数据增量逻辑修正

model/mWeiboBehavior.php

@@ -20,8 +20,8 @@ class mWeiboBehavior extends mBase {
         return $this->obj->replace($this->tbl_data, $data);
     }
 
-    public function getBehaviorByDate($date) {
+    public function getBehaviorByWids($wids, $date) {
-        return $this->obj->selectAll($this->tbl_data, array('sql' => "`date`=?", 'vals' => array($date)), 'id desc', array(0, 10000));
+        return $this->obj->selectIn($this->tbl_data, array('wid' => $wids), array('sql' => "`date`= ?", 'vals' => array($date)));
     }
 
     public function deleteExpireBehaviorData($date) {
@@ -53,7 +53,7 @@ class mWeiboBehavior extends mBase {
         $offset = ($page_num - 1) * $page_size;
 
         $orderby = "created_at desc";
-        if(!$date) $orderby = "wid desc";
+        if (!$date) $orderby = "wid desc";
 
         return $this->obj->selectAll($this->tbl_data, $where, $orderby, array($offset, $page_size));
     }

queue/crontab/spider_behavior_data.php

@@ -10,7 +10,7 @@ class spiderBehaviorData {
     public function __construct() {
         //删除半年前的数据
         $bobj = new mWeiboBehavior();
-        $six_month_ago = date('Y-m-d', strtotime('-6 month'));
+        $six_month_ago = date('Y-m-d', strtotime('-1 month'));
         $bobj->deleteExpireBehaviorData($six_month_ago);
 
         $user_list = array_keys($GLOBALS['WEIBO_USER_LIST']);

queue/crontab/spider_behavior_delta.php

@@ -6,47 +6,70 @@
 
 include_once(dirname(dirname(dirname(__FILE__))) . "/library/publicBase.php");
 
+define('DATE', $argv[1]);
+
 class spiderBehaviorDelta {
     public function __construct() {
         //删除半年前的数据
         $obj = new mWeiboBehavior();
-        $six_month_ago = date('Y-m-d', strtotime('-6 month'));
+        $six_month_ago = date('Y-m-d', strtotime('-1 month'));
         $obj->deleteExpireBehaviorDelta($six_month_ago);
 
         //统计今日增量
-        $today = date('Y-m-d');
+        $today = date('Y-m-d',strtotime(DATE));
-        $today_data = $obj->getBehaviorByDate($today);
-        $today_data = array_column($today_data, null, 'wid');
 
-        $yesterday = date('Y-m-d', strtotime('-1 day'));
+        $page_num = 1;
-        $yesterday_data = $obj->getBehaviorByDate($yesterday);
+        $page_size = 2000;
-        $yesterday_data = array_column($yesterday_data, null, 'wid');
+        while (true) {
+            $today_data = $obj->getSpiderBehavior($today, 0, 0, $page_num, $page_size);
+            if (empty($today_data)) break;
+            $page_num++;
 
-        if (empty($yesterday_data)) return true;
+            $today_data = array_column($today_data, null, 'wid');
 
-        foreach ($today_data as $k => $v) {
+            $wids = array_column($today_data, 'wid');
-            $yesterday_reposts_count = 0;
-            $yesterday_comments_count = 0;
-            $yesterday_attitudes_count = 0;
 
-            if (isset($yesterday_data[$k])) {
+            $yesterday = date('Y-m-d', strtotime('-1 day', strtotime($today)));
-                $yesterday_reposts_count = $yesterday_data[$k]['reposts_count'];
+            $yesterday_data = $obj->getBehaviorByWids($wids, $yesterday);
-                $yesterday_comments_count = $yesterday_data[$k]['comments_count'];
+            $yesterday_data = array_column($yesterday_data, null, 'wid');
-                $yesterday_attitudes_count = $yesterday_data[$k]['attitudes_count'];
+
-            }
+            $three_day_ago = date('Y-m-d', strtotime('-2 day', strtotime($today)));
+            $three_day_ago_data = $obj->getBehaviorByWids($wids, $three_day_ago);
+            $three_day_ago_data = array_column($three_day_ago_data, null, 'wid');
+
+            //将前天的数据查询出来做兼容
+            if (empty($yesterday_data) && empty($three_day_ago_data)) return true;
+
+            foreach ($today_data as $k => $v) {
+                $reposts_delta = 0;
+                $comments_delta = 0;
+                $attitudes_delta = 0;
+
+                if (isset($yesterday_data[$k])) {
+                    $reposts_delta = $v['reposts_count'] - $yesterday_data[$k]['reposts_count'];
+                    $comments_delta = $v['comments_count'] - $yesterday_data[$k]['comments_count'];
+                    $attitudes_delta = $v['attitudes_count'] - $yesterday_data[$k]['attitudes_count'];
+                }
+
+                if (isset($three_day_ago_data[$k]) && !isset($yesterday_data[$k])) {
+                    $reposts_delta = $v['reposts_count'] - $three_day_ago_data[$k]['reposts_count'];
+                    $comments_delta = $v['comments_count'] - $three_day_ago_data[$k]['comments_count'];
+                    $attitudes_delta = $v['attitudes_count'] - $three_day_ago_data[$k]['attitudes_count'];
+                }
+
+                $delta = array(
+                    'uid' => $v['uid'],
+                    'wid' => $v['wid'],
+                    'reposts_delta' => $reposts_delta,
+                    'comments_delta' => $comments_delta,
+                    'attitudes_delta' => $attitudes_delta,
+                    'date' => $v['date'],
+                );
 
-            $delta = array(
+                $res = $obj->saveBehaviorDelta($delta);
-                'uid' => $v['uid'],
+                if (!$res) {
-                'wid' => $v['wid'],
+                    $this->writeLog('spider', 'insert_error.log', json_encode($delta));
-                'reposts_delta' => $v['reposts_count'] - $yesterday_reposts_count,
+                }
-                'comments_delta' => $v['comments_count'] - $yesterday_comments_count,
-                'attitudes_delta' => $v['attitudes_count'] - $yesterday_attitudes_count,
-                'date' => $v['date'],
-            );
-
-            $res = $obj->saveBehaviorDelta($delta);
-            if (!$res) {
-                $this->writeLog('spider', 'insert_error.log', json_encode($delta));
             }
         }