PHP抓取国家统计局行政区划数据-2023最新行政区划编码

因项目需要省市区三级联动行政区划选择，现成的部分存在缺失的情况，最好的办法就是直接到国家统计局网站抓取最新的数据。

<?php


//获取全国最新行政区域编码
function getCityCode(){

	ini_set('memory_limit', '4096M');    // 临时设置最大内存占用
	set_time_limit(0);   // 设置脚本最大执行时间 为0 永不过期
    
    //抓取2023年的全国最新行政区域编码
	$data = file_get_contents('http://www.stats.gov.cn/sj/tjbz/tjyqhdmhcxhfdm/2023/index.html');
	//echo $data;
	
	
	$html = getSubstr($data, '请务必结合有关实际情况', '</td></tr></table></TD></TR></TBODY>');
	//echo $html;
	//<a href="13.html">河北省<br /></a>
	$pat = '/<a(.*?)href="(.*?)"(.*?)>(.*?)<\/a>/i';

	preg_match_all($pat, $html, $province);

	//echo json_encode($province,JSON_UNESCAPED_UNICODE);

	$province_code = [];
	$city_code = [];
	$area_code = [];


	foreach ($province[4] as $key => $value) {

		//补全省的行政编码
		$code = explode('.', $province[2][$key])[0] . '0000000000';
		$province_code[$code] = [
			'id' => (int)$code,
			'name' => trim($value, '<br/>'),
			'parent_id' => 0,       //省级
			'level' => 1,
		];

	}
	//echo json_encode($province_code,JSON_UNESCAPED_UNICODE);

	//市/区/街道
	foreach ($province[2] as $key => $value) {
		$province_code_num = explode('.', $value)[0] . '0000000000';
		$city_data = getMore($value, $province_code_num);

		//组城市
		for ($i = 0; $i < count($city_data[4]); $i += 2) {

			$city_code_num = $city_data[4][$i];
			$city_code[$city_code_num] = [
				'id' => (int)($city_code_num),
				'name' => $city_data[4][$i + 1],
				'parent_id' => (int)$province_code_num,
				'level' => 2,
			];


			//组区数据
			$area_data = getMore($city_data[2][$i]);

			for ($j = 0; $j < count($area_data[4]); $j += 2) {

				$area_code_num = $area_data[4][$j];
				$area_code[$area_code_num] = [
					'id' => (int)($area_code_num),
					'name' => $area_data[4][$j + 1],
					'parent_id' => (int)$city_code_num,
					'level' => 3,
				];




			}

		}
	}

	$all_code = array_merge($province_code, $city_code, $area_code);
	
	//写入文件
	file_put_contents('./city.php', "<?php \n return " . var_export($all_code, true) . ";");

	exit('完毕');

}

function getMore($url){
	
	$data = file_get_contents('http://www.stats.gov.cn/sj/tjbz/tjyqhdmhcxhfdm/2023/' . $url);

	$html = getSubstr($data, '<td>名称</td></tr>', '</td></tr></table></TD></TR></TBODY></TABLE>');


	$pat = '/<a(.*?)href="(.*?)"(.*?)>(.*?)<\/a>/i';

	preg_match_all($pat, $html, $html_data);

	if (count($html_data[2]) > 0) {
		$html_data['is_end'] = 'false';
	} else {
		$html_data['is_end'] = 'true';
	}

	return $html_data;

}


function getSubstr($str, $leftStr, $rightStr){
	$llen = strlen($leftStr);
	$left = strpos($str, $leftStr);
	$right = strpos($str, $rightStr, $left + $llen);

	if ($left < 0 || $right < $left)
		return "";

	return substr($str, $left + $llen, $right - $left - $llen);
}

getCityCode();


?>