hainan/app/Console/Commands/Collector.php

<?php

namespace App\Console\Commands;

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

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 = '驴妈妈产品采集';

    /**
	 * 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)) {
			if (ctype_digit($supplier_id)) {
				$this->import($supplier_id);
			}
			$this->line("供应商 $supplier_id 导入完毕");
		} else {
//    		$this->exists_update = false;
			$this->scenic();
			$this->hotel();
			$this->line('全部采集完毕');
		}
        return Command::SUCCESS;
    }

    //导入
    private function import($supplier_id)
	{
		$limit = $this->option('limit');
		if (!empty($limit)) {
			$arr = explode(',', $limit);
			$count = count($arr);

			$cpModel = CollectProduct::query()->orderBy('id', 'desc');
			if ($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();
			}

			array_walk($import_data, function ($v) use ($supplier_id) {
				$v['supplier_id'] = $supplier_id;
				unset($v['unique_id'], $v['site']);

				Product::query()->updateOrCreate(['supplier_id' => $supplier_id, 'title' => $v['title']], $v);
			});
		}
	}

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

        $mddid = 10030; //10030==三亚
        for ($i=0; $i<10; $i++) {
        	$this->line('开始采集：第 ' . ($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'];

				CollectProduct::updateOrCreate(['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,
					'logitude' => $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);
	}

    //景区采集
    private function scenic()
    {
        $http = Http::withOptions(['verify' => false]);

        for($i=0; $i<10; $i++) {
            $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_2_project'][] = ['name' => $tag, 'num' => '', 'price' => ''];
					}
				}

				CollectProduct::updateOrCreate(['unique_id' => $id, 'site' => 1], [
					'unique_id' => $id,
					'site' => 1,
					'type' => 2, //0:旅游线路、1:酒店、2:景区、3:餐厅、4:车队、5:单项
					'title' => mb_substr($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,
					'logitude' => 0,
					'latitude' => 0,
					'address' => '',
				]);

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