                    ));
                }
                log_message("notice", __FUNCTION__ . " {$file_sequence}번째/총:{$total} MediaType->{$media_type} 작업 완료");
                $file_sequence++;
            }
        }
        log_message("notice", __FUNCTION__ . "=> 게시물 {$url} 작업종료");
        return $storages;
    }
}