景区采集

app/Console/Commands/Collector.php

@@ -53,8 +53,9 @@ class Collector extends Command
 			$this->line("供应商 $supplier_id 导入完毕");
 		} else {
 //    		$this->exists_update = false;
-			$this->tourist_line();
-			$this->hotel();
+//			$this->tourist_line();
+//			$this->hotel();
+			$this->scenic();
 			$this->line('全部采集完毕');
 		}
         return Command::SUCCESS;
@@ -100,6 +101,84 @@ class Collector extends Command
 		}
 	}
 
+	//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];
+
+					$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] ?? '';
+					}
+
+					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' => $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()
     {
@@ -107,7 +186,7 @@ class Collector extends Command
 
         $mddid = 10030; //10030==三亚
         for ($i=0; $i<10; $i++) {
-        	$this->line('开始采集：第 ' . ($i + 1) . ' 页');
+        	$this->line('开始采集酒店：第 ' . ($i + 1) . ' 页');
 			$data = $http->get('https://m.mafengwo.cn/rest/hotel/hotels/', [
 				'filter' => [
 					'mddid' => $mddid
@@ -201,6 +280,7 @@ class Collector extends Command
         $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