117 lines
4.6 KiB
PHP
117 lines
4.6 KiB
PHP
<?php
|
||
// +----------------------------------------------------------------------
|
||
// | YwxApp [ WE CAN DO IT JUST THINK ]
|
||
// +----------------------------------------------------------------------
|
||
// | Copyright (c) 2026-2036 http://ywxapp.cn All rights reserved.
|
||
// +----------------------------------------------------------------------
|
||
// | Author: ywxapp<admin@ywxapp.cn>
|
||
// +----------------------------------------------------------------------
|
||
declare(strict_types=1);
|
||
|
||
namespace ywxapp\middleware;
|
||
|
||
use think\facade\Db;
|
||
use ywxapp\model\BaseModel;
|
||
use ywxapp\library\SpiderDetect;
|
||
|
||
/**
|
||
* 全站搜索蜘蛛统计中间件（全局注册于 app/middleware.php）
|
||
*
|
||
* 设计要点：
|
||
* - 仅命中蜘蛛 UA 才落库，普通用户请求零额外查询；
|
||
* - 落库在响应生成之后（$next 之后），不阻塞蜘蛛抓取响应；
|
||
* - 明细写 spider_log，按日聚合 upsert 到 spider_stat（报表免全表扫描）；
|
||
* - 任何数据库异常静默吞掉（统计绝不能影响业务），缺表时经 BaseModel 自愈引擎 自愈一次；
|
||
* - install.sql 为唯一事实源，此处 DDL 仅为老库运行时兜底。
|
||
*
|
||
* @author ywxapp <admin@ywxapp.cn>
|
||
*/
|
||
class SpiderStat
|
||
{
|
||
public function handle($request, \Closure $next)
|
||
{
|
||
$response = $next($request);
|
||
|
||
try {
|
||
$ua = (string) $request->header('user-agent', '');
|
||
$spider = SpiderDetect::detect($ua);
|
||
if ($spider !== null) {
|
||
$this->record($request, $response, $spider, $ua);
|
||
}
|
||
} catch (\Throwable $e) {
|
||
// 统计失败绝不影响正常响应
|
||
}
|
||
|
||
return $response;
|
||
}
|
||
|
||
/**
|
||
* 落库：明细 + 按日聚合
|
||
*/
|
||
protected function record($request, $response, string $spider, string $ua): void
|
||
{
|
||
$prefix = config('database.connections.mysql.prefix', 'wxapp_');
|
||
$logTable = $prefix . 'spider_log';
|
||
$statTable = $prefix . 'spider_stat';
|
||
|
||
$data = [
|
||
'spider' => $spider,
|
||
'url' => mb_substr((string) $request->url(), 0, 500),
|
||
'ip' => mb_substr((string) $request->ip(), 0, 45),
|
||
'app' => mb_substr((string) (app('http')->getName() ?: ''), 0, 20),
|
||
'user_agent' => mb_substr($ua, 0, 500),
|
||
'http_code' => method_exists($response, 'getCode') ? (int) $response->getCode() : 200,
|
||
'create_at' => time(),
|
||
];
|
||
|
||
try {
|
||
$this->insert($logTable, $statTable, $data, $spider);
|
||
} catch (\Throwable $e) {
|
||
// 表可能不存在（老库未升级）：自愈一次后重试
|
||
$this->ensureTables($logTable, $statTable);
|
||
$this->insert($logTable, $statTable, $data, $spider);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 写明细 + 聚合 upsert
|
||
*/
|
||
protected function insert(string $logTable, string $statTable, array $data, string $spider): void
|
||
{
|
||
Db::table($logTable)->insert($data);
|
||
// 按日聚合：主键(stat_date, spider)，存在则计数+1
|
||
Db::execute(
|
||
"INSERT INTO `{$statTable}` (`stat_date`, `spider`, `count`) VALUES (?, ?, 1) "
|
||
. "ON DUPLICATE KEY UPDATE `count` = `count` + 1",
|
||
[date('Y-m-d'), $spider]
|
||
);
|
||
}
|
||
|
||
/**
|
||
* 缺表自愈（DDL 与 public/install/install.sql 保持一致）
|
||
*/
|
||
protected function ensureTables(string $logTable, string $statTable): void
|
||
{
|
||
BaseModel::ensureTable($logTable, "CREATE TABLE IF NOT EXISTS `{$logTable}` (
|
||
`id` bigint unsigned NOT NULL AUTO_INCREMENT COMMENT '主键ID',
|
||
`spider` varchar(20) NOT NULL DEFAULT '' COMMENT '蜘蛛标识',
|
||
`url` varchar(500) NOT NULL DEFAULT '' COMMENT '抓取URL',
|
||
`ip` varchar(45) NOT NULL DEFAULT '' COMMENT '来源IP',
|
||
`app` varchar(20) NOT NULL DEFAULT '' COMMENT '应用名',
|
||
`user_agent` varchar(500) NOT NULL DEFAULT '' COMMENT 'User-Agent',
|
||
`http_code` smallint unsigned NOT NULL DEFAULT '200' COMMENT '响应状态码',
|
||
`create_at` int NOT NULL DEFAULT '0' COMMENT '抓取时间',
|
||
PRIMARY KEY (`id`),
|
||
KEY `idx_spider_time` (`spider`,`create_at`),
|
||
KEY `idx_create_at` (`create_at`)
|
||
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci COMMENT='搜索蜘蛛抓取日志'");
|
||
|
||
BaseModel::ensureTable($statTable, "CREATE TABLE IF NOT EXISTS `{$statTable}` (
|
||
`stat_date` date NOT NULL COMMENT '统计日期',
|
||
`spider` varchar(20) NOT NULL DEFAULT '' COMMENT '蜘蛛标识',
|
||
`count` int unsigned NOT NULL DEFAULT '0' COMMENT '抓取次数',
|
||
PRIMARY KEY (`stat_date`,`spider`)
|
||
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci COMMENT='搜索蜘蛛按日统计'");
|
||
}
|
||
}
|