Discuz自動加標籤任務計劃

README.md

Discuz自動加標籤任務計劃

此功能在wetalk與herostory都有使用，Discuz計畫任務路徑在*/source/include/cron/*建立後的檔案可以從後台計畫任務設定排程執行 除基本需求外，包含判斷處理時間、發警示信件等附加需求做法。

cron_thread_autoadd_tags_daily.php

<?php
/**
 * 檔案路徑：/source/include/cron/cron_thread_autoadd_tags_daily.php
 * 需求：
 * 1. 建立標籤詞庫(人工或來自DZ系統發文時附加做法)
 * 2. 取出一定數量論壇文章，比對過的文章會配置 pre_forum_thread 新欄位 cronaddtag 次數值加1（後續持續循環找出最小數值持續比對附加）
 * 3. 根據特定規則(標題與內容搜尋)更新文章與標籤關聯，每篇最多5個標籤。
 *  1. 文章標題符合標籤
 *  2. 文章內文抓到符合出現3個以上同一標籤
 * */

if(!defined('IN_DISCUZ')) {
  exit('Access Denied');
}

date_default_timezone_set("Asia/Taipei");

/**
 * 記錄檔儲存參數
 * 1. 目錄夾
 * 2. 檔名規則
 * */
$log_path = DISCUZ_ROOT.'/data/plugindata/AUTOADDTAGS/'.date('Ymd');
if( !is_dir($log_path) ){
  mkdir($log_path, 0777, true);
  chmod($log_path, 0777);
}
$today = date("Ymd");
$log_filename = "tagRecord-".date('H-i').".txt";

/*設定參數*/
$lineString = [];
$limitNum = 350;
$startTime = microtime(true);
$endTime = "";
$lineString["limitNum"] = $limitNum;
$lineString["rowCount"] = 0;
// 處理時間超時(sec)
$toolong_time = 80;
$toMails = array("");


// 取出前100個 cronaddtag
$sql = "SELECT FT.tid, FT.subject, FT.cronaddtag, FP.message
        FROM ".DB::table('forum_thread')." FT
        LEFT JOIN ".DB::table('forum_post')." FP ON FT.tid=FP.tid AND FP.first=1
        ORDER BY cronaddtag ASC
        Limit ".$limitNum;
$rows=DB::fetch_all($sql);

// 取出目前現有標籤數
$tags=C::t('common_tag')->fetch_all_by_status(0);
foreach ($tags as $tag) {
  $namearr[] = $tag['tagname'];
}

$strtag = implode("|",$namearr);

$class_tag = new tag();
foreach($rows as $i => $row){

  $singleStartTime = microtime(true);
  $matchtag=array();
  if( preg_match_all("/$strtag/u", $row['subject'], $m1)){
    foreach($m1[0] as $k => $tag){
        if( $k < 5 ){
            $matchtag[]=$tag;
        }
    }
  }
  // 處理內容過濾htmltag
  $content=strip_tags($row['message']);
  if( preg_match_all("/$strtag/u", $content, $m2)){
    $content_tags= array_flip( array_count_values($m2[0]) );
    krsort($content_tags);
    foreach($content_tags as $times => $tag){
        // 若以設定5個標籤則停止。
        if( count($matchtag)==5 ){
            break;
        // 附加多的標籤與判斷比對到3次以上
        }elseif( !in_array($tag, $matchtag) && $times >= 3){
            $matchtag[]=$tag;
        }
    }
  }
  $lineString["msg"][$i]["subject"] = $row['subject'];
  $lineString["msg"][$i]["newTag"] = implode(", ",$matchtag);

  $sqlt = "SELECT t.tagname
           FROM ".DB::table('common_tagitem')." AS ti, ".DB::table('common_tag')." AS t
           WHERE ti.itemid = ".$row['tid']." AND ti.tagid = t.tagid
           ORDER BY ti.tagid asc";
  $rowt = DB::fetch_all($sqlt);

  $oldTag = [];
  if( $rowt ) {
      foreach($rowt as $t){
        $oldTag[] = $t["tagname"];
      }
  }else {
    $oldTag = "";
  }
  $lineString["msg"][$i]["oldTag"] = implode(", ",$oldTag);
  // if( count($matchtag) == 0 ){ continue; }

  //比對新舊tag是否相同
  $diffArr=array_intersect($matchtag, $oldTag);

  if(count($diffArr) == 0) {
    $lineString["diff"]++;
  }else{
    $lineString["same"]++;
  }

  $addtagstr = implode(',',$matchtag);
  /**
   * 如果原本已經有加入或設定過？
   * 1. 完全排除自動加功能
   * 2. 只先針對舊文設定一段時間之後都改手動加
   * */
  // update tags delete all set item first then add again with new set.
  C::t('common_tagitem')->delete(0, $row['tid'], 'tid');
  $tagarray = $class_tag->add_tag($addtagstr, $row['tid'], 'tid', 1);

  // update cronaddtag field
  $cronaddtag = ( empty($row['cronaddtag']) ) ? 0 : intval($row['cronaddtag']);
  DB::update('forum_thread', ['cronaddtag'=>$cronaddtag+1], "tid={$row['tid']}");

  $lineString["msg"][$i]["sub_totalTime"] = number_format((microtime(true) - $singleStartTime), 3, '.', '');

  unset($row);
  unset($matchtag);
  unset($totalTime);
  $lineString["rowCount"]++;
}

$lineString["totalTime"] = number_format((microtime(true) - $startTime), 3, '.', '');
$lineString["rowCount"] = $rowCount;

$lines_string .= "\r\n共花 " .number_format((microtime(true) - $startTime), 3, '.', '')." 秒\r\n";
$fp = fopen($log_path.DIRECTORY_SEPARATOR.$log_filename,'w');
fwrite($fp, "花費總時長: ".$lineString["totalTime"]." 秒\r\n");
fwrite($fp, "總數：".$lineString["limitNum"]." 筆 異動：".$lineString["diff"]." 筆 未異動：".$lineString["same"]." 筆 \r\n");

foreach ($lineString["msg"] as $key => $value) {
  fwrite($fp, "\r\n==========================================\r\n");
  fwrite($fp, $value["subject"]."\r\n");
  fwrite($fp, "newTag: ".$value["newTag"]."\r\n");
  fwrite($fp, "oldTag: ".$value["oldTag"]."\r\n");
  fwrite($fp, "處理時間: ".$value["sub_totalTime"]."\r\n");
}

fclose($fp);

if($lineString["limitNum"] != ($lineString["diff"] + $lineString["same"]) || $lineString["totalTime"] > 9) {
  include libfile('function/mail');
  $dateInterval = "WeTalk論壇_autotag_alarm_".date("Y/m/d H:i", time());

  $sendMsgs = array();
  if($lineString["limitNum"] != ($lineString["diff"] + $lineString["same"])){
    $sendMsgs[] = "<div><font color='#09C'>處理筆數錯誤</font></div>";
    $sendMsgs[] = "<div>總數：".$lineString["limitNum"]." 筆 異動：".$lineString["diff"]." 筆 未異動：".$lineString["same"]." 筆 </div><br><br>";
  }
  if($lineString["totalTime"] > $toolong_time){
    $sendMsgs[] = "<div><font color='#09C'>處理時間過長</font></div>";
    $sendMsgs[] = "<div>花費總時長: ".$lineString["totalTime"]." 秒 </div><br><br>";
  }

  if( count($sendMsgs) > 0 ){
    foreach ($toMails as $key => $toMail) {
      if(!sendmail( $toMail, $dateInterval, implode("<BR/>",$sendMsgs)) ){
          $errMsg[] = "send mail to ({$toMail}) error accured!";
      }
    }
  }

  if( count($errMsg) > 0 ){
    echo implode("<BR/>", $errMsg);
    echo $sendMsgs;
  }else{
      echo '<p style="display:none;">done!</p>';
  }
}