实现数据库抽取以及文件读取判断功能

README.md

@@ -3,3 +3,5 @@
 * 从数据库提取数据文件，生成包含数据集uuid、path、volume、count信息的文本
 * 逐条检查数据集的文件大小和数量，记录异常数据集
 * 将结果通过邮件进行发送
+
+只检查对外公开的数据集，并且有实体的数据集。

file-in-db.php

@@ -0,0 +1,32 @@
+<?php
+ $link = pg_connect("dbname=metadata user=gis password=gispassword host=210.77.68.241");
+ $host='ftp2.westgis.ac.cn';
+ $sql="select d.uuid,sum(df.filesize) as volume,count(df.id) as fcount,d.path from dataset d left join mdstatus ms on d.uuid=ms.uuid
+  left join datafile df on d.id=df.dsid
+  where ms.status>4 and d.host='".$host."'  group by d.id,d.uuid,d.path";
+ $result = pg_query($link, $sql);
+ $numrows = pg_num_rows($result);
+ $content='';
+ for($ri = 0; $ri < $numrows; $ri++) 
+ {
+  $row = pg_fetch_array($result, $ri);
+  $content.=$row['uuid']."\t";
+  $content.=$row['volume']."\t";
+  $content.=$row['fcount']."\t";
+  $content.=$row['path']."\n";
+ }
+
+ $filename='./dataset-0.txt';
+ if (!empty($content)) 
+  if (!file_put_contents($filename, $content, LOCK_EX))
+  {
+ 	sleep(1);
+ 	if (!file_put_contents($filename, $content, LOCK_EX))
+ 	{
+   sleep(1);
+   			if (!file_put_contents($filename, $content, LOCK_EX))
+   				die("error in write dataset-0.txt");
+   		}
+   	}
+	pg_free_result($result);
+	pg_close($link);

file-in-disk.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+rm dataset-1.txt
+cat dataset-0.txt | while read oneline;
+do
+        uuid=`echo $oneline | awk '{print $1}'`
+        datapath=`echo "$oneline" | awk -F'\t' '{print $4}'`
+		fcount=`find -L "$datapath"/* | wc -l`
+        if [ $fcount -gt 0 ]; then
+                volume=`du -sbL "$datapath"/* | awk '{a+=$1}END{print a}'`
+                echo -e $uuid'\t'$volume'\t'$fcount'\t'$datapath >> dataset-1.txt
+        else
+                echo $datapath;
+        fi
+done
+exit 0