hainan/app/Console/Commands/Collector.php

<?php

namespace App\Console\Commands;

use App\Models\CollectProduct;
use App\Models\Product;
use App\Models\Supplier;
use Illuminate\Console\Command;
use Illuminate\Support\Facades\Http;
use Illuminate\Support\Facades\Storage;

class Collector extends Command
{
    /**
     * The name and signature of the console command.
     *
     * @var string
     */
    protected $signature = 'collector {--import=} {--limit=}'; //php artisan collector --import=$supplier_id

    /**
     * The console command description.
     *
     * @var string
     */
    protected $description = '马蜂窝产品采集';

    /**
	 * 如果已经存在是否要更新，true更新，false不更新
	 * @var bool
	 */
    private bool $exists_update = true;

    /**
     * Create a new command instance.
     *
     * @return void
     */
    public function __construct()
    {
        parent::__construct();
    }

    /**
     * Execute the console command.
     *
     * @return int
     */
    public function handle()
    {
		$supplier_id = $this->option('import');
		if (!empty($supplier_id)) {
			$this->import($supplier_id);
			$this->line("供应商 $supplier_id 导入完毕");
		} else {
//    		$this->exists_update = false;
			$this->tourist_line();
			$this->hotel();
			$this->scenic();
//			$this->db_img_replace();
			$this->line('全部采集完毕');
		}
        return Command::SUCCESS;
    }

    //导入
    private function import($supplier_id)
	{
		$limit = $this->option('limit');
		$cpModel = CollectProduct::query()->orderBy('id', 'desc');
		if (!empty($limit)) {
			$arr = explode(',', $limit);
			$count = count($arr);

			if ($count == 1) {
				$import_data = $cpModel->limit($arr[0])->get()->toArray();
			} else if ($count == 2) {
				$import_data = $cpModel->offset($arr[0])->limit($arr[1])->get()->toArray();
			} else {
				$import_data = $cpModel->get()->toArray();
			}
		} else {
			$import_data = $cpModel->get()->toArray();
		}

		if (ctype_digit($supplier_id)) {
			$ids = [$supplier_id];
		} else if ($supplier_id == 'all') {
			$ids = Supplier::query()->where('id', '>', 1)->pluck('id');
		} else {
			return;
		}

		foreach ($ids as $supplier_id) {
			$this->line('正在导入 ' . $supplier_id);
			array_walk($import_data, function ($v) use ($supplier_id) {
				$v['supplier_id'] = $supplier_id;
				unset($v['unique_id'], $v['site']);

				Product::updateOrCreate(['supplier_id' => $supplier_id, 'title' => $v['title']], $v);
			});
			$this->line("导入 $supplier_id 结束");
		}
	}

	//景区采集
	private function scenic()
	{
		$http = Http::withOptions(['verify' => false])->withHeaders(['User-Agent' => 'Mozilla/5.0 (iPhone; CPU iPhone OS 13_2_3 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/13.0.3 Mobile/15E148 Safari/604.1']);
		for ($i=15; $i<20; $i++) {
			$this->line('开始采集景区：第 ' . ($i + 1) . ' 页');
			$html = $http->get('https://m.mafengwo.cn/jd/10030/gonglve.html?page=' . ($i + 1));
			if (preg_match_all('/data-url="(.*?)"/', $html, $matches)) {
				if(empty($matches[1])) continue;

				foreach ($matches[1] as $url) {
					$this->line('开始采集 ' . $url);
					$res = $http->get('https://m.mafengwo.cn' . $url);

					if (!preg_match('/\/(\d+)\.html/', $url, $match)) continue;
					$id = $match[1];

					if ($this->exists_update == false && CollectProduct::where(['unique_id' => $id, 'site' => 1])->exists()) {
						continue;
					}

					$inc_sales = $http->post('https://m.mafengwo.cn/poi/poi/inc_sales', ['poiid' => $id]);
					preg_match('#<strong><em>￥</em>(\d+)</strong>#', $inc_sales['html'] ?? '', $match_price);

					preg_match('#<h1>(.*?)</h1>#', $res, $match_title);
					preg_match_all('#<div class="swiper-slide">[\s\S]*?<img src="(.*?)"#', $res, $match_pictures);
					preg_match('#蜂蜂点评<br/><strong>(\d+)条#', $res, $match_sale);
					preg_match('#"row h1" data-jump="ticket">\s*([\w\W]*?)\s*</div>\s*<div class="row h1"#s', $res, $match_know);
					preg_match('#<div class="desc" style="display: none;">\s*([\w\W]*?)\s*</div>#s', $res, $match_content);
					preg_match_all('#时间：</dt>\s*<dd>\s*(.*?)\s*</dd>#', $res, $match_open_time);

					//扩展字段
					preg_match('#<p class="t2">地址:(.*?)<i></i></p>#', $res, $match_address);

					$pos_start = strpos($res, '<div class="keywords">');
					$pos_end = strpos($res, '</div>', $pos_start);
					preg_match_all('#<span>([^<>]*?)</span>#s', substr($res, $pos_start, $pos_end - $pos_start), $match_project);
					if (isset($match_project[1]) && is_array($match_project[1])) {
						$extends['field_2_project'] = array_map(function($v) {
							return [
								'name' => $v,
								'num' => '',
								'price' => '',
							];
						}, $match_project[1]);
					}

					$extends['field_2_address'] = $match_address[1] ?? '';
					$extends['field_2_open_time'][0]['node'] = '营业时间';
					$extends['field_2_open_time'][0]['summer'] = '上岛时间';
					$extends['field_2_open_time'][0]['winter'] = '下岛时间';
					if (isset($match_open_time[1]) && is_array($match_open_time[1])) {
						$extends['field_2_open_time'][1]['node'] = $match_open_time[1][0] ?? '';
						$extends['field_2_open_time'][1]['summer'] = $match_open_time[1][1] ?? '';
						$extends['field_2_open_time'][1]['winter'] = $match_open_time[1][2] ?? '';
					}

					$this->save_to_db(['unique_id' => $id, 'site' => 1], [
						'unique_id' => $id,
						'site' => 1,
						'type' => 2, //0:旅游线路、1:酒店、2:景区、3:餐厅、4:车队、5:单项
						'title' => $match_title[1] ?? '',
						'price' => $match_price[1] ?? 0,
						'original_price' => ($match_price[1] ?? 0) * 1.58,
						'pictures' => $match_pictures[1] ?? [],
						'stock' => mt_rand(1000, 9999),
						'sale' => $match_sale[1] ?? 0,
						'status' => -2, //-2下架
						'know' => $match_know[1] ?? '',
						'content' => $match_content[1] ?? '',
						'extends' => $extends,
						'longitude' => 0,
						'latitude' => 0,
						'address' => $match_address[1] ?? '',
					]);
					$this->line("采集 $url 完毕" . PHP_EOL);
				}
			}
			$this->line('第 ' . ($i + 1) . ' 页采集结束' . PHP_EOL);
		}
	}

    //酒店采集
    private function hotel()
    {
        $http = Http::withOptions(['verify' => false]);

        $mddid = 10030; //10030==三亚
        for ($i=0; $i<10; $i++) {
        	$this->line('开始采集酒店：第 ' . ($i + 1) . ' 页');
			$data = $http->get('https://m.mafengwo.cn/rest/hotel/hotels/', [
				'filter' => [
					'mddid' => $mddid
				],
				'page' => [
					'mode' => 'sequential',
					'boundary' => $i * 20, //分页参数
					'num' => 20
				],
			]);

			if (empty($data['data']['list'])) {
				continue;
			}

			foreach ($data['data']['list'] as $v) {
				if (empty($v['id'])) continue;

				if ($this->exists_update == false && CollectProduct::where(['unique_id' => $v['id'], 'site' => 1])->exists()) {
					continue;
				}

				$this->line('采集详情：' . $v['id']);

				//基本信息
				$params = [
					'_ts' => time() . '123',
					'hotel_id' => (string)$v['id'],
					'lat' => '',
					'lng' => '',
					'rmdd_id' => (string)$mddid,
				];
				$params['_sn'] = $this->_sn($params);
				$data = $http->get('https://m.mafengwo.cn/hservice/detail/info/base_info', $params);

				if (empty($data['data']['info'])) {
					continue;
				}
				$base_info = $data['data']['info'];

				//酒店详情
				$data = $http->get('https://m.mafengwo.cn/hservice/detail/info/guide_info', ['hotel_id' => $v['id']]);
				$guide_info = $data['data']['info'] ?? [];

				//旅游须知
				$know = "<p>入住时间：" . (!empty($guide_info['check_in']['title']) ? $guide_info['check_in']['title'] : '') . "</p>";
				$know .= "<p>离店时间：" . (!empty($guide_info['check_out']['title']) ? $guide_info['check_out']['title'] : '') . "</p>";
				$know .= '<p>' . array_reduce($base_info['facility_sort'] ?? [], fn($v1, $v2) => $v1 . $v2['title'] ?? '') . '</p>';

				//扩展字段
				$extends['field_1_tags'] = array_map(fn($v) => $v['title'] ?? '', $guide_info['facility'] ?? []);
				$extends['field_1_name'] = $base_info['name'];
				$extends['field_1_address'] = $base_info['address'];
				$extends['field_1_latitude'] = $base_info['lat'];
				$extends['field_1_longitude'] = $base_info['lng'];

				$this->save_to_db(['unique_id' => $v['id'], 'site' => 1], [
					'unique_id' => $v['id'],
					'site' => 1,
					'type' => 1, //0:旅游线路、1:酒店、2:景区、3:餐厅、4:车队、5:单项
					'title' => mb_substr($base_info['name'] ?? '' . $base_info['level'] ?? '', 0, 255),
					'price' => mt_rand(150, 350),
					'original_price' => mt_rand(350, 550),
					'pictures' => array_map(fn($v) => $v['url'] ?? '', $base_info['album'] ?? []) ?? [],
					'stock' => mt_rand(1000, 9999),
					'sale' => $base_info['num_collect'] ?? 0,
					'status' => -2, //-2下架
					'know' => $know,
					'content' => $guide_info['intro'] ?? '',
					'extends' => $extends,
					'longitude' => $base_info['lng'] ?? 0,
					'latitude' => $base_info['lat'] ?? 0,
					'address' => $base_info['address'] ?? '',
				]);
				$this->line("{$v['id']} 采集结束" . PHP_EOL);
			}
			$this->line('第 ' . ($i + 1) . ' 页采集结束' . PHP_EOL);
        }
    }

    //酒店详情计算_sn
    private function _sn($params): string
	{
		ksort($params);
		return substr(md5(json_encode($params) . 'c9d6618dbc657b41a66eb0af952906f1'), 2, 10);
	}

    //旅游线路采集
    private function tourist_line()
    {
        $http = Http::withOptions(['verify' => false]);

        for($i=0; $i<10; $i++) {
			$this->line('开始采集旅游线路：第 ' . ($i + 1) . ' 页');
            $data = $http->get('https://m.mafengwo.cn/sales/ajax.php', [
                'sF' => 'search_new_list',
                'offset' => $i * 10, //分页参数
            ]);

            if (empty($data['data'])) {
				continue;
			}
			$data = $data['data'];

            preg_match_all('/<a href="\/sales\/(\d+)\.html"/', $data, $matches);
            if (empty($matches[1])) continue;

			foreach ($matches[1] as $id) {
				if ($this->exists_update == false && CollectProduct::where(['unique_id' => $id, 'site' => 1])->exists()) {
					continue;
				}

				$this->line('开始采集：' . $id);
				$info = $http->get('https://m.mafengwo.cn/sales/detail/index/info?id=' . $id);

				//旅游须知
				$know = $info['data']['list']['content'][0]['content'] ?? [];
				if (isset($info['data']['list']['content'][0]['content']) && is_array($info['data']['list']['content'][0]['content'])) {
					$know = current(array_filter($info['data']['list']['content'][0]['content'], fn($v) => isset($v['name']) && $v['name'] == '购买须知'));
					if (isset($know['content']) && is_array($know['content'])) {
						$know = array_reduce(
							$know['content'], fn($v1, $v2) => $v1 .
							(isset($v2['name']) && is_string($v2['name']) ? "<h3>{$v2['name']}</h3>" : '') .
							(isset($v2['content']) && is_string($v2['content']) ? $v2['content'] : '')
						);
					}
				}

				//产品详情
				$content = '';
				if (isset($info['data']['list']['content'][0]['content']) && is_array($info['data']['list']['content'][0]['content'])) {
					$content = current(array_filter($info['data']['list']['content'][0]['content'], fn($v) => isset($v['key']) && $v['key'] == 'introduce'));
					if (isset($content['content']) && is_array($content['content'])) {
						$content = current(array_filter($content['content'], fn($v) => isset($v['key']) && $v['key'] == 'introduction'));
						$content = is_string($content['content']) ? $content['content'] : '';
					} else {
						$content = '';
					}
				}

				//扩展字段
				$extends = [];
				if (isset($info['data']['list']['base']['tags']) && is_array($info['data']['list']['base']['tags'])) {
					foreach ($info['data']['list']['base']['tags'] as $tag) {
						$extends['field_0_project'][] = ['name' => $tag, 'num' => '', 'price' => ''];
					}
				}

				$this->save_to_db(['unique_id' => $id, 'site' => 1], [
					'unique_id' => $id,
					'site' => 1,
					'type' => 0, //0:旅游线路、1:酒店、2:景区、3:餐厅、4:车队、5:单项
					'title' => mb_substr($info['data']['list']['base']['title'] ?? '', 0, 255),
					'price' => $info['data']['list']['base']['price_zhanshi'] ?? 0,
					'original_price' => ($info['data']['list']['base']['price_zhanshi'] ?? 0) * 1.58,
					'pictures' => $info['data']['list']['base']['imgList'] ?? [],
					'stock' => mt_rand(1000, 9999),
					'sale' => $info['data']['list']['base']['sold']['num'] ?? 0,
					'status' => -2, //-2下架
					'know' => $know,
					'content' => $content,
					'extends' => $extends,
					'longitude' => 0,
					'latitude' => 0,
					'address' => '',
				]);

				$this->line($id . ' 采集完毕！' . PHP_EOL);
			}
        }
    }

    //保存到数据库
	private function save_to_db($unique_flag, $data)
	{
		$this->search_img($data);
		CollectProduct::updateOrCreate($unique_flag, $data);
	}

    //搜索数据中的数据，有图片则下载并规格数据
    private function db_img_replace()
	{
		$all = CollectProduct::all(['id', 'pictures', 'know', 'content']);
		foreach ($all as $model) {
			$this->line('当前处理数据：' . $model->id);
			$data = $model->toArray();
			$this->search_img($data);

			$model->pictures = $data['pictures'];
			$model->know = $data['know'];
			$model->content = $data['content'];
			$model->save();
		}
	}

	//从数据中搜索图片并下载
	private function search_img(&$data)
	{
		$img_host = PHP_OS == 'WINNT' ? 'https://yytx.eugyl.com/storage/' : env('APP_URL') . '/storage/';
		//保存产品图片集
		if (!empty($data['pictures']) && is_array($data['pictures'])) {
			foreach ($data['pictures'] as &$url) {
				if (substr($url, 0, 4) != 'http' || str_contains($url, $img_host)) continue;

				$this->line('正在下载图片：' . $url);
				$src = $this->download_img($url);
				if (!$src) continue;
				$url = $src;
			}
		}

		//保存富文本内的图片
		foreach ($data as $key => &$rich_text) {
			//只替换know和content字段
			if ($key == 'know' || $key == 'content') {
				//删除href链接
				$rich_text = preg_replace('#<a.*?>|</a>#', '', $rich_text);

				if (!empty($rich_text) && preg_match_all('/<img\s+.*?src=([\'"])(.*?)\1.*?>/i', $rich_text, $image_url) && !empty($image_url[2]) && is_array($image_url[2])) {
					foreach (array_unique($image_url[2]) as $url) {
						if (substr($url, 0, 4) != 'http' || str_contains($url, $img_host)) continue;

						$this->line('正在下载图片：' . $url);
						$src = $this->download_img($url);
						if (!$src) continue;
						$rich_text = str_replace([$url, ' data-src'], [$img_host . $src, ' src'], $rich_text);
					}
				}
			}
		}
	}

	//下载图片
	private function download_img($url): string
	{
		$file_info = pathinfo($url);
		$ext = 'jpg';
		if (!empty($file_info['extension'])) {
			$pos = strpos($file_info['extension'], '?');
			$ext = $pos ? substr($file_info['extension'], 0, $pos) : $file_info['extension'];
		}

		try {
			$raw = file_get_contents($url);
		} catch (\Exception $e) {
			return '';
		}
		if (empty($raw)) {
			return '';
		}

		$md5 = md5($raw);
		$filename = 'collect/images/' . date('Y-m-d') . '/' . $md5 . '.' . $ext;

		if (file_exists(storage_path($filename)) || Storage::put('public/' . $filename, $raw)) {
			return $filename;
		} else {
			return '';
		}
	}
}