正在显示
1 个修改的文件
包含
0 行增加
和
227 行删除
cmd/to_es.php
已删除
100755 → 0
| 1 | -<?php | ||
| 2 | - | ||
| 3 | -//error_reporting(); | ||
| 4 | - | ||
| 5 | -require_once "../vendor/autoload.php"; | ||
| 6 | - | ||
| 7 | - | ||
| 8 | -if(in_array('v2',$argv)){ | ||
| 9 | - define('S_V2','2'); | ||
| 10 | -}else{ | ||
| 11 | - define('S_V2',''); | ||
| 12 | -} | ||
| 13 | - | ||
| 14 | -/** | ||
| 15 | - * 把mysql的数据同步到es | ||
| 16 | - * @author:dc | ||
| 17 | - * @time 2025/3/4 10:19 | ||
| 18 | - * Class SyncToEsCmd | ||
| 19 | - */ | ||
| 20 | -class SyncToEsCmd { | ||
| 21 | - | ||
| 22 | - public $isStop = false; | ||
| 23 | - | ||
| 24 | - /** | ||
| 25 | - * 文件夹 | ||
| 26 | - * @var array | ||
| 27 | - */ | ||
| 28 | - public $folders = []; | ||
| 29 | - | ||
| 30 | - /** | ||
| 31 | - * @var \Lib\Db | ||
| 32 | - */ | ||
| 33 | - public $fob_db; | ||
| 34 | - | ||
| 35 | - public function handler(){ | ||
| 36 | - | ||
| 37 | - $handler = function ($signal){ | ||
| 38 | - _echo('收到进程信号 '. $signal); | ||
| 39 | - // 可以处理其他程序 | ||
| 40 | - $this->isStop = true; | ||
| 41 | - }; | ||
| 42 | - pcntl_signal(SIGTERM, $handler); // 这个是kill | ||
| 43 | - pcntl_signal(SIGINT, $handler); // 这个是 ctrl+c | ||
| 44 | -// pcntl_signal(SIGHUP, $handler); | ||
| 45 | - | ||
| 46 | - | ||
| 47 | - if(S_V2){ | ||
| 48 | - $es = es('email_lists_copy'); | ||
| 49 | - }else{ | ||
| 50 | - $es = es(); | ||
| 51 | - } | ||
| 52 | - | ||
| 53 | - $db = db(); | ||
| 54 | - $this->fob_db = fob_mysql(); | ||
| 55 | - $startTime = time(); | ||
| 56 | - // 查询gmail的所有邮件文件夹 ai邮件不同步这个文件夹 | ||
| 57 | - $fids = $db->throw()->all("select `id` from `folders` where `origin_folder` in ('[Gmail]/&UWiQ6JD1TvY-','[Gmail]/All Mail','[Gmail]/&YkBnCZCuTvY-')"); | ||
| 58 | - $fids = array_column($fids,'id','id'); | ||
| 59 | - | ||
| 60 | - while (1){ | ||
| 61 | - | ||
| 62 | - if(time()-43200 > $startTime){ | ||
| 63 | - break; | ||
| 64 | - } | ||
| 65 | - | ||
| 66 | - // 检查是否接收到信号 | ||
| 67 | - pcntl_signal_dispatch(); | ||
| 68 | - | ||
| 69 | - if($this->isStop) { | ||
| 70 | - _echo('已退出进程'); | ||
| 71 | - break; | ||
| 72 | - } | ||
| 73 | - | ||
| 74 | - $id = redis()->rPop('sync_to_es'.S_V2); | ||
| 75 | - $code = 500; | ||
| 76 | - if($id){ | ||
| 77 | - $doc_id = ''; | ||
| 78 | - try { | ||
| 79 | - $data = $db->throw()->first('select * from `lists_hot` where `id` = '.$id); | ||
| 80 | - if(!$data){ | ||
| 81 | - $data = $db->throw()->first('select * from `lists` where `id` = '.$id); | ||
| 82 | - } | ||
| 83 | - }catch (Throwable $e){ | ||
| 84 | - redis()->lPush('sync_to_es'.S_V2,$id); | ||
| 85 | - _echo('sync to es '.$e->getMessage()); | ||
| 86 | - break; | ||
| 87 | - } | ||
| 88 | - | ||
| 89 | - if($data){ | ||
| 90 | - // 是否是所有邮件文件夹 是就跳过 | ||
| 91 | - if($fids[$data['folder_id']]??0){ | ||
| 92 | - continue; | ||
| 93 | - } | ||
| 94 | - // 设置 进程 是否在运行 | ||
| 95 | - $data['is_auto'] = $db->count('select count(*) from `lists_auto` where `list_id` = '.$data['id']) ? 1 : 0; | ||
| 96 | - try { | ||
| 97 | - // 文件夹 | ||
| 98 | - if(empty($this->folders[$data['folder_id']])){ | ||
| 99 | - $this->folders[$data['folder_id']] = $db->throw()->value(\Model\folderSql::first($data['folder_id'],'folder')); | ||
| 100 | - } | ||
| 101 | - // 为文件夹打标 方便查询 | ||
| 102 | - $data['folder_as_int'] = folder2int($this->folders[$data['folder_id']]); | ||
| 103 | - | ||
| 104 | - // 再次验证是否是自动回复 | ||
| 105 | - if($data['folder_as_int'] === 1 && !$data['is_auto']){ | ||
| 106 | - $filter = redis()->get('ai_email_filter_lists',[]); | ||
| 107 | - $filter = is_array($filter) ? $filter : []; | ||
| 108 | - foreach ($filter as $f){ | ||
| 109 | - list($t,$str) = $f; | ||
| 110 | - $haystack = ''; | ||
| 111 | - if($t==2){ | ||
| 112 | - $haystack = $data['subject']; | ||
| 113 | - }elseif ($t==1){ | ||
| 114 | - $haystack = $data['from']; | ||
| 115 | - } | ||
| 116 | - if($haystack && $str && stripos($haystack,$str)!==false){ | ||
| 117 | - $data['is_auto'] = 1; | ||
| 118 | - break; | ||
| 119 | - } | ||
| 120 | - } | ||
| 121 | - } | ||
| 122 | - // 再次验证是否是预热邮件 | ||
| 123 | -// if() | ||
| 124 | - | ||
| 125 | - // postid ai邮箱要用 这个是查询黑格 | ||
| 126 | - list($data['postid'],$data['source']) = $this->getPostid($data['email_id']); | ||
| 127 | - }catch (Throwable $e){ | ||
| 128 | - redis()->lPush('sync_to_es'.S_V2,$id); | ||
| 129 | - _echo('sync to es '.$e->getMessage()); | ||
| 130 | - break; | ||
| 131 | - } | ||
| 132 | - | ||
| 133 | - $data = $this->getEsData($data); | ||
| 134 | - $doc_id = $data['email_id'].'_'.$data['folder_id'].'_'.$data['uid']; | ||
| 135 | - | ||
| 136 | - $code = $es->save($doc_id,$data); | ||
| 137 | - } | ||
| 138 | - | ||
| 139 | - if($code!==200){ | ||
| 140 | - @file_put_contents(LOG_PATH.'/sync_es_fail'.S_V2.'.log',$id."\n",FILE_APPEND); | ||
| 141 | - _echo('同步es: '.$doc_id.'===>'.$code); | ||
| 142 | - } | ||
| 143 | - // _echo('同步es: '.$doc_id.'===>'.$code); | ||
| 144 | - // print_r($data); | ||
| 145 | - // var_dump($id); | ||
| 146 | - // exit; | ||
| 147 | - }else{ | ||
| 148 | - sleep(1); | ||
| 149 | - } | ||
| 150 | - } | ||
| 151 | - | ||
| 152 | - } | ||
| 153 | - | ||
| 154 | - | ||
| 155 | - /** | ||
| 156 | - * 查询项目id 和 邮件来源 | ||
| 157 | - * @param int $email_id 邮箱id | ||
| 158 | - * @return array | ||
| 159 | - * @author:dc | ||
| 160 | - * @time 2025/5/29 11:47 | ||
| 161 | - */ | ||
| 162 | - public function getPostid($email_id){ | ||
| 163 | - $data = redis()->getSet('fob_bind_mailx:'.$email_id,300,function ($email_id){ | ||
| 164 | - return $this->fob_db->throw()->first("select `post_id`,`source` from `e_mail_binds` where `email_id` = '{$email_id}' order by `id` desc limit 1"); | ||
| 165 | - },$email_id); | ||
| 166 | - | ||
| 167 | - return [ | ||
| 168 | - $data['post_id']??0, | ||
| 169 | - $data['source']??0, | ||
| 170 | - ]; | ||
| 171 | - } | ||
| 172 | - | ||
| 173 | - | ||
| 174 | - public function getEsData($data){ | ||
| 175 | - if(!empty($data['id'])){ | ||
| 176 | - $data['uuid'] = $data['id']; | ||
| 177 | - unset($data['id']); | ||
| 178 | - } | ||
| 179 | - | ||
| 180 | - $data['from'] = [ | ||
| 181 | - 'email' => $data['from'], | ||
| 182 | - 'name' => $data['from_name']??'' | ||
| 183 | - ]; | ||
| 184 | - unset($data['from_name']); | ||
| 185 | - unset($data['date']); | ||
| 186 | - | ||
| 187 | - if(!empty($data['to_name']) && strlen($data['to_name'])>10000){ | ||
| 188 | - $data['to_name'] = @json_decode($data['to_name'],true); | ||
| 189 | - if(is_array($data['to_name'])){ | ||
| 190 | - $data['to_name'] = array_map(function ($v){ | ||
| 191 | - $v['name'] = ''; | ||
| 192 | - return $v; | ||
| 193 | - },$data['to_name']); | ||
| 194 | - $data['to_name'] = array_slice($data['to_name'],0,100); | ||
| 195 | - $data['to_name'] = json_encode($data['to_name']); | ||
| 196 | - }else{ | ||
| 197 | - $data['to_name'] = ''; | ||
| 198 | - } | ||
| 199 | - } | ||
| 200 | - $data['description'] = str_replace(["\n"],"",$data['description']); | ||
| 201 | - // unset($data['to_name']); | ||
| 202 | - | ||
| 203 | - if(!empty($data['created_at'])){ | ||
| 204 | - $data['created_at'] = date('Y-m-d\TH:i:s',strtotime($data['created_at'])); | ||
| 205 | - } | ||
| 206 | - if(!empty($data['updated_at'])){ | ||
| 207 | - $data['updated_at'] = date('Y-m-d\TH:i:s',strtotime($data['updated_at'])); | ||
| 208 | - } | ||
| 209 | - $data['references'] = empty($data['references']) ? '' : $data['references']; | ||
| 210 | - return $data; | ||
| 211 | - } | ||
| 212 | - | ||
| 213 | - | ||
| 214 | -} | ||
| 215 | - | ||
| 216 | -(new SyncToEsCmd())->handler(); | ||
| 217 | - | ||
| 218 | -return 1; | ||
| 219 | - | ||
| 220 | - | ||
| 221 | - | ||
| 222 | - | ||
| 223 | - | ||
| 224 | - | ||
| 225 | - | ||
| 226 | - | ||
| 227 | - |
-
请 注册 或 登录 后发表评论